Lansil Global — Brand Database (BRAND.md)
The single source of truth for every durable fact about this client. Any note David or Alan sends
(facilities, rates, accounts, contacts, decisions) is filed HERE the same turn it arrives — before any
context compaction. Memory files only POINT here. Never store secrets (tokens, keys, passwords):
reference the env file that holds them. Keep entries dated. Hub renders this file.
Relationship: CLIENT (agency engagement, since 2026-09-10) — and David is also a Lansil customer
(Charm 3PL · NailScience direct-China via 4PX · HumanFrame sourcing quote), which is first-person
customer-discovery material once Alan consents to its use.
0 · Identity & entities
| Field | Value |
| Company | Lansil Global — China product sourcing + manufacturing/QC + freight + 3PL fulfillment for eCommerce/DTC brands. Founded 2015 (site: Hong Kong; interview: built from Shenzhen). |
| Legal entities | Invoicing entity on David's own POs: Shenzhen Lansil Trading Ltd. US entity for NV/PA: TO FILL. HK entity: TO FILL. |
| Founder / decision maker | Alan Coughlan, CEO & Founder — Cork-born, in China since 2011, lost his first import venture (bicycles) to factory quality, founded Lansil 2015. VERIFIED: Pulse 2.0 interview 2024-12-13 (pulse2.com/lansil-global-profile-alan-coughlan-interview/). |
| Dwight Hijastro (kickoff 2026-10-09) | Alan's marketing-ops lead, ~2 years in (from the original Philippines team of eight); runs the GoHighLevel CRM + the inbound lead pipeline with RBD in China; weekly report to the China team. A China-based support person supplies warehouse photos / video. The previous China marketing team (~2 months) was let go — ad overspend, no sales-qualified leads; the previous Dubai team never handed over Meta Business Manager admin. |
| Leadership (site, 2026-09-10) | Peki (GM) · Sophia (Head of Ops) · Jenifer (Head of Ops, USA) · Robert (Head of Business Development) · Nick (Warehouse Manager, USA) · Heather (Customer Support Mgr, USA) · Ivy (Head of Customer Support) · Kathy (Purchasing) · Lily (Finance) · Sharon (Logistics) · Tex (Head of Packaging) |
| Support | support@lansilglobal.com · CN +86 755-3290-2756 · US +1 702-462-2035 · WhatsApp chat widget on site |
| Addresses | Shenzhen: 4/F, Bldg A1, ABao International Optical Communication Industrial Zone, Shiyan St, Bao'an · PA: 270 Old Silver Spring Rd, Mechanicsburg, PA 17050 · NV: 6455 E Tropical Pkwy, Suite 140, North Las Vegas, NV 89115 (a 3PL-Hub listing says "Henderson, NV" — reconcile) · Dubai office (interview) |
| Ownership | 100% privately owned, no VC / PE / outside investors (Alan, verbatim: "pursue organic growth and focus on service quality rather than acquiring market share") — VERIFIED |
| Service stance | "100% human customer service, rather than depending on chatbots and automated replies" — VERIFIED (interview) |
| Scale (as stated, dates matter) | Interview 2024: ~$150M revenue · 15M+ packages · 4,000+ items sourced · 2,000+ factories · 171+ countries. Homepage counters 2026-09-11: 3 warehouses · 20M orders · 1,492+ factory relations · 74+ team. About page: 100+ employees. → numbers disagree across surfaces; get the current set from Alan before any appears in copy. |
| Brand colours (site CSS) | accent #EE4127 · ink #231F20 · white · logo = red "LG" mark + LANSIL GLOBAL wordmark |
| Our contact / owner | David Sanghera (agency) — Alan is the client contact; approval flow TO CONFIRM |
1 · Service lines (the "products")
| Code | Service | Promise on site | URL | Notes |
| S1 | China Product Sourcing | "China sourcing agents with 2000+ trusted vendors for quality products at competitive prices" | /china-product-sourcing/ | 4,000+ unique items sourced (interview) |
| S2 | Manufacturing & Quality Control | "Bring your product to market in no time" · in-house QC team | /china-manufacturing/ | the founder-origin proof point |
| S3 | Direct China Fulfillment | "over 200 countries in 5–10 days" (hero: 171+ countries) · via 4PX (David's NailScience experience) | /direct-china-fulfillment/ | post–de minimis economics changed — position honestly |
| S4 | USA Order Fulfillment | "across the States in 2–4 days" from NV + PA | /usa-order-fulfillment/ | the growth lever (new capacity) |
| S5 | Amazon FBA Prep | "from our China and USA prep centers" | /amazon-fba-prep/ | |
| S6 | China Global Freight | "Reliable China air and sea freight… start to finish" | /china-global-freight-services | customs brokerage bundled (NailScience: bundled in 4PX rate) |
| Add-ons | Trademark filing · customs brokerage · returns handling (China + USA) | trademark filing = legal-service claim, check wording | ||
| Integrations | Shopify · Amazon · manual CSV (site). DR reports claim Woo/BigCommerce — unverified. | |||
| Financing | Up to $1M credit on 60-day payment terms (site). Qualification rules: TO FILL. | the #1-ranked mechanism |
- Operations facts confirmed by David 2026-10-20 (Alan's questions 6–17): one SKU can run Shenzhen-direct and from a US warehouse at once; stock moves between Shenzhen, NV and PA; no minimums (first run, monthly volume, dropship accounts); US import clearance in-house; NV + PA ship Amazon Seller Fulfilled Prime; custom packaging is a service; freight ships internationally; clients = ecommerce companies; no category refused; QC checklist / reject rate and capacity: not available (not sayable).
2 · Footprint & supply chain
| Node | What we know | Verify |
| Shenzhen HQ + warehouse | fulfils direct-to-consumer via 4PX; in-house QC; FBA prep | size: "45,000 sq ft" (07) vs "98,221 sq ft" (04) |
| North Las Vegas, NV | opened 2020 (interview) | size: "106,624 sq ft" (3PL Hub listing, 07) |
| Mechanicsburg, PA | opened 2023 (interview), 32,000 sq ft (interview) / 32,069 (07) | |
| US network total | "236,000" / "138,000" / "157,000" sq ft — three different figures across DR steps | get real numbers from Alan |
| Planned | Sunbelt-state + Europe warehouses; Dubai office / Middle East (interview) | current status |
| Delivery SLAs | US 2–4 days · China→US 5–8 days · China→UK 3–5 days · worldwide 5–10 days | as claimed on site |
| Capacity | orders/day, onboarding slots/month, BD headcount — unknown, decides the CAC ceiling | kickoff |
| David's live experience | Charm: Hansford shipped 7,993 / 7,018 / 3,002 units 2026-09-01, Lansil receiving counts pending · NailScience: customs + brokerage bundled in the 4PX rate (Lansil-confirmed) · HumanFrame: shoulder-massager quote 2026-05-13 EXW Shenzhen, class-9 DG paperwork (UN38.3 / MSDS) open |
3 · Offer & pricing
- Today: quote-only ("Get a Quote" / "Get Started" / eBook download). No published rate card, minimums, storage or pick/pack fees anywhere on the site.
- Alan's brief (2026-09-10): the new website should carry rates. Competitor research validates it — the three best-liked competitors (Portless, Ecomflow, NextSmartShip) publish; the two most hated (ShipBob, ShipMonk) gate or bury them.
- Financing hook: up to $1M credit / 60-day terms — David 2026-10-20 (Q4): the terms cover the factory invoice and the freight, not fulfillment; anyone qualifies, based on order volume.
- Guarantees Lansil stands behind (David 2026-10-20, Q5 “yes”): no hidden fees, no exit fees, no postage markups, free rework when QC fails — Truth Gate f11/f12; the research brief had invented them, now confirmed. The offer numbers (Q1–Q3): sourced cost about 30% under the Alibaba price; fulfillment about 40% under the US average; the calculator's rate card confirmed as built. The calculator's fee list can carry “Exit fees $0” as one more zero line (not built yet).
- NextSmartShip's entry offer to beat or ignore: free onboarding, 60 days free storage, no minimums, $4.10/0.9 kg published.
- The calculator / tripwire (David, 2026-10-13): the pricing page's lead magnet is a modular cost calculator — "find your landed cost, find how much your pick-and-pack fees are" — whose job is to show Lansil's low pick-and-pack against the competitors' fee stacks. Pick & pack slides with volume: $2.50 per order at 1,000 orders/month down to about $1.50 at the top of the slider. One flat fee per order (no per-unit / additional-pick fee), no storage fees, no extra pick-and-pack fees; every hidden line the others charge (transfer fees, kitting fees, receiving, packaging, minimums, account fees) appears as a $0 line item because Lansil does not charge it. Gate: orders per month + email + phone, then they manipulate it (daily cost, per-order cost, monthly total vs a named rival) — persuasive enough to jump on a call. Numbers sit on the Truth Gate as p1–p3 until Alan confirms the rate card. Spec:
briefs/lansilglobal/site/calculator.md. - The calculator's numbers, David's confirmations (2026-10-13, answering the spec's “what it needs from Alan”): (1) the pick-and-pack tiers as proposed ($2.50 at 1,000 → $2.25 at 2,500 → $2.00 at 5,000 → $1.75 at 10,000 → $1.50 at 20,000+); (2) returns handling $0 and packaging $0; (3) free storage, forever; (4) freight at the standard rates by cubic volume and weight, no markup; (5) last mile on DHL eCommerce (the ecom saver service), no margin taken — the same cheap postage every brand gets, so postage is a pass-through line, not a differentiator; (6) the 60-day terms stand (o4). Usable in the calculator build now on David's word; Alan's sign-off is a launch check (§9 cutover list).
- No competitor names on the calculator (David, 2026-10-14: "Lets not use competitor names on the calculator"): the other column is “a typical US 3PL” from the dated survey averages, the only rival the page shows; the ShipMonk / ShipBob research stays in
site/calculator.mdfor our reading, never on the page, its presets, its sources line or its copy. The lanes andpricing/calc/rivals.jsoncarry one preset. - The lock — blur everything until Name, Email and Phone are in (David, 2026-10-14: "we first need to collect their Name, Email, Phone, and then somehow pass that into the CRM with their orders range they put in the calculator as they may play with the numbers"): the controls move, every output is blurred with the form on top; the submit upserts the GoHighLevel contact (source “Landed cost calculator”, tags
calculator-lead+website, the numbers on screen ascalc_*custom fields) and lifts the blur; every change after that updates the contact, with the lowest and highest orders explored. Built:products/lansilglobal/functions/lead.mjs(Netlify function, ships with--functions; envLANSIL_GHL_TOKEN+LANSIL_GHL_LOCATIONon the site, never the repo; without them it answers “prototype” and stores nothing),pricing/calc/ui.jsLansilUI.lock+lock.css,clients/lansilglobal/ingest/ghl_fields.py ensurefor the ten fields. Verified on staging 2026-10-14: locked on load, unlock on submit, an update sent on the next slider move, remembered on reload.
4 · Accounts & IDs (no secrets)
| System | ID / handle | Notes | |
| Domain / DNS | lansilglobal.com — DNS at Hostinger (Dwight, kickoff 2026-10-09); the new site goes live by forwarding the DNS to Netlify. Zone as read 2026-10-18: nameservers ns1/ns2.dns-parking.com (= Hostinger's own DNS, so the Hostinger API edits this zone directly), apex A 191.101.104.166 + 195.35.60.254, www CNAME www.lansilglobal.com.cdn.hstgr.net (Hostinger CDN in front of WordPress — the CDN hostname is part of the cutover when WordPress moves to its own host), no rates record yet. Route for our own DNS edits (David 2026-10-18): the official Hostinger Claude plugin (hostinger@claude-plugins-official, remote MCP mcp.hostinger.com, browser OAuth sign-in by David); fallback = the REST API with a token David creates at hpanel.hostinger.com/api, kept only in ~/.lansil/hostinger.env. Every write goes overwrite: false + validate + snapshot/diff — a full-zone overwrite would take the live site down. Read 2026-10-19 through the plugin, signed in as Lansil's Hostinger account: the account hosts lansilglobal.com (WordPress, main vhost since 2025-02-22, hosting user u264441994) and sandbox.lansilglobal.com (subdomain vhost, 2026-09-07); its subscriptions show a KVM 2 VPS ($203.88/yr, since 2026-02-27, auto-renew) and one unused free-domain credit from the same day; the domain itself is registered at eNom (created 2015-09-24, expires 2027-09-24), not in Hostinger's domain portfolio, so the Hostinger DNS API refuses the zone (403 DNS:4002 Customer does not own) even though Hostinger's nameservers serve it — the API/plugin cannot add rates; the zone is edited in hPanel's DNS Zone Editor under the hosting website (browser), or Dwight adds it. Full zone inventory 2026-10-19: clients/lansilglobal/dns/2026-10-19-zone.md — apex ALIAS + www CNAME on the Hostinger CDN; ship/link/unbouncevalidation = the old Google Ads funnel on Unbounce (ludicrous.cloud); quote A 76.13.221.55 (origin unknown); go + tiktok A 182.92.94.202 (a Chinese-hosted redirect); Google Workspace MX/SPF/DKIM on the apex, no apex DMARC; reply = the CRM's sending subdomain on Mailgun (LeadConnector SPF, DKIM, DMARC p=none). Hostinger plan facts from hPanel: Premium, owner alan@lansilglobal.com, expires 2029-04-29; David's own Hostinger login sees it under “Shared with you” with admin access | /faqs/ nav link 404s | |
| Site CMS / host | WordPress (confirmed 2026-10-09); not touched — the new site is a separate build on Netlify, DNS switched at launch | WhatsApp widget present | |
| Google Ads | Three accounts, all open to david@davidsanghera.com. (a) “Lansil Global” 873-360-9807 — the MANAGER (Standard, 2026-10-14). (b) “ECOM LG FZE” · 218-766-7313 — owned by Lansil Global, paid on Lansil's credit card (Dwight, 2026-10-15; a direct grant, not under the manager) — David 2026-10-15: “please use the ECOM LG FZE account” = the relaunch runs here. Not reachable yet (2026-10-15, after the Google re-auth): listAccessibleCustomers on david@davidsanghera.com shows 18 accounts including the Lansil manager, not 218-766-7313, and the API answers 403 under both logins — the invite was not accepted by that address or went to another email (the GA4/GTM pattern). Ask: accept the Google Ads invitation as david@davidsanghera.com, or Dwight re-sends it to that address; then gads_pull.py pull --customer 2187667313 (the puller signs in as the account itself for a direct grant). READABLE 2026-10-20 night (David: “just got access to it”; the puller signs in as the account itself): ECOM LG FZE · USD · Asia/Dubai · auto-tagging on · conversion tracking id 16666567492 — which is the id the morning's GTM read called “a different account” and the Google tag's second destination: AW-16666567492 IS ECOM LG FZE. 47 campaigns (36 paused, 11 removed; 42 search, 3 Performance Max, 2 video); $44,911 spent in 15 of the last 24 months; the 2025 months count a page view as a conversion (2025-06: 3,438 clicks → 2,388 “conversions”). Full read → clients/lansilglobal/ads/google/2187667313/2026-10-20/SUMMARY.md. Conversion actions created 2026-10-21 (David's go: “wire these up … for Google Ads events”), both website, one per click, 30-day click window, no value: Calculator Lead (id 7862650519, Submit lead form, PRIMARY, label 6Qb2CJeNmqUdEMTGnos-) on the calculator unlock and Calculator WhatsApp click (id 7862650522, Contact, secondary, label iAHjCJqNmqUdEMTGnos-) on the WhatsApp button — fired by the runtime's AW_LABELS map through the Google tag with the event id as transaction_id. The page-view “Lead” action stays primary until David's separate go to demote it. (c) 蓝斯尔贸易-GG-XD4PX · 627-986-4143 — owned by the Chinese agency (Dwight), under the manager; the history, read-only for us. (c) read 2026-10-14 by clients/lansilglobal/ingest/gads_pull.py pull --customer 6279864143 (API v24; v21 is sunset) → clients/lansilglobal/ads/google/6279864143/2026-10-14/SUMMARY.md. The read: conversion tracking id 18265417989, auto-tagging on. 8 search campaigns, all PAUSED, all Maximize-clicks (TARGET_SPEND) at $20–30/day, split by country (US · AU · UK · CA · AU&US phrase · Global) for “China 3PL” and “US 3PL”. Spend in only 3 of the last 24 months: Jul 2026 $1,825 (166 clicks, 6 conv) · Aug $3,394 (342, 8) · Sep $1,157 (51, 5) = $6,376, 559 clicks, 19 conversions, CPC ~$11, ~$336 per conversion. Countries: US $3,096 → 4 conversions; Australia $1,381 → 11; UK $911 → 1; CA $634 → 2. Devices: desktop only. Conversion actions: “Submit lead form” (87 all-conv, NOT primary), “Lead” = a codeless PAGE VIEW marked primary (counts a page view as a lead), two codeless form submits (lansilglobal.com + ship.lansilglobal.com) primary, one removed. 3,155 keywords with traffic (exact 1,621 · phrase 1,246 · broad 288), the same terms duplicated across the country campaigns (“3pl china” exact in five), QS mostly 1–5; 1,046 search terms cost $3,289 of which 46 terms with $20+ and zero conversions ate $2,400 (“china 3pl” $234, “3pl services in usa” $137, “3pl usa” $136, competitor names such as “sj henderson fulfilment ltd” $87, “cainiao”, “shenzhen 1st international logistics”). Negatives: 0 campaign-level, 1,991 ad-group-level, 8 shared lists (three Chinese “广告法” advertising-law lists of 1,054 / 1,242 / 511 terms, “Negative Keywords 3PL” 263, “Negative Keywords US” 226). 36 RSAs, strength mostly AVERAGE (three POOR), generic headlines (“For eCommerce Brands · China 3PL for eCommerce Brands · China Ecommerce Fulfillment”). Landing pages = ship.lansilglobal.com funnel pages (/3pl-usa, /3pl-china, /3pl-in-china, /china-sourcing) with doubled, malformed UTM templates, plus lansilglobal.com/china-3pl-for-ecommerce-brands and /china-product-sourcing. Change history: Dwight made 616 changes in the last 29 days (581 keyword/negative edits) while the account sat paused | ads paused for October (Q4 slowdown) by Lansil · the manager account is clean for the relaunch structure | |
| Meta BM · ad account · pixel | Lansil's BM = “Lansil Global” 838347503664640 — BM admin exists at Lansil, the Dubai-era gap is closed. Ad account 1221470026065851 “Lansil Global V2”: ACTIVE, USD, timezone Asia/Manila (reporting days are Manila days — note before any US campaign), created 2025-04-07, Mastercard on file, lifetime spend $376, 8 campaigns all PAUSED since 2025-06-24 (the FBFY25Q2 run: tariff-angle videos “Still shipping US orders from China?”, a China-sourcing-playbook download, WhatsApp-click engagement ads; 9 leads at ~$17, 18 WhatsApp conversations at ~$4–10, 911 landing-page views from a retargeting link-click campaign); audiences LG Engagers IG · LG Website Visits · LG FB Engagers; Instagram @lansilglobal (17841402930217458) attached. Shared to AMV LLC 113824746806442 (our BM) — our adscaler system user asu reads it and can launch in it (verified 2026-10-13). Facebook Page “Lansil Global” 1145315282000023 — SHARED 2026-10-20 (David: “added 1145315282000023 to the ASU”): the system user holds CREATE_CONTENT · MODERATE · MESSAGING · ADVERTISE · ANALYZE · MANAGE_LEADS on it (45 followers, category Business service, not verified, published; probe 2026-10-20). act_1221470026065851/promote_pages does not list it yet — confirm at the first ad creation (dry run) that the ad account can reference the Page; if not, the Page needs assigning to the ad account inside Lansil's Business Manager. Pixels on the account: “Lansil Global” 1182666090073735 (Lansil's BM, created 2025-04-07, alive on the WordPress site — last fired 2026-10-12; last 28 days PageView 1,887 and nothing else: no Lead / Contact / form event exists) and “Lansil Pixel 2.0” 1803603354256711 (created 2026-10-13 INSIDE AMV LLC, never fired = the new site's pixel; David: “full access” 2026-10-13). The CAPI token David pasted (~/.lansil/meta.env; app “Conversions API Application”, scope read_ads_dataset_quality only) still cannot see Pixel 2.0 — regenerate one for Pixel 2.0 when the relay is wired, or the relay sends as asu. Page: pending (no promotable Page on the account yet). Pull: clients/lansilglobal/ingest/meta_pull.py pull → clients/lansilglobal/ads/meta/2026-10-13/SUMMARY.md | Earlier read (Dwight, 2026-10-09: admin never handed over by the Dubai team) — superseded | |
| LinkedIn company page · Instagram · YouTube · X · Facebook | present in site header — admin access TO FILL | ||
| GA4 · GTM | Access granted 2026-10-13 (David: "I have google analytics access, tag manager") — most likely to David's Lansil Workspace login (david@lansilglobal.com, credentials re-sent by Dwight 2026-10-09, §9), because the API probe on david@davidsanghera.com (2026-10-13, fresh consent: Analytics + Tag Manager + Drive) sees none of Lansil's properties or containers. Fix: from the Lansil login add david@davidsanghera.com to the GA4 property (Admin → Property access management, Editor) and to the GTM container (Admin → User management, Publish) — or Dwight does it. Then clients/lansilglobal/ingest/ga4_pull.py discover / gtm_pull.py discover flag them and pull --auto writes clients/lansilglobal/tracking/. Re-checked 2026-10-19 through the GTM MCP server (David authenticated it that night; list_accounts sees five accounts — MGI Clinic, Nail Science, Human Frame, byskinscience.com, M-G-I — and no Lansil account), so the add was still owed. GTM DONE 2026-10-20 evening — “Lansil Global” (account 6283574419, container www.lansilglobal.com GTM-WH5KZ45S) now lists for david@davidsanghera.com (list_accounts + David's screenshot); GA4 DONE the same evening — after the gcloud re-login ga4_pull.py discover sees “LG Analytics” property 465005552 (account “Lansil Global”, created 2024-10-28, America/Los_Angeles, 14-month retention), one web stream “Lansil Global” G-GDFTX1H761 (https://lansilglobal.com/), key events purchase · lead_generate (custom, 2025-05-27) · form_submit (2026-04-14), custom dimension googlesitekit_post_type (the site runs Google Site Kit), linked to Google Ads 2187667313 = ECOM LG FZE (link 2024-11-22). pull --auto → clients/lansilglobal/tracking/ga4/465005552/SUMMARY.md (90 days: 6,575 sessions · 6,200 users · 17,873 page views · 61 key events; Direct 62 %, Organic Search 29 %, Paid Social 146 sessions / 6 key events, Paid Search 59). Correction of the morning's read: the measurement id is NOT dead — the property receives the site's traffic through Site Kit's Google tag GT-MJM9PB4Z; what 404s at Google is the standalone loader gtag/js?id=G-GDFTX1H761 (and the destination fetch gtag makes when a page configures the G- id directly), so any page that wants this property must load the GT tag. Earlier that day the container already answered: lookup_container GTM-WH5KZ45S → container 214372234 in account 6283574419 “www.lansilglobal.com”, and get_live_version reads it (container-level access exists for the GTM server's Google identity even though the account is not listed). Live version 11 inventoried → clients/lansilglobal/tracking/gtm/2026-10-20-GTM-WH5KZ45S-live.md: Google tag GT-MJM9PB4Z fires ONLY on the thank-you page; Google tag AW-18265417989 + Conversion Linker on all pages; GA4 events website_view / engagement to G-GDFTX1H761 (= the property's measurement id); the primary Google Ads conversion (AW-18265417989 / zI5oCJ2l6MUcEIXS0IVE) fires on a page view of /tell-us-about-your-business; enhanced conversions point at a DIFFERENT account, conversionId 16666567492; a Reddit pixel. No UTM/click-id variables. Outside GTM, the WordPress theme hard-codes Meta pixels 1182666090073735 and 1005023722227372 (PageView only) — Pixel 2.0 1803603354256711 is nowhere on the site. The GA4 Data API still needs the house Google login (invalid_rapt again 2026-10-20) | ||
| Calculator tracking (built 2026-10-20) | Live on rates.lansilglobal.com — the house attribution stack, lead-gen cut. products/lansilglobal/tracking/lg-attribution.js (runtime: click ids + UTMs persisted 90 d in _lg_* cookies with localStorage fallback, first-touch kept separate from last-touch, device-level external_id _lg_eid, one event_id per event shared by the browser pixel and the server beacon, Pattern M settle, no _fbp/_fbc writes, PII SHA-256 hashed client-side because shared/functions/capi.mjs forwards user_data as sent). Head of every lane: MutationObserver guard → LG_TRACKING config → Meta base code (autoConfig off, init only) → the runtime (Pattern F PageView). Events: PageView · CalculatorPlay (first slider move) · Lead on the unlock (browser + Conversions API, em/ph/fn/ln/external_id/fbp/fbc/gclid). Relay: products/lansilglobal/functions/capi.mjs re-exports the house relay at /api/capi (POST/OPTIONS only — a GET is 404 by design; the deploy smoke sends a preflight); env FB_PIXEL_ID + FB_CAPI_ACCESS_TOKEN (secret) on the site. CRM: 17 attr_* custom fields written on create (UTMs first + last touch, gclid/gbraid/wbraid/fbclid, fbc/fbp, external id, landing page, referrer, first seen). Proof 2026-10-20: headed Chromium — PageView, CalculatorPlay and Lead reached facebook.com/tr on Pixel 2.0 with the same ids as the three /api/capi beacons (dedup proven); the Lead beacon carried em, ph, fn, ln, external_id, fbp, fbc; the CRM contact carried every attribution field; test contacts deleted, counters reset. Headless Chromium never surfaces the pixel's own beacon (true on a proven house site too) — prove the browser side headed or in Chrome. GA4 live the same evening: the lanes load Lansil's Google tag GT-MJM9PB4Z (the only loader that serves G-GDFTX1H761) with consent defaults ad_storage / ad_user_data / ad_personalization denied, analytics_storage granted, ads_data_redaction on and ga-disable-AW-16666567492 — because that Google tag carries a second destination, Google Ads AW-16666567492, an account that is not in Lansil's map, which without the consent defaults started collecting the opt-in form's data (google.com/ccm/form-data/16666567492) and remarketing pings; with them it receives only cookieless, id-redacted pings. Resolved the same night: AW-16666567492 is ECOM LG FZE's own conversion-tracking id (customer 218-766-7313; a Google Ads account's tag id and customer id differ), so the muzzle came off — the lanes configure the Google tag plainly (GA4 + the ECOM LG FZE conversion linker, enhanced conversions for leads and remarketing), the runtime's AW_ID is AW-16666567492, and the Lead conversion fires since 2026-10-21 (Calculator Lead, primary) with the WhatsApp click as a secondary Contact conversion — see the Google Ads row. GA4 events from the calculator: page_view · calculator_play · form_start (the tag's own enhanced measurement) · lead_generate on the unlock — the property's existing key event, so the calculator's leads count where their site's leads already count (proven headless 2026-10-20: four hits to tid G-GDFTX1H761, no other destination). Google Ads Lead conversion route: GA4 is linked to ECOM LG FZE, so the conversion is an import of the GA4 key event lead_generate in that account (Dwight, or us once the invite lands) — no tag needed; the AW_ID / AW_LEAD_LABEL slots stay for a direct tag if ever wanted (AW_ID + AW_LEAD_LABEL slots in the runtime, gclid already on every contact for offline import); Phase 4 (Events Manager 24 h: Lead browser+server deduplicated, EMQ ≥ 6, fbp ≥ 90 %) due 2026-10-21; the new site's own pages get the same head block at cutover | GA4 stream id · ECOM LG FZE access · the Phase 4 read | Property id · measurement id · GTM id TO FILL after the first pull |
| CRM / quote-form destination | GoHighLevel sub-account “Lansil Global GHL” vQjXuHT0MIrxWoFoP5vk (MainStreet CRM white-label; Central, Hong Kong; timezone Asia/Shanghai). Private Integration token in ~/.lansil/ghl.env (David, 2026-10-13; every endpoint reads); write permission on the Private Integration landed 2026-10-15 (David: “We have write permission now on CRM”): the ten calc_* custom fields created by ghl_fields.py ensure, the token + location on the Netlify site lansil-site as LANSIL_GHL_TOKEN (secret, production / deploy-preview / branch-deploy contexts) + LANSIL_GHL_LOCATION via the Netlify API, and the calculator writes real contacts — verified end to end on staging 2026-10-15: the unlock created a contact with 9 fields, the next slider move updated it with the explored range (1,000–4,000), the test contact was then deleted. Re-proven on the LIVE address 2026-10-20 (David: “did they actually put the lead in the CRM?”): a Playwright visitor on https://rates.lansilglobal.com/lanes/bill/ moved the slider, opted in, the blur lifted, and MainStreet CRM held the contact within a second — source “Landed cost calculator”, tags calculator-lead + website (+ the CRM's own new contact), nine calc_* fields; the next slider move rewrote them (orders 7,000, range explored 650–7,000); the test contact deleted. The contact carries attribution: null — the calculator sends no UTM, landing page, referrer, gclid or fbclid, so a CRM lead cannot be tied to an ad yet (the §9 tracking build). Stage glossary (David 2026-10-20) → clients/lansilglobal/crm/STAGES.md: New Lead = all leads · Low MOQ = below minimum volume · Qualifying = two follow-ups in progress · Future Opportunity = potential later · No Response = no reply after two follow-ups · Disqualified = bad fit or spam · Opportunity = sales-qualified · Negotiation = pricing sent, under discussion · Pricing Sent, No Response · Signed Up = won · Lost · Unqualified. Maps onto the live stages (Bad Fit + Spam = Disqualified; Qualifying + To Follow Up + 2nd Follow Up = Qualifying; Opportunity - WhatsApp/WeChat + Booked Call = Opportunity) and onto ECOM LG FZE's offline conversion uploads (Interest → Opportunity → Quote Negotiation → Signed Up). No glossary line for Booked - No Show and W&F; no response-time target per stage; the two “Signed Up” stages (Sales 34, Close 22) still unreconciled. Still Dwight's: the workflow on the calculator-lead tag (instant text · the setter's call · pipeline stage). What is in it (365 days, pulled 2026-10-13 → clients/lansilglobal/crm/2026-10-13/SUMMARY.md): 3 pipelines — Sales Pipeline (18 stages; open today: No Response 377 · Bad Fit 318 · Future Opportunity 213 · Opportunity WhatsApp/WeChat 49 · Low MOQ 28 · Pricing Sent/No Response 28 · Pricing Sent/Negotiation 18 · Booked Call 1 · Booked No Show 4; won 34 at Signed Up, lost 42), Ebook Lead (8 open), Close: Sales Pipeline (Post-Discovery Call 32 open · Proposal Sent 9 open worth $9,000 · Signed Up 22 won). 1,267 opportunities: 50–95 new a month, 161 in Sep 2025, 132 / 182 in Jul–Aug 2026 (the Meta lead-gen run), 54 in Sep, 14 in Oct (ads paused); about 2 won a month. Sources: Sales Form Completed 249 · none 243 · Facebook Ad 208 · website form 200 · 1st Contact Form 132 · WhatsApp chatbot 101 · GA Ads Growth Lab (Google Ads) 39 · TikTok Ads 36 · website chatbot 12. 9,704 contacts, of which 7,702 were imported in Sep 2025 as an outbound list (tags outbound 8,145 / inbound 973); inbound runs 60–140 a month, 213 in Aug 2026. No UTM attribution on any contact — first- and last-touch source/medium are empty on all 9,704: the CRM knows which form, never which campaign. 65 workflows (published: the two guide opt-ins, FB / Meta lead gen, 1st + 2nd contact-form opt-ins, lead scoring, nurtures for newsletter / low MOQ / future opportunity / no response / quote no response, the pipeline moves, WhatsApp chatbot opt-ins, TikTok welcome, cold-email follow-up; ~25 drafts incl. an outreach set from Jan 2026). 26 forms, 1,613 submissions in 12 months (1st Contact Form 50–90 a month, 2nd Sales Contact Form 25–48, the Meta lead form 67 in Jul + 165 in Aug 2026, Growth Labs landing form Jan–Mar). 3 calendars (Schedule a Call with Robert · with Dwight · the agency's personal) with 0 bookings in range. 2,152 conversations (phone type). Team: Alan Coughlan, Dwight Hijastro, David L (account admins), Famille Ipaw (agency admin) + 3. 40 custom fields, 67 tags, 31 funnels, 1 survey, no email campaigns | Audit next: stage hygiene (908 of 1,267 parked in No Response / Bad Fit / Future Opportunity), speed-to-lead, the two “Signed Up” definitions, who Robert is (the closer's calendar + pipeline), UTM capture on the new site's forms | |
| Photo & Video Bank (Google Drive) | Folder 1vt4DnVdkS8MRo0BJmiXpZhOB__AvNTvP (David, 2026-10-13), shared to david@davidsanghera.com; some sub-folders are not link-shared, so gdown stops at the first one — the Drive API reads it as the account (API enabled on the analytics-mcp Cloud project 2026-10-13). 6,640 files · 4,852 photos · 1,626 clips (3 h 51 min) · 70.8 GB — 3,821 unique (39.7 GB): 2,819 files are the same bytes under a second folder name (“US & China Videos 2025” = “USA & CHINA Footage”, “CN Warehouse Photos” = “Warehouse Photos”, “20million party” and “Christmas party” filed twice) and 140 are Premiere scratch. Two roots: Warehouse 2025 (CN Warehouse Photos + China Warehouse 2025, each by Factory · Label Printing · Office · Pick & Pack · QC · Warehouse · Tiktok Landing Page; China Loading; Factory Visit; US & China Videos 2025 — 20 Million Party, CN Warehouse, QC journey, QC-cosmetics, china-to-US walk-through, high five across the world, label printing, pack, script 1, other USA footage; the 2025/2026 parties, anniversary, SZ team trip) and China (Christmas 01, Portraits, CS team). Mostly iPhone 4032×3024 HEIC and 4K .MOV; shot Jul 2025 – Mar 2026, bulk Nov–Dec 2025, plus Dec 2024 Christmas. Catalogue clients/lansilglobal/assets/bank/CATALOG.md + catalog.json (tracked); bytes via clients/lansilglobal/ingest/drive_bank.py mirror into the MAIN checkout's clients/lansilglobal/assets/bank/files/ (git-ignored; duplicates hard-linked; scratch skipped) | Next: thumbnails per folder → a shot-selection board, David picks the hero shots per page | |
| Brand Assets (Google Drive) | Folder 1vEx0JkA_mE3PYM4jRE_LUg4weaG30-u5 (David, 2026-10-13): 113 files · 73 unique · 4.1 GB — Brand Guidelines (Lansil Global - Brand Guidelines.pdf, 64 MB), logo masters (.ai / .eps + Lansil-01…07 PNG at 4500×4500, Asset 1@4x 6299×1562), fonts Kanit + PP Neue Montreal (the latter is the free-for-personal-use cut — its EULA sits in the folder; a commercial web licence is needed before it ships on the site), collateral (invoice, quotation, pen, T-shirt, bomber), brochure (Lancil Brochure.ai / .pdf, 795 MB — the file name's typo is theirs — plus two Chinese brochure PDFs), box design, PPT template. Catalogue clients/lansilglobal/assets/brand/CATALOG.md; bytes in the main checkout's assets/brand/files/ (git-ignored) | Read the guidelines PDF before the design lanes (colours / type); confirm the hub accent #EE4127 / #231F20 against it | |
| Content Archive (Google Drive) | Folder 1Q5Ftn14bLWid90WNhGqNeaM_euC_vBeD (David, 2026-10-13): 44 docs, 2.8 MB — 27 published blogs (two parent guides: China Sourcing Guide, Ecommerce Fulfillment Guide; posts on choosing a US fulfillment company, the China warehouse, sourcing cities, DDP vs DAP, OEM vs ODM, MOQ, Temu, AliExpress, TikTok fulfillment, BFCM / peak-season checklists, 3PL basics…), 12 YouTube scripts (how to ship from China, quality control, choosing a China sourcing company, sourcing mistakes, the Shenzhen office tour, the affiliate program, weekly batches) and social calendars 2023–2025. Rendered to Markdown with tracked copies in clients/lansilglobal/content-archive/ (3.4 MB) | SEO: the 27 blogs are live URLs on the WordPress site — the new site needs a /blog/ (or 301s) before the DNS switch or that equity 404s — §9 | |
| Email platform / list | TO FILL (size, segments) | ||
| Netlify site (Lansil surfaces) | lansil-site (EcomScalers team, 30301728-1284-4894-a1ed-13f2ead15c85) — the NEW website, staging until the DNS switch, deploy only via products/lansilglobal/deploy.sh; custom domain rates.lansilglobal.com set 2026-10-18 (the calculator's own subdomain: its root serves the /pricing/ door, /lanes/ and /calc/ map onto /pricing/ — rules in netlify.toml + twins; Netlify needed a primary custom domain before any alias; the staging URL still serves) — DNS DONE 2026-10-19 — rates CNAME lansil-site.netlify.app TTL 300 added by David in hPanel's DNS Zone Editor (his Hostinger login has admin access to Alan's account); resolves at 1.1.1.1 / 8.8.8.8 within two minutes, Netlify answers 200 on http://rates.lansilglobal.com; HTTPS LIVE the same night: Let's Encrypt certificate CN=rates.lansilglobal.com issued ~19:31 UTC 2026-10-19 (valid to 2027-01-17, Netlify renews), force_ssl on (http → 301 https), every route 200 (/ door, /lanes/bill/, /lanes/wizard/, /calc/*, /assets/logo.png), the lead function answers on the subdomain, Playwright smoke on both test screens clean. The first provision call was 422 bad dns for custom domain two minutes after the add (Netlify's negative cache); the retry 60 s later was accepted and the certificate landed ~8 min after that. https://rates.lansilglobal.com/ is now the link for ads and the client board; the staging URL keeps serving. at the main-site launch lansilglobal.com becomes the primary and rates stays an alias; lansil-launch-board (EcomScalers team teo-h3uxfci) — the client-facing launch board, deploy only via clients/lansilglobal/deploy.sh; the new website gets its own site when built | ||
| Apify (the price check's lookups) | Lansil's own account lansil_global ("Lansil Global", support@lansilglobal.com), created for Lansil 2026-11-11 (sign-in sent by David; he changes the password). FREE plan: $5 of credit a month, 5 jobs at once. All three actors are pay-per-event, so nothing to rent. | Its API key lives ONLY in lansil-site's env LANSIL_APIFY_TOKEN (secret; production, deploy-preview, branch-deploy), never in the repo. It replaced the house account's key on 2026-11-11. | |
| Launch Center workspace | https://launch-center-hq.netlify.app/lansilglobal/ | console lane lansilglobal |
5 · Compliance ceiling
- Substantiate before use: 2–4 day US delivery · 5–10 day worldwide · 15M/20M orders · 1,492/2,000+ factories · 74/100+ staff · $150M revenue · "$1M credit / 60-day terms" (financing language — check Google/Meta financial-services ad policy).
- Tariff / customs claims ("de minimis suspended", Entry Type 86/11/01, "12.5% forced-labor duty", per-parcel broker fees) — cite primary sources (Federal Register, CBP) and date them; the landscape moves monthly. Never phrase sourcing savings as "suppressing declared customs value" — reads as undervaluation; say "lower factory-gate cost".
- Competitor references: Reddit sentiment on ShipBob/ShipMonk is real but must never be quoted or paraphrased in ads (defamation exposure) — frame positively ("no hidden fees", "named account manager") without naming them.
- No "only" / "#1" / "the only provider with both China and US" — NextSmartShip already has both (DR 02).
- No "automated / effortless / AI-routed fulfillment" language — it is the exact register this market distrusts (DR 03/05/06).
- Trademark-filing service = a legal-adjacent claim; confirm how it is delivered before advertising it.
6 · Decisions & taste log (dated; David's verdicts = ground truth)
- 2026-11-15 — Why WhatsApp messages fail in the CRM. David, on a screenshot of "Message blocked to maintain quality standards. Review Meta's Per-User Marketing Template limits": "Why does this keep showing up on GoHighLevel when messaging?" - The read (read-only, outbound WhatsApp, Oct 25 – Nov 15): 74 sent, 34 failed. 30 of the 34 failures were sent by workflows. - 13 × Meta's marketing limit (error 131049), 11 of them to +1 numbers. Meta's own page: "WhatsApp does not currently deliver marketing template messages to WhatsApp users with United States phone numbers." Elsewhere it caps marketing templates per person by recent engagement. The blocked templates: "Welcome to Lansil Global! Let's get started" (6), "Hi …, just checking in. Are you still looking…" and the other "checking in" follow-ups (4), "Following up on your inquiry" (2). - 13 × outside the 24-hour window: the workflow message "Hope you've been well. You previously…" goes out as free text, which WhatsApp allows only within 24 hours of the person's last reply. - 7 × not a WhatsApp number (3 US, 4 other; includes 2 of the +852 numbers). 1 × a template sent with an empty name. - Why it is rising: round one targets the US, so almost every new Google Ads lead has a +1 number, and every WhatsApp template in the follow-up flow fails for them. - What works for US leads: messages inside the 24 hours after they write first (the calculator's WhatsApp button opens that window), templates in a non-marketing category tied to their own request, SMS (if Lansil's US texting is registered) and email. Retrying a blocked template inside 24 hours makes Meta block that person for longer. - These are changes to Lansil's GHL workflows (Dwight's side; there is no workflow API) (§9).
- 2026-11-13 — Business name is filled with the lead's website. David, on a CRM screenshot where the first GA-route lead had none: "how is it possible that their business name is not being attached". - Lansil's convention: Business name = the website. All 13 recent 1st Contact Form leads have it, and so do the two calculator leads that went through "1st Contact Form Opt-ins" (Nov 13 01:43 and 04:36 UTC). The copy is a step in that workflow. - Why it broke: "GA Ads Opt-ins" has no such step. The first Google-ad lead on the GA route (Nov 13 16:58 UTC) had a website, an opportunity in New Lead for Dwight and its UTMs, but no Business name. - Fixed: the lead function now writes Business name = the website, copied as stored, on every create while the contact has none (production
6af7be8cb35b02866dbaa55c, live Fri Nov 13 5:51 PM PT). The price check gets it too. That lead's Business name was filled the same evening (read back). The six older leads have no website, so there is nothing to copy. - Lansil's side could add the same step to GA Ads Opt-ins, so that real submissions of the GA form get it too. - 2026-11-12 (night) — The earlier ad leads now carry their tracking. David, on a screenshot of the CRM's "GA Ads" contact list with the utm columns boxed: "The previous leads from the ads were not tracked properly". - Why they were blank: the 6 calculator leads of Nov 11–12 came in before the new lead function (07:57 UTC Nov 12). The old code saved the click id and UTMs only in our own
attr_*fields, not in Lansil'sutm_*and Google Ad Click ID fields. Those are the columns their list shows and the fields the GA form's hidden inputs fill. Which field the offline-conversion uploads (clients/lansilglobal/crm/STAGES.md) match a click on has not been checked. - Copied over the same night: on all 6, fromattr_*into utm_source, utm_medium, utm_campaign, utm_term, utm_content, Google Ad Click ID and gbraid (where the visitor had one), and Where did you hear about us = Ads. Only empty fields were written, each contact was re-read just before, and 6 of 6 read back correctly. - Two contacts in that list stay blank: the +92 one (Nov 11, 05:19 UTC) and the +65 one (Nov 13, 04:34 UTC), both tagged "google ads" on Lansil's side and with no email. Each is a WhatsApp chat that opened with the calculator's own button message ("Hi Lansil, I just ran your pricing calculator…"). A WhatsApp click passes no click id or UTMs, so there is nothing to copy. - Phones: someone on Lansil's side changed two of the five +852 numbers to +1 by hand (the leads of Nov 11 23:55 and Nov 12 06:48 UTC). Three still read +852 (Nov 11 01:02, 09:54 and 16:18 UTC); they were left alone pending David's word (§9). - 2026-11-12 (evening) — Calculator leads from a Google ad come in as "GA Form Ads FY26" leads. David, on a screenshot of the CRM's forms list: "please use this form for leads". - The form (
oUHrYbOQYEnLgfnc4tre) asks for the calculator's four boxes (Full Name, Website or Store URL, Email, Phone) and a required consent box: "I agree to receive email, SMS, and WhatsApp communications from Lansil Global, including service updates and marketing messages. Reply STOP to opt out." - Hidden fields capture the Google Ad Click ID (gclid) andutm_source,utm_medium,utm_campaign,utm_termandutm_content. - On submit it emails dwight@lansilglobal.com and redirects to lansilglobal.com/tell-us-about-your-business/ (step 2). - Its workflow is "GA Ads Opt-ins" (4e190309-7422-44ae-ba7e-52ababa8e0fe), created and published 2026-11-12 around 4:57 PM PT with a draft "GA Contact Form Opt-ins". The form had no submissions yet. - Done the supported way (production6af69d56ae8eb4859f1c2b08, live 2026-11-12 9:17 PM PT, §8): - A calculator lead with a Google click (gclid, gbraid, wbraid, or utm google / cpc) gets the form's name as its source. - It is enrolled in GA Ads Opt-ins, not 1st Contact Form Opt-ins. - It keeps the click id and UTMs the form's hidden fields would hold. - What it does not do: - The leads do not appear in the form's Submissions tab, and the form's own email to Dwight is not sent. MainStreet has no API that submits a form, and the form sits behind a bot check. - The form's required box records SMS and WhatsApp consent. The calculator never shows that line, so a lead cannot be posted as a submission of that form without recording a consent nobody gave. Asked David whether to add the line to the calculator's lock (§9). - Any other calculator lead keeps the old route (source "Landed cost calculator", 1st Contact Form Opt-ins). The price check is unchanged. The 6 leads from Nov 11–12 stay where they are. - 2026-11-11 (night) — The calculator's leads now say who they are. David: "It seems like Lansil had a lot of leads coming in from the cost calculator. Is there a way to pass more information about them into the lead, because right now it doesn't say much? Can you check what it says in the [MainStreet CRM]?", then "I dont think their store URL is passing into their profile" and "Where do the leads go in [MainStreet CRM]? Which workflow or whatever?" - What the CRM held (read 2026-11-12, about 07:00 UTC): 6 real calculator leads since the ads went live (Nov 11 01:02 to Nov 12 06:48 UTC), all from Google Ads, searching "chinese distribution companies", "china agent shipping", "logistics from china" and "pack and ship". Each had the name, email and phone, the tags, the
calc_*andattr_*custom fields, Lansil's workflow and a New Lead card for Dwight at $0. None had a note, a website or a company, and every one was filed in Hong Kong. - Phones were broken: the CRM reads a number typed without a country code as a Hong Kong one. 5 of the 6 were saved as +852 in front of a US number, so no call, text or WhatsApp could reach them. The sixth had typed +1. - The numbers say little: 5 of the 6 never touched the calculator, so their 1,000 orders, 1.4 items and 10 pallets are the page's starting numbers. 8 of the 10 tool contacts use a free mailbox, so the email names no store, and the calculator never asked for one. - Where they go (answered in chat): the contact (source "Landed cost calculator", tagscalculator-lead+website), then Lansil's "1st Contact Form Opt-ins" workflow, then a Sales Pipeline card at New Lead assigned to Dwight Hijastro, then their follow-ups. - Shipped the same night (production6af57157bdffffad63a01df2, §8): - Phones are saved with the country code. - The calculator has a Store website box, optional like the price check's. A work email's domain fills it when it is empty. - Where they are: city, state, country and time zone, from their connection. - Lansil's own fields are filled where they are empty. - Every calculator lead gets one note, which follows their numbers. - Asked David: fix the 5 broken numbers by hand, and give the 6 leads the note. - 2026-11-12 — all four boxes are required (David, on a screenshot of the form: "all of these are required"). The Store website box now has to hold a website before the bills unblur. An empty box, or text that is not a website, keeps the form locked with "A name, a working email, a phone number and your store's website, please." Production6af578731a58ad9d2662e541. The price check's box stays optional, because its product link usually names the store. - 2026-11-09 (afternoon, last) — Every opt-in from the calculator and the price check now joins Lansil's own lead flow, assigned to Dwight. David: "set up the GHL workflow for price-check leads, you did it for the rate card right?", then, asked how: "I just want whatever information is input in the opt-in form to be sent to GoHighLevel. Then that would just be turned into a lead and put into whatever existing workflow or whatever existing system they use for leads", and "Dwight" for who gets them. - The rate card had no workflow. The calculator-lead workflow was Dwight's open step since 2026-10-15 and was never built. The 4 contacts tagged
calculator-lead(Oct 19–22) are all TESTS, not prospects: - David's two: david@davidsanghera.com and david@lansilglobal.com. - Dwight's two: "Test" dghija@gmail.com and "TEST Gerald Hijastro" dghijastrotest@gmail.com. - So no real calculator lead has arrived yet (the ads start tonight), and none needs routing after the fact (corrected the same afternoon after David asked for the emails). - Lansil's own system: their website's 1st Contact Form feeds the workflow "1st Contact Form Opt-ins" (4c129c32-bfc8-4c40-86bd-f4681a36b3c4, published). Its leads arrive taggedinterest, with a Sales Pipeline opportunity at New Lead assigned to Dwight (47 of the last 60 opportunities were his), followed by their WhatsApp and email follow-ups. - Now:functions/lead.mjsenrols each NEW opt-in in that workflow through the API (POST /contacts/{id}/workflow/{id}; the integration's key allows it). - That covers the calculator's first unlock and the price check's sign-up. Slider updates, the price check's "result" and "pick" notes and?selftest=1never enrol. - A failed enrolment never fails the lead. -LANSIL_GHL_LEAD_WORKFLOWoverrides the workflow id; an empty value turns enrolment off. - Apify, David: "apify will need to be created by Alan, because it should be his account right". Yes: Lansil's own account and bill. Alan signs up and sends an API token, which replacesLANSIL_APIFY_TOKENon the site, then one redeploy. Done 2026-11-11: David sent the accountlansil_global(support@lansilglobal.com); its key was swapped in from his clipboard without being written down, then draft and production were deployed. A real lookup on each ran on Lansil's account, and the house account logged no run. - 2026-11-09 (afternoon, later still) — The supplier price check goes LIVE on Lansil's domain (David: "lets go ahead and deploy this to the lansil domain like we did with the other"). - Address: https://rates.lansilglobal.com/price-check/, on the subdomain the calculator runs on. - The switch:
LANSIL_PRICE_CHECK_LIVE=1was set for production first. - The deploy: production deploy6af222252e2bda4aaa0f99e0viaproducts/lansilglobal/deploy.sh. It also carried the one-line tracking guard from 2026-11-07, which only switches tracking off inside the Page Studio editor frame. The calculator pages are unchanged. - Timing: he gave the go before tonight's Google Ads launch, so the earlier "after the ads launch" timing advice is moot. - 2026-11-09 (evening) — The price check's headline says "up to 20%". David: "Pay 20% less than your supplier. => We need it to be up to 20% off type language as teh headline". - The headline and title read "Pay up to 20% less than your supplier." It is his approved line with only his two changes, "up to" and 20%. Copy chief:
products/lansilglobal/pricing/COPYCHIEF-price-check-hero-r2.md(U1 picked, 7/10, capped by "up to"; "product costs" killed under his earlier ruling). - The tool follows the headline back to 20%: the price is 20% under what they pay, or under the factory quote when they don't say or it is lower. It is rounded UP to the cent, so no quote is ever more than 20% under the price it starts from (checked on 13,432 quotes from $0.10 to $500; on a 14-cent unit the round-up makes it 14%). Example: other factories quote $1.50, Lansil $1.20, keep $150 a month at 500 units. - The one place the card beats the headline: a reader who pays more than the factory quote is priced from the quote, so their saving runs past 20% (pays $2.00, quote $1.50, Lansil $1.20). It is in the reader's favor. - The "already close" floor is unchanged: no price when 20% off what they pay would land under 70% of the quote. - LIVE on rates.lansilglobal.com/price-check/ (production deploy6af2a776143f72b6d21b52ba). It supersedes the 21% entry below: its headline, its math and its round-down. - 2026-11-09 (afternoon, later) — The price check promises 21%, not 20%. David: "Pay 21% less than your supplier. is a good headline". - The headline and title read "Pay 21% less than your supplier." - The tool's math matches it: the price is 21% under what they pay (or under the quote when they don't say). It is rounded DOWN to the cent, so no visitor ever sees less than 21% (checked on every quote from $0.10 to $500). Example: other factories quote $1.50, Lansil $1.18, keep $160 a month at 500 units. - The "already close" floor stays where it was: no price when 21% off what they pay would land under 70% of the quote. - This supersedes the 20% in the 2026-10-26 pricing rule. Truth Gate o1 (sourced cost about 30% under the marketplace price) leaves room for it. - He then asked: "What do you think about Save 21% On your Product Costs or something". Answer given: keep "Pay 21% less than your supplier." - It names the enemy the sub explains. - It is exactly the tool's math, per unit off the supplier's price. - "Product costs" reads as the whole landed cost (freight, duty, packaging don't drop 21%), and "Save 21% on…" is coupon register. - The honest cost-language version, if he wants one: "Cut your unit cost by 21%." - He confirmed: "keep Pay 21% less than your supplier" (2026-11-09). That is final; "Save 21% on your product costs" is not used. - Hero round closed in chat: the sub is B1's, the headline is his 21% line.
- 2026-11-09 (afternoon) — The price check's sub is decided: B1's sub, APPROVED in chat. David: "Your supplier might not be the factory, and you pay their markup on every unit. Paste a product link and see the factory price in about a minute. is good". - It's on the page under his headline "Pay 20% less than your supplier.", which stays (he picked no headline row). - Round 1's S8 is now DENY (superseded). - The tripwire's promise line follows the new wording. - Draft:
6af21e8763e2420b32adee6f. - The 27 headline rows of the hero round stay open; only the sub is decided. - 2026-11-09 (early afternoon) — David opens his own price-check headline and sub to the copy chief. He pasted "Pay 20% less than your supplier." and the middleman sub: "lets copy-chief these 2 headline and subheadline". Review:
products/lansilglobal/pricing/COPYCHIEF-price-check-hero-r1.md. - Scores: H1 7/10 (a strong control, "too good but not great"); sub 6/10 (its "if" hands every reader the exit "not me"). - The chief's sub: "Your supplier might not be the factory, and you pay their markup on every unit. Paste a product link and see the factory price in about a minute." - The chief's headline pick: W1 "Your supplier has a supplier." - Posted for his marks: 8 bucketed pairs plus the weird twenty, 28 rows as asklg-copychief-price-check-hero-r1on Copy Sheet HQ. These rows replace round 1's S8. - 2026-11-09 (midday) — Price check: Alibaba and AliExpress come out of the input note too, and the copy chief runs (David: "yes take Alibaba and Aliexpress out of the lane. And lets copychief this if we havent?"). The note now reads "A link from your store or Amazon works." Pasted Alibaba and AliExpress links still work; the page just doesn't name them anywhere. Copy-chief round 1 is written (
products/lansilglobal/pricing/COPYCHIEF-price-check-r1.md, 7/10 now, 8/10 with the top three) and posted as 13 APPROVE / DENY / FEEDBACK rows: asklg-copychief-price-check-r1on Copy Sheet HQ. - 2026-11-09 (late morning) — Price check: a progress bar after the sign-up, and the page never says Alibaba. David, on the spinner line "Matching your product to its factory on Alibaba…": "There should be a progress bar that shows up, starting from 0% to 100%. That way, we capture their information, and then they see this kind of loading bar until we have the results." Then: "I don't want to say that we're scouring Alibaba. You can say something like 'as listed from our Chinese factories,' 'as listed internally,' or 'as quoted from other factories.'" - The flow now: 1. The visitor pastes a link. 2. The lock asks for name, work email, mobile and website. 3. The lead goes to GoHighLevel as soon as it's sent. That first note says "still matching"; the result follows as a second note. 4. A bar runs from 0% to 100% through five steps: "Reading your product page", "Checking quotes from factories in China", "Comparing factory quotes", "Pricing it at N units", "Done." 5. The result opens. - Edge cases: a result that is already in still runs the bar for about two seconds. A returning visitor goes straight to the bar. A visitor who leaves during the bar is still a lead. - Wording on the page: "Other factories quote $X at N units · we buy it from the factory", the card label "Quoted from other factories", and the list heading "Closest quotes from other factories". - The one Alibaba left on the page is the input note "A link from your store, Amazon, Alibaba or AliExpress all work." It names links a visitor may paste, not where we search; it was kept and put to David. The CRM notes still name Alibaba (internal).
- 2026-11-09 (morning) — The supplier price check searches Alibaba by the product's NAME and what it does, and shows the matches (David, after his two tries on thesuperiormane.com both went to "by hand": "When I look for a product, it should go find what the product's name is or what the product does on the website's product link, and then search for it in Alibaba and return results. Why is it not doing that now?"). Why it failed: it searched by photo only, which matches shapes (the Eye Repair Stick → a makeup foundation stick), and that listing had no price tiers. - Now: the page's name, brand, category and description are read. A plain search phrase is made from the name, with the brand, sizes and sales words taken out. The area goes in front when the name lacks it ("Sculpting Serum" → "skin sculpting serum"). A model name with no product type is searched by type plus the description's words ("Tree Runner" → "shoe breathable lightweight"). - Two searches run side by side on Apify: by name (
blueskyscraper/alibaba-products-scraper, 20 listings with quantity price tiers, about $0.02, 3–5 s) and by photo (as before, $0.02). The pool is ranked by the search words in each listing's title, then by the description's own words, with a bonus when both searches agree. The price checks are kept, and the floor is lowered to 1% of shelf price, because skincare costs 1–4% of retail. - The page shows what it matched: up to four Alibaba listings (photo, short title, minimum order, price at their units), the closest one picked. A tap on another re-prices the page. The listing link and supplier stay server-side for the CRM note, which now also says what was searched, which match they priced and the other matches. A tap after sign-up adds one short note. - Results on the draft (?selftest=1), all about 9–12 s: - Eye Repair Stick → retinol eye sticks, $1.06 at 500 units, so Lansil $0.85. - Sculpting Serum → sculpting and peptide firming serums. - Allbirds Tree Runner → lightweight breathable running sneakers. - Wildflower iPhone case → magnetic iPhone cases. - Prime Face System ($99.99 bundle) → by hand. - 2026-11-08 (night, later) — The supplier price check (LG-T09) way A is picked and built (David: "Lets start building this out: https://launch-center-hq.netlify.app/docs/lansilglobal/tripwire-mocks/t09/a/ - with alibabas mcp, so it finds the right product,a nd then we can put the pricing down by 20%"). Way A = "Paste one link"; B and C stay mock-ups. - The Alibaba lookup: Alibaba publishes no MCP or product API (searched 2026-11-08; Accio is a closed app). The lookup runs on the server through Apify's Alibaba actors, which double as MCP servers: a photo search,
piotrv1001/alibaba-image-search-scraper(about $0.02, 7–15 s), then the matched listing's price ladder,blueskyscraper/alibaba-product-details-scraper(about $0.005, 3–5 s). A pasted Alibaba link skips the photo search. Billing runs on the house Apify account (FREE plan, $5 a month, about 200 lookups). - Pricing on screen (David's rule, plus three edges set by default and put to him): 20% under what they pay, or under the Alibaba listing at their monthly units when they don't say. When they pay MORE than the listing, the price is 20% under the listing. When they pay under 87.5% of the listing, no price shows; the page says "You already pay close to the factory price. We check it in Shenzhen and text you what we can do". A link that can't be read, or that matches no plausible listing, goes to the sourcing team by hand. - Leads: the GoHighLevel contact gets the tagstripwire-lead+price-check+website, source "Supplier price check", and a note for the sourcing team. The note holds the product, units, what they pay, the factory listing's link and supplier, and what the page showed. The factory listing never reaches the browser. The page sends no Google Ads conversion (AW_LABELS: {}), so round one's "Calculator Lead only" bidding stays clean. GA4lead_generateand MetaLeadfire with laneprice-check. - 2026-11-08 (night) — Google Ads round one: the go. David in chat: "i think it stime we deployed the google ads media plan for Round 1", then "Which lansil account are we using to launch on?" (answer: ECOM LG FZE 218-766-7313, Lansil's own account on Lansil's card, his pick of 2026-10-15; not the agency's 627-986-4143 or the manager 873-360-9807), then "Go for it." on four steps shown first: (1) a campaign goal "Calculator Lead only"; (2) both campaigns created paused; (3) Google's review read, then both switched on, stopping if anything is disapproved; (4) row 4d on the client board updated with the real start. - His answers in chat: run all 15 ad groups as written (the Copy-Chief rows were never marked); $100 a day as two $50 campaigns, US only, 14 days; the campaigns bid on calculator opt-ins only through the campaign goal, so the account's nine older primary conversions (thank-you page, Unbounce form, home-page view, three offline uploads + signed-up, two YouTube) stay untouched and the page-view demotion from the plan is not done. - Start: Monday Nov 9, about 7:30 PM Pacific. His first pick was noon Pacific = 00:00 on the account's clock (ECOM LG FZE runs on Asia/Dubai, so its budget day begins at 12:00 PT; 00:00 PT is mid-day there, which defeats the midnight rule's purpose). The house developer token turned out to be Explorer access (2,880 operations a day on a live account); round one is ~3,400 operations with both match types, a failed request counted against the allowance, and Google cut the key off for 23.6 hours mid-build. Given "the soonest start, ~7 hours into the account's day" or "Tuesday noon PT", he took the soonest. - Built 2026-11-08 night (inert, nothing can serve): the goal
customConversionGoals/6459856393, 12 assets (4 sitelinks, 7 callouts, the Services snippet), the shared negative listsharedSets/12307777267(932 exact). Monday 19:35 PT, by a detached runner in this session: both campaigns with every keyword in phrase match, Google's review read, switch-on, then the exact twins. Phrase covers the exact searches meanwhile. - The script's fixes (clients/lansilglobal/ads/google/build_round1.py): as first written it set the lead-form category biddable at campaign level, which would have counted the old thank-you page and Unbounce form too, so a custom goal replaces it; 49 keywords held characters Google refuses (commas, brackets, ?, %), now cleaned (punctuation doesn't change matching); negatives became one shared list (half the operations); every request is all-or-nothing and a stopped run resumes fromround1-created.json. - 2026-11-03 (night) — Home page, round one judged: Route and Depth are the best, and round two builds on them. David in chat, word for word: "So based off of the route 1 and the depth one, those are the best. The reason why I like the depth one is the big hero, the background images, and the images and the videos that are auto-playing on the hero. That's very nice. The route one I like is because of that line that's drawn along the page as you scroll down. In the other ones, I like it when the factory, QC company, and China 3PL are all animated. That's nice. I also like, on the depth one, the fact that there are people who are on our Lansil team doing work, like 2 to 4 days to the door and 5 to 8 days to the US doorstep. That's very nice. I think we should be able to apply these to the route one as the base, or make some kind of hybrid. Biggest feedback for the route is that there's a lack of real photos of people, heavy text, and a lack of images." - His taste, for every page of the site that follows: a big photographic hero with video playing on it; one drawn line that leads you down the page; the steps of the trip moving; Lansil's real people doing the actual work, large; few words on screen. - Measured at the verdict: Route shows 1,354 words and 14 pictures on a laptop, Depth 1,240 and 19. - Round two has four lanes. K keeps Route as the base (his first option). L threads the line through Depth (his second). M makes the team the campaign, on Lansil red, with the fewest words. N turns the line into a sideways trip through the pictures. - New to round two: 40 real team members cut out of their photos on this Mac (Apple Vision, nothing uploaded), so the line can pass behind them and they can sit on their own depth layer. Three team photos added from the bank, including the whole team at the 11th anniversary. - The copy rule for "heavy text": every sentence stays, word for word (135 checked), but sentences become captions on pictures, or fold away. Nothing is rewritten without David seeing it. - Built, checked and reviewed 2026-11-04. Words and pictures on a laptop: K 725 and 64, L 729 and 57, M 499 and 59, N 688 and 75. All four keep all 135 sentences and clear the checker. - Our read on the board: M is the strongest, K is the safe choice, N has the biggest single moment, L is the most polished but keeps the most words. David picks; his chain preference decides which round-one animation carries forward (K Isometric, L Control Tower, M a film in colour, N Product).
- 2026-11-03 — The home page explored in ten design lanes, judged section by section. Judged in chat, see the entry above. David's ask: "lets do the maximum design variations and boards for this home page … maybe 10 different lanes … from parallax to motion to stop graphics to visual imagery, animation … look like the page and site design was from a $100,000 agency". His choices: all three design resources, all ten lanes, and imagery from the bank plus AI shots for what the bank lacks. - Board: https://launch-center-hq.netlify.app/docs/lansilglobal/home-lanes/. Each of the 13 sections appears in eleven versions, on a laptop and a phone: today's page plus the ten lanes. The lanes are A Ledger, B Route, C Depth, D Manifest, E Control Tower, F Isometric, G Product, H Cinematic, I Evolution and J Comparable. - He can pick per section or take a whole lane. Picks go to the verdict log: board
lansilglobal--home-lanes, candidatesNN:<lane folder>, actionwinner; a section note is candidatesNN, actionnote. - Every lane uses every word of the live copy, in order: 32 lines checked word for word. - Every lane clears the floors: 16px sentences, 44px targets, 4.5:1 contrast, nothing hidden under reduced motion. Today's page fails 10, 15 and 26 of those checks. - Every lane went through a Web Interface Guidelines review, and its must-fix and should-fix items are applied. - Our read on the board: - B is the strongest whole page. - G sells hardest: it presents the landed-cost calculator as the product. - D is the most memorable. - Sections picked from several lanes get rebuilt in one look, set by the lane with the most picks. - Photo captions follow the bank's facts: US shots say "US warehouse", never Nevada or Pennsylvania, and no portrait is named. - The lanes use open-licence Google fonts. Whichever lane wins, the site's type is still open (§9: PP Neue Montreal is the personal-use cut). - 2026-11-03 — The media plan lives on Media Plan HQ, as the Funnel map (David: "a simple spreadsheet doesn't work" → four template lanes → "lane D." → "lets make a mediaplan-hq.netlify where we house these … update all the reference links and remove all the old boards"; the schedule board: "Move it into Media Plan HQ"): https://mediaplan-hq.netlify.app/lansilglobal/. Each round is the path a buyer takes (targeting → budget → ads → landing page → goal) with the counts on every step and the 30% page bar on the drop; the plan page names no other brand. Retired: the 2026-11-02 sheet (Copy Sheet HQ's Media plan tab), the Launch Center plan page and board (their addresses forward to the HQ), the design-lane boards.
- 2026-11-02 — Every media plan is a sheet, like the copy sheet (David: "Want the media plans made just like the copy sheet in a spreadsheet-like format so it's easier to read. Make sure to change all of the other media plans to be this way as well."). - Copy Sheet HQ has a ninth tab, Media plan: https://copy-sheet-hq.netlify.app/lansilglobal/?tab=mediaplan. - This plan is 59 rows: the three sentences, the 15 ad groups in each round (round 3 keeps 6), the settings, the 7 rules and the 5 decisions. The decisions are read-only there; they are answered on the plan's ask. - Each row takes 👍 keep or 👎 change plus a note. They are read with
python3 tools/copy-sheet-hq/mediaplan.py verdicts --brand lansilglobal. These are records only; the account still changes only on his go. - The plan's hub address is the same sheet, read-only, so the links to it keep working. - CellScience, NailScience and MGI's plans changed the same way the same day. - 2026-11-02 — The bill screen puts the cost per order at the top of each bill (David: "Move the per-order unit economics to the top of the bill so it's easier to read."). - Desktop: each paper bill shows a per-order box under its name, before the line items. At 1,000 orders that is $2.50 against $4.01. The footer keeps Total and Per day. - Phones: the single receipt shows the two per-order figures side by side under the two names, above the lines. - The figures sit behind the lock like the rest of the bill. - Tested unlocked at 1280 and 390 wide from a saved test lead, with every outside call blocked, so no lead and no pixel event was sent.
- 2026-11-02 — Hire the organic lane's community lead now; the role also messages operators on LinkedIn (David: "Need someone from the "how one tool gets amplified" organic community lead that is in every single group (Reddit, Ecomfuel, pretty much anywhere there are D-to-C people) and also messages them on LinkedIn as well. I need someone like that. What kind of job description is it?"). - The role: community-led growth, half community marketer and half LinkedIn prospector. Posted as Community Growth Lead — Ecommerce. - The job description is
clients/lansilglobal/hiring/community-growth-lead.md. - Duties: the communities, one-at-a-time LinkedIn messages, the referral bench, client stories, a weekly read of calls by channel. - Starting 90-day targets: 10 qualified calls a month, 5 referral partners, 2 client stories a month. - Pay: a base salary, plus a bonus per qualified call that shows up and commission per signed client. - Applying: three community posts, the groups they're in and one LinkedIn message. No resume. - EcomFuel lets in store owners only. No vendors and no pitching (its own site, read 2026-11-02). - The one exception it names is a veteran provider who comes to give, not sell. That would be Alan, not a new hire. - A vendor's ways in are the member-written Review Directory (clients who are members review Lansil) and event sponsorship (ECF Live, eComm Camp, meetups). - So the hire has to already be inside the groups, for example a store operator with real standing. That is the first screen, ahead of marketing experience. - David's calls, the same day (David: "Lansil hires, they report to me, ignore base pay on. the JD, Alan's profile"): - Lansil hires and pays (Alan 2026-10-12: "We can be hiring all those people once we start getting leads in."). - The hire reports to David. - No base pay on the job description. The pay section names the bonus per qualified call that shows up and the commission per signed client, and says the full package comes up when they talk. - The LinkedIn messages go out from Alan's profile, in Alan's voice. The job description says so, and the application asks for the message they would send as Alan. - Needed from Alan: access to Alan's LinkedIn for the hire, on one fixed device. LinkedIn checks sign-ins from new devices (the code goes to Alan's email) and its terms don't allow shared logins, so the volume stays human. No login details go in this file. - The amplify diagram on Alan's board still lists LinkedIn only under the content team (posts in Alan's voice). It has not been changed. - Published the same day (David: "Can you publish this to the external site of lansil, so i can link it to them"): https://rates.lansilglobal.com/careers/community-growth-lead/ (§8). - 2026-10-26 — Two notes on the mock-ups, applied the same day. (1) The delivery map shows the postage both coasts save (David: "This ship from both coasts should give us now a way to get cheaper shipping rates in there too right?"). Yes: two warehouses mean fewer zones, so the same pass-through rates cost less per order. All three map mock-ups now carry a parcel-weight pick (under 1 lb · 1–3 lb · 3–5 lb) and postage per order from both coasts against one warehouse in Pennsylvania (A: the monthly amount kept; B: today's warehouse against both coasts; C: the pasted orders' total and the amount an order). The copy frames the saving as the shorter trip at the same carrier rates, passed through with no margin, never a cheaper rate, so it stays inside Truth Gate p5/f11 (postage is a pass-through, never a differentiator). Rates are SAMPLE (ground-parcel shape by zone, $4.75–$14.80); the live tool needs Lansil's DHL eCommerce rates by zone and weight. The bill calculator keeps postage equal on both sides on purpose: it compares against a typical 3PL, and many large 3PLs already ship from several warehouses. (2) The supplier price check says "Pay 20% less than your supplier." (David: "for the Supplier price check, oyu can say pay 20% less than your supplier. youre being taken advantage of because of the middle man or something along those lines. and they can give us their store link"). All three now open on that line with "If your supplier isn't the factory, you're paying a middleman's markup." A takes a product link from their own store (or Amazon, Alibaba, AliExpress) plus an optional "what you pay per unit", which sets the 20% when given; B prices three factory matches; C is now "your store link" (was "your product list"): we find the store's products, they untick what they make themselves and set units a month, and see the total kept a month; its sign-up asks only name, email and mobile because the store link is the website. The comparison basis stays the Alibaba listing unless they say what they pay.
- 2026-10-26 — Four tripwires to build, mock-ups first; the supplier price check prices off Alibaba at 20% under (David: "lets do every one except invoice teardown. show me mock-ups first for feedback" · "for supplier price check, we can use Alibaba MCP to source the items and price them 20% less"). The invoice teardown (T02) is out of this round (parked on the tripwire board as his chat call). The four: supplier price check (T09), delivery-days map (T07), credit-terms check (T03), freight estimate (T05). T09 changes shape: the visitor's product is found on Alibaba and Lansil's price shows on the spot at 20% under the Alibaba price; only the landed cost goes by text. Note the gap with the Google ad line "Sourced 30% Under List Price" (B5, from o1's 30%): the inputs card asks whether the ad moves to 20%. No Alibaba MCP is connected in this session; the live tool needs a server-side Alibaba lookup. Mock-ups: twelve clickable pages in the live bill screen's look (
tools/brand-hub/docs/lansilglobal/tripwire-mocks/<tool>/<a|b|c>/, sharedkit.css/kit.js,map.js+zones.jsfor the map,freight.js), three ways per tool that differ in how the tool works: T09 paste one link / describe it / your product list; T07 the map first / today against both coasts / paste your ZIP codes; T03 the pre-approval card / the cash-flow timeline / three questions; T05 the carton calculator / pick what you ship / deadline first. Sample numbers flagged on each: Alibaba prices, zones from state population centres (both warehouses put 59% of Americans within zone 4, against 40% from Pennsylvania alone and 19% from Nevada alone), the credit figure at twice monthly factory spend up to $1M, freight at $150–$250 per cubic metre by sea and $6–$9 per kg by air. Audited at 1280 and 390 wide: no sideways scroll, nothing under 13px, no contrast or tap-target failures, no console errors. Boardlansilglobal--tripwire-mocks(new kindmocks,tools/brand-hub/mock_board.py: PICK one way per tool + feedback notes), pointerlg-tripwire-mockson Needs You; the inputs cardlg-tripwires-inputsupdated to the four (+ the 30%→20% ad question). - 2026-10-25 (evening) — The media-plan row goes on Alan's board (David: "publish the media plan row too"). Row 4d unwrapped and deployed through
clients/lansilglobal/deploy.shwith two accuracy fixes: "bidding only for calculator opt-ins" (Google charges per click, the row had said "paying only for") and "932 off-target search terms blocked" (not "searches"). Chip still "Queued for Nov 3"; the start stays "on the go". - 2026-10-25 (evening) — Five tripwires picked to build (David: "i judged the tripwires"; marks on the board, no notes): T02 invoice teardown, T09 supplier price check, T07 delivery-days map, T03 credit-terms check, T05 freight estimate. The other fifteen left unmarked = not this round (the warehouse tour, the store shipping check, the switch planner and the presets among them); they stay on the board. Build order: round one T02 + T09, round two T07, round three T03 + T05 (Alan's numbers). Correction to my catalogue: T07 is not input-free — only "2–4 days anywhere in the US" is on the Truth Gate (f8), so the map shows distance zones unless Lansil gives days by zone. Inputs card
lg-tripwires-inputs; folded withtripwires.py verdicts(tripwires.json → picked,verdictper tripwire); the board's note names the picks. - 2026-10-25 (later) — The diagram for Alan is lane A, live on his board (David: "lets go with A, put it on alans board"). https://lansil-launch-board.netlify.app/amplify/ — three engines (paid ads by David, content with Alan on camera, organic by a community lead) converging on the tool, the follow-up chain, the loop; a row "How the marketing works, in one picture" under "What happens next". Deployed through
clients/lansilglobal/deploy.sh(hygiene PASS, /amplify/ 200, noindex header). The held media-plan row 4d did not ship: it is wrapped in an HTML comment in the source, which the hygiene gate strips, until his go on the media plan. - 2026-10-25 (later) — The tripwires get their own board, written for reading (David: "make the tripwires on a different board easier to read"). Board
lansilglobal--tripwires(kindtripwires,tools/brand-hub/tripwire_board.py, built bytools/site-studio/tripwires.py build): one card per tripwire in the plan's build order (live · round one · round two · round three once Alan's numbers are in · more · the weirder ones), 16px body, 19px promise, 13px labels at the smallest, 48px buttons, filled red darkened to #CD3721 for 5:1 under white text; plain group names ("3PL companies", "Fulfillment centers"…) and the share of each group's searches per tripwire with its biggest searches; short no-fit reasons. Audited at 1280 and 390 wide: no sideways scroll, no contrast or tap-target failures. The Site Studio section is now a pointer ("Judge the tripwires"), the in-tab cardlg-tripwires-r1is withdrawn (no marks were on it), the five decisions moved to Needs You. - 2026-10-25 — Tripwires per keyword, TO JUDGE (David: "Need to create a list of trip wires per keyword that would be able to get their phone number, order volume, and website information. When I say per keyword, I mean per keyword in the media plan, based on the group the keywords are in. Trip wires would be like how we made the pricing calculator. Some more ideas like that, so we can start working on ways to capture people"). Twenty tripwires (LG-T01 the bill calculator, live, to LG-T20), each asking for the mobile number, the orders a month and the website because the tool needs them: the website is the input or what gets checked, the order count sizes the answer, the phone is where the answer is texted. Every keyword in the Google Ads plan (1,364 across rounds 1–3; 1,183 in round one) gets its group's tripwire unless its words point to a better one. Split by searches a month: bill calculator 713 keywords / 118,721 (61%) · switch planner 61 / 18,757 (round 2, rival names) · warehouse tour 140 / 16,546 · supplier price check 11 / 12,280 · freight estimate 63 / 3,430 · dropshipping-or-US-stock 8 / 2,198 · store shipping check 61 / 1,836 · Shenzhen receiving address 27 / 1,736 · rate per product 86 / 1,056 · delivery-days map 82 / 1,014 · FBA prep quote 17 / 457 · packaging price 13 / 40 · no tripwire fits 79 / 15,326 (round one: 34 / 6,450 — "3PL Central", "4PL", "process" searches, micro-fulfillment, print-on-demand). His Oct 12 ideas are in it: the credit-terms check (LG-T03) on a new inventory-loan keyword group (FIN-INV, landing on /financing/), the rate per product (LG-T04). Build order = the plan's calendar: the website asked right after the calculator unlocks (now); warehouse tour, supplier price check, invoice teardown (round 1); calculator presets, store shipping check, dropshipping-or-US-stock, switch planner (round 2); the China-lane quotes once Alan's numbers exist (round 3). Files:
briefs/lansilglobal/site/tripwires.json(catalogue + rules) →tools/site-studio/tripwires.py build --brand lansilglobal→tripwires-by-keyword.csv+tripwires-map.json; planbriefs/lansilglobal/site/TRIPWIRES.md. Askslg-tripwires-r1(BUILD / PARK / FEEDBACK per tripwire) andlg-tripwires-decisions(D1 website after the unlock · D2 add the inventory-loan group · D3 drop the 34 no-fit round-one keywords · D4 a group under 5 opt-ins per 100 clicks moves to its own tripwire from round two · D5 Lansil's team runs the human tripwires), both on the Site Studio tab. Tripwire names are working lines; each gets the chief pass when it is built (CL-21). - 2026-10-25 — The diagram for Alan: how one tool gets amplified, three lanes to pick from (David: "create a visual diagram about how a piece of content, like a tripwire, is for Alan, the client, so that we can visualize and show Alan how it'll be amplified: How we market things: content marketing · How it's going to be through paid ads, like Facebook and Google. That's going to be by me. · How there's the organic stuff, which has its own categories, like through community, like Reddit, Facebook groups, and professional communities like Ecom Fuel, referral, and UGC. … Those people would be a different person doing it."). Lanes (HARD design rule, three skills): A the streams (minimalist: three engines converge on the tool, then the follow-up chain), B who does what (Swiss print: owners × stages, the tool as the column every lane runs into), C the flywheel (soft premium: make, reach, capture, close, prove). One set of facts on all three: paid ads = David from day one; content = an editor and a writer with Alan on camera, and organic = a community lead, a separate hire (community, referral, UGC), both hired as the leads come in; the follow-up = the setter (David takes the first calls), Dwight on the CRM; every lead keeps its channel. Audited at 1280 and 390 wide (no sideways scroll, nothing under 13px except lane A's desktop figure labels at 12.8px, no contrast, target or console failures). Chief pass
clients/lansilglobal/launch-board/COPYCHIEF-amplify-r1.md(the setter not "a Lansil rep", "can be" a testimonial, "community lead" not "separate person"). Boardlansilglobal--amplify-lanes, asklg-amplify-diagram; my recommendation A. The pick goes on the client board at/amplify/with a row, only on his verdict. - 2026-10-12 (pulled 2026-10-25) — Nine hub answers that had never reached the lane (pulled with the main checkout's
pull_answers.py): economics — "We need to think of this as Cost per Lead. Because these clients once they are onboarded have an insane LTV. So the idea is getting volume of leads to qualify and nurture and get on phone calls with so we can understnad if they high volume orders or low volume orders." · priority service — USA Order Fulfillment (fill NV/PA) · first content build — the Top-5 ranking ("best China-sourcing fulfillment partner"), search + retargeting · rate card + guarantees — "We should be putting a YOU QUALIFY for the 60-DAY CREDIT TERMS of $X Dollars. Check to see if your company qualifies for credit terms angle as part of the calculator. We should also be bidding for inventory loan keywords so that we can capture audiences looking for inventory loan. Take calls with them. That way, we can convert them over. We can make an inventory loan calculator to see if you qualify, kind of like what credit card companies do, along with a rate calculator. If people who are looking for new 3PL companies put in their product information, we will give them a rate per product." · Alan on camera — yes, video and LinkedIn in his voice · approval — David decides, Alan reviews weekly · hiring — "We can be hiring all those people once we start getting leads in." · own brands as case studies — no · the Truth Gate round — folded intruth-gate.md(the "one with both" positioning is usable again because NextSmartShip does not own its warehousing; the 60-day method needs a catchier name; x4/x5/x8 FALSE that day, superseded by the 2026-10-20 confirmations). - 2026-10-21 — Google Ads media plan for the calculator, built from the Keyword Studio verdicts and the rivals' ads (David: "lets setup a media plan for this rates.lansilglobal.com for Google Ads, include the keywords based on the competitor keywords we judged, and ads based on the competitor ads, using this rates.lansil page") — WAITING ON HIS GO, nothing in the account.
briefs/lansilglobal/media-plan-google-2026-11.md+.json→ plan page https://launch-center-hq.netlify.app/docs/lansilglobal-media-plan-google-2026-11.html. Two search campaigns in ECOM LG FZE at $50 a day each (A: US 3PL & fulfillment = US-3PL · US-FUL · US-OPS · US-NEAR · US-SHOPIFY · US-PRICE (TEST) · US-COAST; B: the China lane = CN-CORE · CN-DROPCN · CN-CITY · CN-DROP · SRC-DROP · FR-CNUS · FR-AGENT · FBA-CN), phrase + exact only (1,181 keywords), every SKIP group as exact campaign negatives (932), Calculator Lead the only goal, maximize conversions 14 days then tCPA, US presence, one RSA per ad group (3 own headlines + a 12-line spine + 4 descriptions), sitelinks on four live lansilglobal.com pages, landingrates.lansilglobal.com/lanes/bill/with the UTM suffix. BR-SHIPMONK + BR-OTHER (BID) wait for a round-2 rival-name campaign; the TEST groups US-NV · US-PA · FR-US · MFG-PACK for round 3. Rules: no verdicts days 1–3; $150 with no opt-in = pause; 2+ opt-ins under $75 = more budget; converting search terms become exact keywords; $30 no-opt-in terms become negatives; under 5 opt-ins per 100 clicks after 200 = the page changes first; nothing edited live. The rivals' Google previews (decoded 2026-10-21) sell rank and adjectives (ShipBob "#1", ShipMonk "Guaranteed SLAs") or a vague number (NextSmartShip "40% Saving") — no US rival shows a price, so every Lansil ad does ($2.50/order, thirteen $0 lines, 26-31%, NV + PA, 2-4 days); rival names never in ad text. Chief pass r1 applied (clients/lansilglobal/ads/google/COPYCHIEF-google-ads-r1.md, 8/10 capped by the format: the terms headline scoped to the factory invoice, "Shenzhen Fulfillment Center" out (4PX, f7), the dropship line sells US stock, "at cost" → "no markup", the wa.me sitelink out). Build =clients/lansilglobal/ads/google/build_round1.py(dry run OK;--gocreates everything PAUSED, only on the go). Asks:lg-google-media-plan-go(go + date · cap · US only · demote the page-view conversion · the Mon/Thu reads) andlg-copychief-google-ads-r1(19 rows on the Copy-Chief tab). Client board row 4d written, not deployed until the budget and date are his. - 2026-10-21 — The calculator runs on the bill screen alone for now (David: “which two are we going to split test? Maybe lets just run …/lanes/bill/ this one for now”): the door at
rates.lansilglobal.com/sends every visitor to/lanes/bill/; the wizard, slider and dashboard stay built and reachable by URL, off the door; the scoreboard keeps counting the bill screen. Google Ads events for the calculator = the go (“wire these up to Google Analytics and … Google Ads events”): two conversion actions in ECOM LG FZE — Calculator Lead (website, Submit lead form, primary, one per click) on the unlock and Calculator WhatsApp click (website, Contact, secondary) on the button — fired by the runtime'sAW_LABELSmap, deduplicated on the event id. GA4 already receives page_view · calculator_play · form_start · lead_generate. - 2026-10-21 — The bill screen, David's own words (Copy Studio, applied verbatim per CL-12): H1 “Our Pricing vs. Their Pricing.” (first “Our Pricing vs. Their Pricing Calculator”, trimmed the same night); hero line “See why our 3PL is 26-31% cheaper than any other 3PL in the country.” (first “Save 26-31% of your bill with our startup-friendly pricing…”, replaced the same night); the lock opens “Your pricing is ready!” over “Just let us know who you are so we can see if there are any more specialized discounts we can offer.”; the small print under the button deleted; the sources paragraph (“Where the other column comes from…”) removed (“remove this”). The CTA block = ONE button that opens WhatsApp (“we need this to open WhatsApp on their phone if they have it - remove Get a Quote button”):
wa.me/17024622035with a prefilled “Hi Lansil, I just ran your pricing calculator at N orders a month. Can we talk?” (N follows the slider), the number under the label, WhatsApp green with ink text (contrast floor), aContactevent (browser + CAPI) on the click; the Get a quote button is gone from the block (the nav still carries it). The toggles sit on one row on phones (“all of these on one line on Mobile”): a grid at ≤620px, labels wrap inside their cell, 14px, 44px tap height. The Q4 month toggle is removed (“remove Q4 month”) — two toggles remain, an insert in every box and stock past 90 days; the Q4 uplift stays in the engine, just not switchable on this screen. Lansil's WhatsApp line = +971 52 674 7824 (David 2026-10-21, the Dubai office number); the button openswa.me/971526747824and shows that number — the published +1 702-462-2035 stays the phone line in the bar and footer. Open: whether 26-31% follows the page's own number (38% at default inputs; o3 says ~40%), and whether the lock wording and the sources removal mirror onto the wizard, slider and dashboard (the split test currently compares a changed bill screen with an unchanged wizard). - 2026-10-20 — David answers Alan's questions 1–23 (ground truth; Alan's sign-off on the rate card = launch check): sourcing 30% under Alibaba · fulfillment 40% under the US average · rate card confirmed · 60-day terms = factory invoice + freight, not fulfillment, anyone qualifies by order volume · guarantees: no hidden fees, no exit fees, no postage markups, free rework on QC failure · founder story confirmed · eleven operations facts (BRAND §1, Truth Gate f14–f21) · proof (Q19) and customers on camera (Q20) become their own programs — client-board pages
/proof/and/casting/· Alan on camera, staff cast from the team · Trustpilot usable. Open: which company figures to use (Q18), questions 24–34. - 2026-10-20 — Google Ads destination = ECOM LG FZE 218-766-7313 only (David: “ignore 166, the one we use is 218 766 7313”) — and AW-16666567492 turned out to be that very account's conversion-tracking id once the account became readable that night, so the Google tag's Ads destination is wired on the calculator after all. The Lead conversion for the calculator = a conversion action in ECOM LG FZE (website Lead on the unlock through the runtime's
AW_ID+ label, or an import of the GA4 key eventlead_generate), created only on David's per-action go. - 2026-10-12 — Sitemap v2 judged: 21 of 22 pages KEEP, nothing split, merged or cut (one dropshipping page, one compare page); LGS-14 /compare/shipmonk/ not judged. No notes.
- 2026-10-14 — Blur everything until Name, Email and Phone (David). The calculator is gated first; the lead is created in GoHighLevel at the unlock with the numbers on screen and kept current as they play (orders, the range explored, items, pallets, the Lansil month, the typical-3PL month, the saving, the lane, who ships today). Dwight's side (§9): write scopes on the Private Integration, the
calc_*fields, a workflow on thecalculator-leadtag → instant text + the setter. - 2026-10-20 — Tracking built on the calculator (David: “build the tracking”). Pixel 2.0 browser + Conversions API with shared event ids (PageView · CalculatorPlay · Lead on the unlock, full hashed PII on the server Lead), first-touch attribution (UTMs, gclid/gbraid/wbraid/fbclid, fbc/fbp, device id, landing, referrer) on every CRM contact in 17 new
attr_*fields, the house relay at/api/capithrough a re-export in the site's functions folder. Proven headed (dedup) + headless (CRM) on the live address, test contacts deleted, counters reset. Found on the way (corrected the same evening):gtag/js?id=G-GDFTX1H761404s at Google, which first read as a dead id — it is not: the property receives the site's traffic through Site Kit's Google tag GT-MJM9PB4Z, and only the standalone loader is unserved. Phase 4 EMQ read due 2026-10-21. - 2026-10-21 (evening, +1) — Google Ads conversions live for the calculator: after the Google re-login, two conversion actions created in ECOM LG FZE (Calculator Lead primary, Calculator WhatsApp click secondary), labels in the runtime, verified with a headless opt-in and a WhatsApp click (both pings carry their label). Meta + GA4 + Google Ads + the CRM all receive the unlock now.
- 2026-10-21 (evening) — Door → bill only (test paused, David's call). Runtime:
AW_LABELSmap fires a Google Ads conversion for any standard event that has one (Lead, Contact). The conversion actions themselves wait on the Google re-login (the refresh token hit invalid_rapt again, a day after the last one) — created the moment it is back, labels into the runtime, verified with the headless opt-in. - 2026-10-21 (later) — Copy Studio loop on the bill screen, three batches in one evening: six of David's edits applied verbatim and live (H1, hero line, lock title + line, small print off, sources paragraph off), replies posted, Studio re-extracted after each deploy. §6 holds the words.
- 2026-10-21 — COPYCHIEF round 1 on the calculator (
products/lansilglobal/pricing/COPYCHIEF-calculator-r1.md; bill 8 · wizard 7.5 · slider 7.5 · dashboard 6; 15 rewrites) and the Copy-Chief tab on Launch Center HQ (David: “make a Copy-Chief section so there is context like the copy studio”): every rewrite inside its Copy Studio section with the current line highlighted, the chief's note, the rewrite and APPROVE / DENY / FEEDBACK — the card is the console ask itself (lg-copychief-calc-r1, the Needs You lane points to the tab). Approved lines ship to both test screens in one deploy. - 2026-10-20 (night, +3) — Copy Studio on the calculators: markers on the four lanes, a
renderedextract mode (live page + local markers), four screens on the Studio with ?selftest=1 view links; hub deployed. - 2026-10-20 (night, +2) — Pipeline stage glossary from David (Q26, half): twelve definitions mapped to the live stages and to the offline conversion chain →
clients/lansilglobal/crm/STAGES.md. Open: an SLA per stage; the two Signed Up stages; Q18 figures; Q24–25, 27–34. - 2026-10-20 (night, +1) — Facebook Page shared:
1145315282000023“Lansil Global” on the adscaler system user with advertise rights (David). With the ad account, the pixel and the Page in hand, Meta ads can be built; first creation = a dry run to confirm the ad account can use the Page. - 2026-10-20 (late night) — Alan's questions 1–23 answered by David. Truth Gate: o1 → 30% under Alibaba, o3 → 40% under the US average, o4 corrected (terms = invoice + freight, not fulfillment; anyone qualifies by volume), x4/x5/x8 confirmed, f11–f23 added. BRAND §1/§3 updated. Two new programs on the client board:
/proof/(names, testimonials, three case studies, one set of company figures) and/casting/(five customers on video, Alan, three staff roles). Board rows 8, 9, 11 → received. Open: Q18 (which figures), Q24–34. - 2026-10-20 (night) — ECOM LG FZE readable (David accepted the invite). Its conversion-tracking id is 16666567492 — the “unknown” second destination in Lansil's Google tag was ECOM LG FZE all along; the morning's GTM note (“enhanced conversions point at a different account”) and the evening's muzzle were wrong on that point and are corrected: the Google tag runs plainly on the calculator (GA4 + ECOM LG FZE linker, enhanced conversions, remarketing). The agency account's conversions (AW-18265417989) are what the GTM conversion tag fires on the WordPress site. Full account read pulled (
ads/google/2187667313/2026-10-20/): 47 campaigns all paused or removed, $44.9K over 15 months, page-view conversions inflating 2025. Next (ask-first): the Lead conversion action in ECOM LG FZE for the calculator unlock. - 2026-10-20 (late) — David's answers to the status: ECOM LG FZE invite “coming soon”; AW-16666567492 ignored, 218-766-7313 is the account; the Facebook Page share is still outstanding (the ad account and pixel are shared, the Page is not — no ad can be created without it).
- 2026-10-20 (evening) — GTM + GA4 access landed (Dwight's add; David's screenshot; gcloud re-login). GA4 property 465005552 read (
tracking/ga4/465005552/), GA4 wired on the calculator through the Google tag with ad consent denied so the tag's second destination — Google Ads AW-16666567492, an account nobody has named — gets no cookies, no form data, no click ids (it had started receiving the opt-in form's data on the first try).lead_generate= their existing key event. Asks for Dwight (board 4c): say what AW-16666567492 is and remove it from the Google tag if it is not Lansil's; import the GA4 key eventlead_generateas the Lead conversion in ECOM LG FZE (GA4 is already linked to it). The Meta pixel's own log explained the headless silence: “Bot traffic detected” — fbevents.js drops its beacon under automation; the browser side is proven headed.
- 2026-10-20 — “Did they actually put the lead in the CRM, and the attribution on Facebook and Google?” CRM: yes, proven live (contact + nine fields + the update path, test deleted). Attribution: no —
rates.lansilglobal.comcarries no pixel, no Google tag, no GTM, no UTM/click-id capture, and the CRM contact hasattribution: null. The current WordPress site's container GTM-WH5KZ45S inventoried (clients/lansilglobal/tracking/gtm/2026-10-20-GTM-WH5KZ45S-live.md): a page-view “Lead” conversion, enhanced conversions aimed at another Google Ads account (16666567492), GA4 G-GDFTX1H761 through two event tags, two old Meta pixels hard-coded in the theme, Pixel 2.0 absent. The §9 tracking build is the next thing the calculator needs before a dollar of paid traffic. - 2026-10-19 — The calculator's subdomain is wired.
rates.lansilglobal.com→lansil-site.netlify.app(CNAME, TTL 300) added by David in hPanel's DNS Zone Editor after the Hostinger plugin route died (the domain is registered at eNom and only points at Hostinger's nameservers, so the Hostinger API answers 403 for the zone; the official pluginhostinger@claude-plugins-officialis installed and signed in — useful for hosting/WordPress reads, useless for this zone). Resolves publicly within two minutes; Let's Encrypt certificate issued by Netlify ~10 minutes later, HTTPS live, http → 301 https —https://rates.lansilglobal.com/is the calculator's address from tonight. The whole zone inventoried inclients/lansilglobal/dns/2026-10-19-zone.mdfor the cutover (no apex DMARC; old Unbounce funnel hosts; the CRM'sreplysending domain). GTM MCP authenticated the same night: five accounts visible, none Lansil's — the GA4/GTM add to david@davidsanghera.com is still owed by Dwight. - 2026-10-18 — The calculator goes live on its own subdomain, with the real bar (David: "lets integrate them into CRM and lets actually create a subdomain on the netlify, and the logo nav bar and stuff"). The two screens stop being prototypes: the brand kit's horizontal wordmark in the bar (
products/lansilglobal/assets/logo.png+ a white cut for the dark bars, 720 px fromAsset 1@4x.png), the phone and a Get a quote button that goes to the current site's quote page (lansilglobal.com/get-a-quote-for-us-3pl/, the only one that exists today), the three addresses + contact in a footer, the “Prototype · lane” badges gone, the lock's small print now a promise Lansil can keep today (“a Lansil account manager follows up by text or email within one business day”). The home page bar carries the same logo. The subdomain isrates.lansilglobal.com(on the A10 angle) — one CNAME from Dwight and ads can send people there before the main site moves. CRM: already live (every opt-in writes the contact). - 2026-10-18 — Two calculator screens go live as a split test (David: "I don't know which quiz will convert better, so we can set 2 quizzes live to see which one has a better opt-in rate").
/pricing/is the door: each visitor is sent to one of two screens at random, sticky per browser (localStorage),?v=bill|wizardforces one; the screen counts itself once per browser as “shown”, the form unlock counts as the opt-in; counts live in Netlify Blobs on the staging site (numbers only),GET /.netlify/functions/lead?stats=1serves them, and the Site Studio tab shows shown / opted in / rate per screen. The pair: bill + wizard (my default — David has handled both; swap by editing thelaneslist inproducts/lansilglobal/pricing/index.html). Our own checks pass?selftest=1and are not counted. No traffic reaches it until the site and the ads go live; nothing means anything before ~100 opt-ins per screen. - 2026-10-15 — The calculator is live into the CRM. Dwight added write permission to the Private Integration; the
calc_*fields exist; the lead function on staging creates and updates GoHighLevel contacts for real (verified with a test contact, then deleted). The remaining piece is Dwight's workflow on thecalculator-leadtag. - 2026-10-15 — The relaunch runs from ECOM LG FZE (218-766-7313), Lansil's own account paid on Lansil's card (David: “please use the ECOM LG FZE account”, on Dwight's note listing it next to the agency-owned 蓝斯尔贸易-GG-XD4PX 627-986-4143). The agency account stays read-only history; nothing new is built in it. Any change in ECOM LG FZE waits for David's per-action go (ads-account rule).
- 2026-10-14 — No competitor names on the calculator (David). One rival column, “a typical US 3PL” (survey averages); the named presets removed from the engine and every lane; the wizard's “who ships your orders today?” keeps generic answers (a US 3PL · direct from China · ourselves · nobody yet) as a qualifying question that does not change the maths. Spec v1.2.
- 2026-10-13 — The calculator's numbers confirmed by David (“those exact tiers are fine… even return handling is zero, and packaging costs are zero… free storage forever… freight on basic rates by cubic volume and weight… last mile DHL ecom saver, really cheap, no margin… the terms are fine”). Truth Gate p1–p5 + o4 move from PENDING to confirmed by David; Alan signs at launch. The fee stack on the page therefore reads $0 on returns, packaging, storage, receiving, kitting, transfers, minimums and account fees; postage is a pass-through at DHL eCommerce rates.
- 2026-10-13 — GoHighLevel read end to end (Private Integration token + location id from David). The headline: ~2 won deals a month from 50–180 new opportunities; 908 of 1,267 opportunities parked in No Response / Bad Fit / Future Opportunity; 7,702 of the 9,704 contacts are an imported outbound list; no UTM attribution exists on any contact, so the new site's forms must carry source, medium, campaign, content, gclid and fbclid as hidden fields into GoHighLevel from day one. Full read in §4 and
clients/lansilglobal/crm/2026-10-13/SUMMARY.md. - 2026-10-13 — The calculator brief (David, pasted note): a modular tripwire calculator on /pricing/ — a volume slider for pick & pack ($2.50 at 1,000 orders/mo sliding to ~$1.50), one flat fee per order, no storage fees, the rivals' hidden fees as $0 line items, email + phone + order volume as the gate, outputs the per-order / daily / monthly cost next to a rival so they book the call. Spec + competitor fee benchmark in
briefs/lansilglobal/site/calculator.md; the rate numbers are Truth Gate PENDING p1–p3 until Alan confirms. - 2026-10-13 — The blog stays where it is (David: "I think we just keep the /blog as it is, we dont want to mess with that, just have the blog link to there"). The 27 published WordPress posts are neither rebuilt nor migrated; the new site carries a Blog link (nav + footer) to
/blog, which Netlify proxies to WordPress (products/lansilglobal/netlify.toml,_redirectstwin regenerated) so every post keeps its URL and its search equity. Sitemap v2 nav updated (site/SITEMAP.md,sitemap.json). Cutover requirement (§9): the proxy target is the live domain today; at the DNS switch WordPress needs its own hostname at Hostinger (e.g. wp.lansilglobal.com) and must serve it without a canonical redirect, or/blogloops back into Netlify. - 2026-10-12 — Copy, one page at a time, in the Copy Studio (David: "the copy matters… I'd rather do one page at a time because I don't want it to get overwhelming"). Page 1 = the HOME page (LGS-01), written from GO angles A07 / A09 / A10 with the A01 / A04 / A15 facts (H1 = his best-rated hook, "You pay the factory after the product has already sold."); every line through Copy-Chief (
products/lansilglobal/COPYCHIEF-home-r1.md, 8/10 inside the rules: 6 rewrites, 1 kill, 4 launch checks); readability FK 5.09, 0 sentences over 24 words. Live on the staging site https://lansil-site.netlify.app/ (noindex; Netlifylansil-site, deploy only viaproducts/lansilglobal/deploy.sh) and in the Copy Studio for row-by-row verdicts. Out by rule until Alan confirms: the four offer lines (o1–o4), any guarantee, a minimum-volume answer. Launch checks the page depends on: one product in both lanes at once; /pricing/ with the rate card + calculator live; the appointment setter + instant text live. - 2026-10-09 — Kickoff call with Dwight Hijastro (transcript:
clients/lansilglobal/meetings/2026-10-09-kickoff-dwight.md). Aligned: (1) paid ads stay paused for October (Lansil's call, Q4 slowdown); Lansil's own lane right now is industry events (Alan at an event in China the week of Oct 13 — competitors invest in B2B events). (2) The new site is built from the keyword + competitor research as a funnel (sitemap → one page per keyword group → design + brand treatment → ads), with lead magnets (the landed-cost / lead calculator) and customer video testimonials; hosted on Netlify, DNS forwarded from Hostinger when live, WordPress untouched until then. (3) Post-lead flow: instant call button, immediate text, an appointment setter / SDR hired to call every lead at once (David takes the first few calls to learn the flow); Dwight keeps the CRM orchestration in GoHighLevel. Two phases: pre-lead (site, angles) and post-lead (nurture). (4) The China design team's proposed Figma redesign was reviewed on the call: copy far too formal, the form poorly designed — not used; its assets may be reused. (5) David opens a launch center the whole team can read: the client-facing Lansil launch board (clients/lansilglobal/launch-board/, Netlifylansil-launch-board), built on the MGI client-board pattern in Lansil's red + black with their logo (David 2026-10-09, design-framework pick: house pattern). - 2026-10-08 — New website = a brand-new build (David: "we are not using the existing website at all. This is a brand new website"). First step: the sitemap, built from the Keyword Studio verdicts (every BID / TEST group gets a page whose headline matches its searches) and the Angle Studio GO angles. Draft v2:
briefs/lansilglobal/site/SITEMAP.md— 22 pages (US fulfillment hub + Las Vegas / Pennsylvania / Shopify / pick & pack; China fulfillment + dropshipping + Amazon FBA prep; shipping from China + freight forwarder; pricing with the landed-cost calculator; financing; two compare pages; how it works; sourcing & QC with no ads; about / FAQ / get a quote / legal). Open: one dropshipping page or two; one compare page or one per rival brand. - 2026-10-06 — Angle Studio round 1 verdicts (David): GO LG-A07 Escaping the RBF trap · LG-A09 One pool of stock, every channel · LG-A10 Rates on the page. KILL LG-A06 The week de minimis died · LG-A08 The founder lost his first business to a factory · LG-A11 Best China-sourcing 3PL after de minimis (after a first GO) · LG-A12 Effortless is the word they don't believe — standing bans (CL-18). Hooks: 20 GOOD (A01 H1 H2 H3 H5 · A04 H4 · A05 H3 · A07 H1 H3 H4 · A09 H1 H2 H3 · A10 H1 H2 H3 · A11 H1 H2 · A13 H2 H3 H5), 1 BAD (A13-H4 "One carton, two lanes, one company."); A01-H4 noted "Doesnt make sense".
- 2026-10-06 — His notes, acted on: LG-A02 "De minimis is old news. Re-watch the transacripts and videos for these ads and make new angles based off of that." → competitor video sweep (66 videos downloaded from the Meta Ad Library, Whisper transcripts in
ads/research/lansil-3pl/scripts/) → new LG-A16 The more you sell, the less cash you have + LG-A17 Launch new products from Shenzhen, then ship the container; A02's hooks withdrawn, the card awaits his PARK / KILL. LG-A03 "This needs to be about Lead Times" → rewritten: Sell from Shenzhen while the container is at sea. LG-A04 (the usual chain: factory → QC company → China 3PL → freight forwarder → import agent → US 3PL) → rewritten: Six companies between your factory and your customer. LG-A13 "the spirit of this is correct… china to vegas to usa, with all one credit terms… generate more hooks and angles" → batch-2 hooks + new LG-A14 One set of terms from Shenzhen to Las Vegas + LG-A15 Shenzhen, Las Vegas and Pennsylvania under one partner. LG-A10 → the landed-cost calculator (Alibaba price −20% as COGS, internal freight / pick-pack fees), free storage, 10% below the US average, reference landing.shipvelocity.com/lp (savings calculator: orders, weight, SKU count → their rate vs yours) — the offer numbers sit on the Truth Gate as PENDING o1–o4 until Alan confirms. - 2026-10-06 — Hook taste, round 1: 20 of 21 judged GOOD — an operator saying the move plainly with a real noun (the factory, the invoice, the container, the rate card, FBA prep, Shenzhen, Nevada, sixty days); 'the question they already ask' 6/6, 'the split' 4/4, statics 17/17. The losers were riddles ("two lanes") and a question whose answer needs decoding.
- 2026-10-05 — Keyword Studio verdicts (David, all 43 groups; folded 2026-10-06): BID 16 (1,221 keywords · 190K searches/mo) · TEST 5 · SKIP 22 (981 keywords → negatives). BID: the big generic head terms US-3PL + US-FUL (704 kw, ~121K searches/mo, median CPC ~$11 → ≈$1,100 per 100 clicks), 'near me' (US-NEAR), pick & pack (US-OPS), Shopify (US-SHOPIFY), East/West coast (US-COAST), China core + cities (CN-CORE, CN-CITY), dropshipping (SRC-DROP 12.6K/mo, CN-DROP, CN-DROPCN), China → FBA (FBA-CN), China → US freight + forwarders (FR-CNUS, FR-AGENT), conquest ShipMonk (BR-SHIPMONK) + other 3PL brands (BR-OTHER: Deliverr, Easyship, ShipHero, Stord…). TEST: Nevada (note: "3PL WestcoasT Keywords"), Pennsylvania ("Eastcoast 3PL keywords"), pricing (US-PRICE), packaging (MFG-PACK), US freight (FR-US, note: "shipping agent"). SKIP: China sourcing agents (SRC-CN — "Sourcing is more initial business starting up. Too low of a volume."), China manufacturing, international, Amazon SFP/FBM, US prep, verticals, small business, DDP, other cities, platforms, learning searches, conquest ShipBob / Portless / NextSmartShip, own-name (one address search), both negative groups. Taste: volume first — 31 of 43 verdicts differ from the studio's read; he goes after the head terms ShipBob and ShipMonk own, dropshipping and rival-3PL names, and drops low-volume niches (sourcing = start-ups). Plan:
briefs/lansilglobal/keywords/PLAN.md· https://launch-center-hq.netlify.app/docs/lansilglobal/keywords/plan.html. Budget note: the BID set is ~190K searches/mo at $4–15 CPCs vs Lansil's ~$2.2K/mo today — the CAC ceiling (kickoff ask 1) sets how much of it to buy. - 2026-10-04 — Keyword Studio built (David: "shouldnt you build a keyword studio on the left workspace nav?" + "make sure to put the competitor keywords somewere in there too"): a Keyword Studio tab in every brand's Launch Center workspace nav (under Angle Studio). For Lansil it holds the competitor keywords in place (each rival's budget, paid keywords, what it keeps paying for, its full keyword table opening in the tab) and 43 keyword groups / 2,345 keywords (every keyword a rival keeps paying for, or Lansil already bids on, by service line × intent) for David's BID / TEST / SKIP verdict on the board
lansilglobal--keywords. Studio reads (ours, not verdicts): BID 7 — SRC-CN China sourcing agents, CN-CORE China fulfillment core terms, US-NV Las Vegas & Nevada, US-PA Pennsylvania, FBA-CN China → Amazon FBA, FR-CNUS China → US shipping, BR-OWN own name; SKIP the generic 3PL/fulfillment head terms (US-3PL 436 kw, US-FUL 268 kw — ShipBob + ShipMonk outspend Lansil ~40–100×). Findings: Lansil's ads show on 113 searches for named Chinese logistics companies and LinkedIn look-ups (NEG-NAMES) — broad match leaking; and it bids now and then on 241 of the 436 generic 3PL terms it can't win. Same day: the research classifier missed '3 pl', 'third party shipping' and China-shipping spellings — fixed (176 keywords out of off-target), keyword pages rebuilt. Verdicts: pending (David). - 2026-10-04 — Keyword research pages built (David: "build the per-competitor keyword pages on the hub"): one page per competitor + Lansil baseline + index at https://launch-center-hq.netlify.app/docs/lansilglobal/keywords/index.html, keywords sorted into Lansil's six service lines. All SpyFu data re-exported 2026-10-04 (SpyFu roll-up 2026-10-02; ShipMonk complete for the first time — 15,353 rows via two exports split by advertiser count 0–20 / 21+). "Core" = competitor's ad in 3+ of SpyFu's six latest SERP captures. Findings: ShipMonk 1,341 core (1,089 US fulfillment/3PL), ShipBob 579 (464 US), Portless 125 (58 China fulfillment, 46 freight), Bestfulfill 11 (8 sourcing), NextSmartShip 11, Ecomflow 2, EcommOps 0, DayOne not on Google. 153 keywords are core for 2+ competitors (154 after the same-day classifier fix); Lansil is absent from 103 — 100 of them generic US 3PL head terms held by ShipBob + ShipMonk (fulfillment 31.2K/mo, 3pl 21.1K, fulfillment center 6K). In the China lines Lansil is already in the fight: it shows up on 44 of Portless's 125 core keywords. Lansil's own Google Ads (SpyFu est.): 826 keywords but only 12 core (7 China fulfillment, 4 US incl. 'fulfillment center in nevada/pennsylvania', 1 FBA prep) — 766 seen only once or twice, i.e. spread thin; est. budget $2,218/mo for ~330 clicks (Sep 12 estimate: $6,614 for ~327 clicks) — confirm against the real account at kickoff.
- 2026-09-16 — ANGLE STUDIO round 1 boarded (
briefs/lansilglobal/angles/): 13 angles inseeds.md(LG-A01 you pay the factory after the product has already sold · A02 one customs entry, not ten thousand · A03 find the defect in Shenzhen · A04 no hand-offs · A05 the 3PL that's a software company · A06 the week de minimis died · A07 escaping the RBF trap · A08 the founder lost his first business to a factory · A09 one pool of stock · A10 rates on the page · A11 the ranking · A12 effortless is the word they don't believe · A13 2–4 days from NV/PA), 65 hooks, every seed on the gate's TRUE ids (20 TRUE / 13 FALSE parsed; x1–x13 untouched). Evidence: the competitor-intel program's Meta + LinkedIn ads converted into studio manifests (tools/angle-studio/ingest_competitor_program.py→ads/research/lansil-3pl/, 591 creatives) + 83 operator quotes from the Deep Research 08 step. 18 Trustpilot reviews of Lansil itself came back in the pull (praise for account managers Ivey and Chloe; complaints about production restarts, packages on the wrong truck, unanswered emails, one about Alan's communication) — kept inangles/_client-mentions-private.json(outside voc/, so the distil never reads it), never on a board or in copy; a reputation read for David before kickoff. Boardlansilglobal--angles; asklg-angle-board-r1. CL-20 in force. - 2026-09-10 — Client onboarded. Alan's founding notes (verbatim in
clients/lansilglobal/brief/raw-brief-2026-09-10.md): new website (features, rates), referral program, email marketing, Google Ads with a landing page per service, Instagram video/image editor, warehouse shoots in NV/PA (we plan + manage; Chinese team shoots Shenzhen), LinkedIn video/image/content writer, "more staff footage — real content", "who do we hire", competitor analysis via the Gemini deep-research stack. Named competitors: Bestfulfil, EcommOps, DayOne, Ecomflow, Portless (China) · ShipMonk, ShipBob (USA). Alan's read: "their websites are much better than mine, more hip, younger, direct — which is directly my market"; "we should be bigger than all of them". - 2026-09-10 — David: run the ENTIRE 7-step deep-research pipeline, not just competitors (B2B-adapted templates; first non-DTC run).
- 2026-09-11 — Alan's positioning thesis ("China players have no US warehouse; US players have no China sourcing") holds for all 7 named competitors but NextSmartShip already offers both → "the only one with both" is RETIRED. Differentiator set = sourcing + QC + freight + NV/PA + the credit line nobody else advertises + private/human.
- 2026-09-11 — Research complete 7/7. Mechanism ranking (06): Working-Capital Inversion 9.5 · Omni-Shore Tariff Dilution 9.0 · Point-of-Origin QC 8.5 · Omnichannel pooling 8.0 · Single-Custody 7.5.
- Standing rule (David): before drafting ANY prelander/LP, run the 4-framework selection (ask
lansil-prelander-frameworkon the Launch Center).
7 · Asset map
- Repo working files:
clients/lansilglobal/(README,brief/raw-brief-2026-09-10.md,meetings/for the kickoff recording + transcript) - Deep research (7 steps + MANIFEST + Truth Gate):
tools/deep-researcher/runs/lansilglobal/fulfillment-3pl/— rendered in the hub under Docs → Research - Distilled notes (Obsidian
Clients/Lansil Global/, mirrored in Docs): 00 Client Hub · 03 Buyer Avatar & Pain Points · 04 Campaign Angles & LP Ideas · 05 Brand Story & Messaging · 07 Raw Brief · 09 Competitor Analysis - Competitor intel program (2026-09-12):
clients/lansilglobal/competitors/— registry, raw ad-library/SpyFu captures, oneprofile.mdper competitor,README.mdsynthesis; rendered at https://launch-center-hq.netlify.app/docs/lansilglobal/competitors/index.html; ad-swipe boards per competitor in the hub - Google Ads keyword research (2026-10-04): pages at https://launch-center-hq.netlify.app/docs/lansilglobal/keywords/index.html (+ one per competitor, CSV download on each); builder
tools/competitor-keywords-build.py --client lansilglobal; service-line taxonomyclients/lansilglobal/competitors/keyword-taxonomy.json; raw SpyFu exportsclients/lansilglobal/competitors/<slug>/google-ads/spyfu/2026-10-04/(+overview.json). Hub docs markdown now staged inclients/lansilglobal/hub-md/(moved out of the deployed docs folder). - Keyword Studio (2026-10-04):
briefs/lansilglobal/keywords/—themes.json(the group rules + the studio's read per group; group IDs are permanent),bank.json(built),PLAN.md(written when verdicts are folded); boardtools/brand-hub/boards/lansilglobal--keywords.json; tooltools/keyword-studio/keywords.py build|verdicts|status --brand lansilglobal; skillskills/keyword-studio/SKILL.md. - Website benchmark board: homepage screenshots of Lansil + 8 competitors (2026-09-11) → hub Boards → "Website benchmark"
- Brand assets from Lansil (logo files, fonts, photography, warehouse footage): TO FILL — request at kickoff. The live site already uses real staff photography (orange vests / black tees) — no stock forklifts.
- Press: Pulse 2.0 interview (verified) · ecommercetimes "This Digital Nomad Built a Global Supply Chain Business" · thinkbusiness.ie "Alan Coughlan: The Irishman transforming global logistics" · The Echo (Cork) "Coronavirus has cost Cork man's business millions" (2020) · GlobeNewswire "Lansil Global Expands China and U.S. Fulfillment Network" — last four not yet re-verified (fetch blocked)
- Access ingest (2026-10-13):
clients/lansilglobal/ingest/README.md— one read-only tool per source (drive_bank.py,ga4_pull.py,gtm_pull.py,meta_pull.py,ghl_pull.py;_google.pyshares the house Google login); secrets in~/.lansil/*.env(600, never the repo); outputsclients/lansilglobal/assets/bank/(catalogue tracked, bytes ignored),tracking/ga4/,tracking/gtm/,ads/meta/,crm/.
8 · Live surfaces
- Supplier price check, LG-T09 way A — LIVE since 2026-11-09: https://rates.lansilglobal.com/price-check/ (also /price-check without the slash). Add
?selftest=1to keep a test out of the CRM. - It was built 2026-11-08 night. It has searched by name since 2026-11-09 morning, run the progress bar since late morning, said "Pay 21% less than your supplier." from the afternoon, and "Pay up to 20% less than your supplier." since the evening, with the tool back at 20%. - Production deploy6af222252e2bda4aaa0f99e0;LANSIL_PRICE_CHECK_LIVE=1is set (production). - Checked on the live domain: - The page answers 200 with the 21% headline. - The ads' landing page/lanes/bill/answers 200. - The lead function answers a test POST. - A live lookup (Sculpting Serum) finished in 11.5 s with four matches. - The full flow (sign-up → bar 0 → 100% → result) ran on desktop and phone with no errors and no "Alibaba" on the page. - Progress bar tested 2026-11-09 with Playwright: - Desktop, signing up straight away: 0% → 100% through all five steps, open at 12 s. - A returning visitor: straight to the bar, 11 s. - Phone, signing up after the result was in: 0% → 100% in 2.7 s. - Throughout: the percentage never went backwards, and the form stayed hidden during the bar. - Lead calls: the lead (pending) with the job id, then the "result" update with what was shown. - "Alibaba" appears once on the result page (the input note); no console errors. - Apify plan limits that shape it: FREE allows 5 jobs at once (a lookup uses 2), so a refused start is retried on each poll for 20 s. About $0.04 a lookup, so about 125 lookups on the $5 month. - Tested 2026-11-09 with Playwright on desktop and phone: four matches shown; a tap on the third re-prices and keeps focus; the lead POST carries the pick and a later tap sends theupdate: "pick"note; no sideways scroll; options at least 74 px tall; no console errors. - Source:products/lansilglobal/pricing/price-check/(index.html+check.js),functions/price-check.mjs(/api/price-check),functions/lead.mjs(kind: "tripwire"). - Netlify env:LANSIL_APIFY_TOKEN(secret; production, deploy-preview, branch-deploy). Since 2026-11-11 it holds the key of Lansil's own Apify accountlansil_global(§4), not the house account's. - Go-live switch:LANSIL_PRICE_CHECK_LIVE=1is SET for production (2026-11-09). Without it, the production addresses (rates.lansilglobal.com, lansilglobal.com, lansil-site.netlify.app) look nothing up and answer "We'll price this one by hand." To switch the lookups off again, delete the variable and redeploy. - Blobs storeprice-check: jobs, a 7-day cache by link, daily caps (6 a visitor, 40 in all). - Tested 2026-11-08 on the draft: a Shopify store (15–24 s), Amazon (15 s), an Alibaba link (3–6 s), an unreadable link (falls back to the team in 0.2 s), a returning visitor, desktop and phone. Every sign-up was a?selftest=1run, so nothing reached the CRM. - Home page design lanes, round two (2026-11-04, waiting on picks): the board is at https://launch-center-hq.netlify.app/docs/lansilglobal/home-lanes/r2/. Each lane has its own address under
/docs/lansilglobal/home-lanes/:k-route-people/,l-depth-line/,m-campaign/,n-journey/. - Source: the same folder as round one, with the board inr2/, the cut-outs inassets/cut/and three team photos added toassets/photos/. - Brief and cards:tools/design-lanes/R2-BRIEF.md,tools/design-lanes/lansil-home-lanes-r2.json. - Picks go to the verdict log on boardlansilglobal--home-lanes-r2, same candidates as round one. Ask:lg-home-lanes-r2. - Home page design lanes, round one (2026-11-03, judged in chat, see §6): the board is at https://launch-center-hq.netlify.app/docs/lansilglobal/home-lanes/. Each lane has its own address:
/docs/lansilglobal/home-lanes/<lane>/, froma-ledgertoj-comparable, plustoday/. - Source:tools/brand-hub/docs/lansilglobal/home-lanes/. -assets/holds the bank photos, the AI shots, the motion libraries and the posters for the six loops. - The loops themselves are hosted on the UGC Studio. Their source files arecreative/lansilglobal/videos/lansilglobal-home-loop-*, git-ignored in the main checkout. - The board is rebuilt withtools/design-lanes/(see its README). - Ask:lg-home-lanes. - Careers: the Community Growth Lead job post (LIVE 2026-11-02): https://rates.lansilglobal.com/careers/community-growth-lead/ (the same page is at lansil-site.netlify.app/careers/community-growth-lead/). - Noindex, like the rest of staging. LGS-23 in the site map, no ads. - Source:
products/lansilglobal/careers/community-growth-lead/index.html. It uses the site's look; the logo links to lansilglobal.com. - Applications go by email to david@lansilglobal.com. The subject and the three asks are filled in. - No tracking on the page, so candidates never enter the lead pipeline. - Link preview card:assets/careers-community-growth-lead.png(1200×630). - The deploy check refuses "Sanghera" and "David," on a shipped page, so the page says "Reports to Lansil's growth lead, David". - Tripwire mock-ups (2026-10-26, to pick): board https://launch-center-hq.netlify.app/lansilglobal/boards/lansilglobal--tripwire-mocks/ — four tools, three clickable mock-ups each (https://launch-center-hq.netlify.app/docs/lansilglobal/tripwire-mocks/t09/a/ … /t05/c/), PICK one per tool + feedback.
- Tripwires board (2026-10-25, to judge): https://launch-center-hq.netlify.app/lansilglobal/boards/lansilglobal--tripwires/ — one readable card per tripwire in the plan's build order (the line the visitor reads, what they get, how it gets the phone / orders / website, the searches it meets with real examples, build time, who runs it, what Lansil gives), BUILD / PARK / note on the verdict log; at the bottom, group by group, which searches meet which tripwire. Linked from Site Studio ("Judge the tripwires"), Needs You, Quick links and Judgement Boards. Plan https://launch-center-hq.netlify.app/docs/lansilglobal/site/tripwires.html · every keyword https://launch-center-hq.netlify.app/docs/lansilglobal/site/tripwires-by-keyword.html. Fold the marks with
python3 tools/site-studio/tripwires.py verdicts --brand lansilglobal. - The diagram for Alan, LIVE on the client board since 2026-10-25: https://lansil-launch-board.netlify.app/amplify/ (lane A, "How one tool gets amplified"; source
clients/lansilglobal/launch-board/amplify/index.html, a copy oftools/brand-hub/docs/lansilglobal/amplify/a/with the back link../), linked from the board's "What happens next" as "How the marketing works, in one picture". Lanes B and C stay on the hub boardlansilglobal--amplify-lanesas history. - Google Ads round one LIVE since 2026-11-09 19:36 PT (the go 2026-11-08, §6) in ECOM LG FZE 218-766-7313 (account time Asia/Dubai). Both campaigns were built and switched on by the runner (
build.logendsEXIT 0); they serve once Google approves the keywords (all 2,360 were under review at 19:36). The record isclients/lansilglobal/ads/google/round1-created.json, copied back from the frozen copy and committed: - A24445030068"LG · Search · US · 3PL & fulfillment · calculator": $50/day, 7 ad groups A1–A7. - B24450332765"LG · Search · US · China lane · calculator": $50/day, 8 ad groups B1–B8. - Both start on creation (07:35 Dubai Nov 10 = 19:35 PT Nov 9), status ENABLED, primary status LEARNING. - 2,360 keywords: 1,180 phrase plus 1,180 exact twins (A 955, B 225). 15 ads, 12 assets. - The goal "Calculator Lead only" iscustomConversionGoals/6459856393, on both campaigns. The shared negative list issharedSets/12307777267(932 exact). - Columns (2026-11-12, David: "setup my google ads columns in campaigns to be conversion oriented liek you did for MGI Clinic"): the Campaigns view of ECOM LG FZE uses the saved column set "Lansil Leads", built in the Google Ads web interface (custom columns have no API): - Order: Cost → Calculator Leads → Cost / Lead → Lead Rate → WhatsApp Clicks → Clicks → Avg. CPC → CTR → Impr. → Search impr. share → Search lost IS (budget) → Conversions → Cost / conv. → Bid strategy type → Campaign type. - Custom columns (account-level, so every view can add them): - Calculator Leads =All_conversions.conversion_action("Calculator Lead"). This counts the calculator's and the price check's opt-ins for every campaign, whatever goal it bids on. - Cost / Lead = Cost ÷ that, in dollars. - Lead Rate = that ÷ Clicks, as a percentage. - WhatsApp Clicks =All_conversions.conversion_action("Calculator WhatsApp click"). - Removed from the default view: Campaign subtype, Conv. value, TrueView views, TrueView avg. CPV, Conv. rate and Optimization score. - The four "Offline -" actions (Interest, Opportunity, Quote Negotiation, Signed Up; CRM stage imports from the old setup) get no column until they receive data. - First read, Nov 9–11 (Dubai time): - A: $25.54, 1 lead from 4 clicks, 1 WhatsApp click. - B: $65.96, 2 leads from 16 clicks. - Total: $91.50, 3 leads, $30.50 a lead, 15% of clicks. The CRM agrees (3 leads in those days). - Search impression share is under 10% with 0% lost to budget, so rank, not budget, limits how often the ads show. - Check the next time a filtered column is built: the editor saved 3 of the 4 filters as plain "All conversions". The Calculator Leads column must equal Conversions while the campaigns bid on Calculator Lead only. - The named set did not save the first time (David 2026-11-12: "you didnt save the column name, so it resset"). The first "Save & apply" applied the columns without saving the set, and the view fell back to Google's defaults. It was rebuilt and saved again the same morning, then verified: Columns → Your column sets lists "Lansil Leads", and a full reload keeps the lead columns. The save is Google'sCustomerUserAppDataService.Mutatecall. - Gone again by 09:15 PT the same day (David: "Did you fix the campaign columns? i dont see it, try again"): - At 01:39 PT the Columns menu listed only "Lansil Leads". By 09:15 PT that set was gone, and a set named "1.0", holding Google's default columns, had taken its place under the same login (david@davidsanghera.com). The view showed the defaults. - No Claude session made "1.0": neither the session transcript nor claude-mem shows it. The likely cause is a Google Ads tab opened before the save, which can write its own column list back when touched. This is not confirmed. - Rebuilt the same 15 columns in the same order at 09:20 PT and saved them as "Lansil Leads" (threeMutatecalls, all 200). Calculator Leads matched Conversions (3 = 3), so the filters held. - Checked after a fresh open of the account through the account picker: the lead columns stayed, and the menu lists "1.0" and "Lansil Leads". - "1.0" was left alone; it is David's to keep or delete. Saved sets appear to belong to one account and one login (MGI's set does not show here), so another Google login, or the phone app, shows its own columns. - Reads: Thu Nov 12 (numbers only) and Mon Nov 16 (first keep or pause), on the rules inmedia-plan-google-2026-11.md. Account changes only on David's word. - How it was built: a frozen copy outside the repo,~/.lansil/round1-build/(runner.py, logbuild.log), so no worktree cleanup or pull into /dev could change it. It waited out one lapsed Google sign-in (Nov 9, 11:19 to 12:38 PT). Its record was copied back over the repo'sround1-created.jsonthe same evening. - Client board row 4d (live since 2026-11-08) says: "Round one goes live on Monday Nov 9 in the evening, Pacific time, and runs two weeks, to Nov 23" (chip "Starts Nov 9"). - Media Plan HQ still shows the plan as waiting. Its next deploy also re-labels MGI's and NailScience's plans as past their dates, so it waits for David. - Google Ads media plan (2026-10-21): https://mediaplan-hq.netlify.app/lansilglobal/ on Media Plan HQ since 2026-11-03 — the Funnel map: each round as the path (keyword groups → $50 campaigns → the text ads → the bill screen → opt-ins), the round's story above, Every round below; the old Launch Center address forwards there. The chief's review: https://launch-center-hq.netlify.app/docs/lansilglobal/copychief/copychief-google-ads-r1.html. Built by
tools/mediaplan-hq/build.pyfrom the twin, deployed bytools/deploy-mediaplan-hq.py. - The calculator's CRM lead (updated 2026-11-11 night, production
6af57157bdffffad63a01df2):functions/lead.mjs+pricing/calc/ui.js. - Phones: saved with the country code. A number typed with + or 00 is kept; a 10-digit US or Canadian number gets +1 (round one is US-only); anything else gets the code of the visitor's country from Netlify's geo. If the CRM refuses the corrected number, the lead is saved with the number as typed, so it is never lost. The price check gets the same fix. - Store website: a fourth box on the lock, REQUIRED since 2026-11-12 (David: "all of these are required"), with the price check's label and placeholder. The page accepts any address with a dot in its host (mystore.com,https://www.mystore.com/x,brand.myshopify.com); the function does not refuse a lead without one, so an older cached page never loses a lead. It goes to the profile's Website and to Lansil's "Website or Store URL" field. When the box is empty, a work email's domain fills it; a free mailbox (gmail, yahoo, icloud and so on) fills nothing. - Where they are: city, state, country and time zone from the visitor's connection (Netlifycontext.geo). These fill empty fields only; the country also replaces the sub-account's default Hong Kong. - Lansil's own fields, filled only where empty (a returning contact keeps what it told them): - Where did you hear about us = Ads (a click id or paid medium) or Online Search (a search engine referrer). - What service = USA Order Fulfillment; the price check sets China Product Sourcing / Manufacturing. - utm_source, utm_medium, utm_campaign, utm_term and utm_content, plus Google Ad Click ID, gbraid and wbraid. - Monthly Order and Monthly Order (Text), filled only from a number the visitor set themselves, never from the starting 1,000. Both are rewritten as they play, as long as the text field is empty or still ends in "(pricing calculator)". - The note, written at the unlock and rewritten in place while they keep changing numbers: - The store, where they are and their local time, the search and campaign that brought them. - Their numbers, including the insert and 90-day toggles, and which of them they set themselves. If they changed nothing, it says so and tells the team to ask their real monthly orders. - What the page showed, how long they looked before signing up, and the device. - The page keeps the note's id with the contact id. - Tested: offline against a mocked CRM (14 phone formats, never overwriting a filled field, test mode stores nothing). The draft and the live site were checked with Playwright at 1440 and 390 px in?selftest=1: four fields at 17 px and 51 px high, the form inside the blurred block, no sideways scroll, and the payload carries the store, the inputs they changed, the toggles and the seconds. The first real lead after the deploy confirms the CRM side. - First real leads on the new code (2026-11-13 01:43 and 04:36 UTC = Thu 5:43 and 8:36 PM PT): both have the store website, the phone with its country code (+1, +65), Lansil's UTM fields and Google Ad Click ID, Where did you hear about us = Ads, the note, and an opportunity for Dwight. Both came in before the GA-form route went live (9:17 PM PT), so their source is "Landed cost calculator". - Business name (2026-11-13, production6af7be8cb35b02866dbaa55c, §6): on a create,fillEmptysets Business name (companyName) to the contact's website as stored, only while it is empty. This is the same value Lansil's 1st Contact Form Opt-ins workflow writes, so it no longer depends on which workflow runs. Offline test: 13 of 13 pass. - Route by click (2026-11-12, production6af69d56ae8eb4859f1c2b08, §6): a create whose first-touch attribution carries a Google click gets source "GA Form Ads FY26" and the workflow GA Ads Opt-ins. Every other create gets "Landed cost calculator" and 1st Contact Form Opt-ins. An update never rewrites the source.LANSIL_GHL_GA_WORKFLOWon the site overrides the id; an empty value sends Google-ad leads the old way. The function answersroute: "ga-form" | "calculator"on a create. Offline test: 12 of 12 pass, including 5 routes and the empty-value switch. - Google Ads vs the CRM, reconciled 2026-11-12 (David: "I noticed today there are 3 conversions but zero have been posted to mainstreamcrm?"): - Every conversion is in the CRM. Google Ads runs on Dubai time, so its "today" (Thu Nov 12) began at noon Pacific on Wednesday. The 3 conversions are the leads of Wed 3:55 PM, 7:54 PM and 10:48 PM PT (China-lane campaign, Google click id on each). - All 6 calculator leads since launch (Nov 11–12 UTC) are contacts with source "Landed cost calculator" and tag calculator-lead. Each has an open opportunity in Sales Pipeline / New Lead, assigned to Dwight. The page's own opt-in counter agrees: 4 on Nov 11 and 2 on Nov 12 (UTC days). - All 6 came in before the new lead function went live (07:57 UTC = Wed 11:57 PM PT): no note, no website, and 5 of 6 phones saved as +852. No calculator lead had come in since then (as of 09:30 PT Thursday). The live page passed a?selftest=1unlock at 1440 and 390 px, and the offline test oflead.mjspasses. - To compare the two, read Google Ads by Dubai day and the CRM by the time of each contact. The read-only check: GoHighLevel/contacts/searchsorted by dateAdded, then/opportunities/search?contact_id=for each lead. - The calculator — four lanes on staging + the judging board (2026-10-13): engine
products/lansilglobal/pricing/calc/(rates.json= David's tiers,fees.json= the fee lines,rivals.json= typical US 3PL / ShipMonk reported / ShipBob reported with sources,calc.js+ui.js); lanes https://lansil-site.netlify.app/pricing/lanes/ (slider · bill · wizard · dashboard, each with the gate as a prototype form that sends nothing yet); board https://launch-center-hq.netlify.app/lansilglobal/boards/lansilglobal--calculator-lanes/ (★ #1 = the screen we build, #2–4 = elements to borrow; desktop + phone shot per card, “Open the page ↗”). At 1,000 orders / 1.4 items / 10 pallets the engine reads Lansil $2,500 a month vs $4,007 at a typical US 3PL (38% less); 5,000 orders → $10,000 vs $19,027. - Asset Bank (LIVE 2026-10-13): tab https://launch-center-hq.netlify.app/lansilglobal/#assetbank · board https://launch-center-hq.netlify.app/lansilglobal/boards/lansilglobal--bank/ (1,439 photos + 552 clips, ★ HERO / KEEP / SKIP + notes, same verdict log) · thumbnails + the brand kit's logos and guidelines PDF on https://lansil-bank.netlify.app (Netlify
lansil-bank, 9edba97c-1d3e-4872-9852-3c9e80b368a2, noindex; deploy only viatools/deploy-lansil-bank.sh) · runbookskills/asset-bank/SKILL.md - Client launch board (LIVE, updated 2026-10-13): https://lansil-launch-board.netlify.app (noindex;
clients/lansilglobal/launch-board/, deploy only viaclients/lansilglobal/deploy.sh) · the new site, staging (home page LGS-01): https://lansil-site.netlify.app (noindex; deploy only viaproducts/lansilglobal/deploy.sh) · Copy Studio https://launch-center-hq.netlify.app/lansilglobal/copy/ · Site Studio https://launch-center-hq.netlify.app/lansilglobal/#sitestudio - Keyword plan (from David's verdicts, 2026-10-06): https://launch-center-hq.netlify.app/docs/lansilglobal/keywords/plan.html
- Keyword Studio (verdicts in — BID 16 · TEST 5 · SKIP 22): tab https://launch-center-hq.netlify.app/lansilglobal/#keywordstudio · board https://launch-center-hq.netlify.app/lansilglobal/boards/lansilglobal--keywords/ · 43 groups · BID / TEST / SKIP
- Hook Studio (hooks judged on their own screen — 👍 / 👎, keys G · B · N): https://copy-sheet-hq.netlify.app/lansilglobal/?tab=hooks · 65 hooks · same verdict log as the angle board · taste digest
briefs/lansilglobal/angles/TASTE.mdafter the first pass - Angle Studio board (round 2 boarded 2026-10-06 — rewrites, 4 new angles, hooks batch 2): https://copy-sheet-hq.netlify.app/lansilglobal/?tab=angles · bank
briefs/lansilglobal/angles/ - lansilglobal.com — home, 6 service pages, About, How It Works, Blog, FAQs (404), Contact
- Launch Center workspace: https://launch-center-hq.netlify.app/lansilglobal/ — Needs You (kickoff asks + Truth Gate + framework pick) · Brand DB (this file) · Copy & Claims (
truth-gate.md) · Boards (website benchmark) · Docs (research + notes) - Console lane:
tools/launch-console/launches/lansilglobal/ - (2026-09: no Lansil-owned Netlify surfaces yet — superseded by the three sites above)
- Copy Studio on the calculator (2026-10-20, David: “setup copystudio on the calculators”): Launch Center → Lansil → Copy Studio now lists the four screens next to the home page —
calc-bill(7 sections) ·calc-wizard(8) ·calc-slider(7) ·calc-dashboard(6). They extract in the Studio's newrenderedmode (tools/copy-studio/extract.py): headless Chromium opens the live screen but answers the document request with the repo file, so the<!-- [NN] -->markers survive while the page's own JS writes the bills, the ledger, the lock form and the sources line — every word the visitor sees is on the board, including what lives incalc/ui.jsandcalc/fees.json. The View-screen links carry?selftest=1(a review visit counts nowhere). Edits land in the source that owns the line (lane HTML,ui.jslock options,fees.jsonline labels), thenextract.py lansilglobal→deploy-launch-center.sh --build; verdicts pull withtools/copy-studio/feedback.py pull lansilglobal. - Copy sheet (2026-10-28, David: every headline and primary-text idea judged on one spreadsheet-style grid, on every brand — copy law CL-23): https://copy-sheet-hq.netlify.app/lansilglobal/ — 109 rows (109 hook), nothing judged yet;
python3 tools/angle-studio/copysheet.py build --brand lansilglobal --auto·verdictsfolds the 👍 / 👎 intobriefs/lansilglobal/copy/. · since 2026-10-30 the brand's page on Copy Sheet HQ carries every tab: angles · hooks · hook buckets · ad headlines · ad primary text · email subject lines · the truth gate · every verdict list (the Launch Center's grids and card views retired behind redirects)
9 · Open questions / asks (mirrored as Needs You cards)
- WhatsApp follow-ups (2026-11-15, §6), for Dwight: (1) send +1 leads their follow-ups by email and SMS, not WhatsApp templates; (2) make "Hope you've been well…" an approved template or send it only inside the 24-hour window; (3) stop workflows resending a blocked template within 24 hours; (4) fix the template that went out with an empty name. Offered to write Dwight the change list.
- Calculator leads (2026-11-11 night, §6, §8): - GA form (2026-11-12): read the first Google Ads lead after 9:17 PM PT Nov 12: source "GA Form Ads FY26", enrolled in GA Ads Opt-ins, and the opportunity that workflow opens. - Consent line, David's call: add the GA form's consent sentence to the calculator's lock, either under the button or as a box. Today the calculator's leads get Lansil's email, SMS and WhatsApp follow-ups without having seen it. - David's yes: fix the 3 leads still saved with +852 in front of a US number (Nov 11 01:02, 09:54 and 16:18 UTC; Lansil's side fixed the other two by hand), and give the 6 earlier leads the note. These are writes to Lansil's records, and a workflow could fire on the change. Their tracking fields were filled on 2026-11-12 at David's ask (§6). - ~~Read the first real lead after the deploy~~ — done 2026-11-12 night: the two leads of Nov 13 01:43 and 04:36 UTC are complete (§8). - Optional: make the calculator's WhatsApp button carry a short reference (campaign or keyword) so a chat can be matched to its ad. Today a WhatsApp lead has no click data. - Optional: put a value on the New Lead card, such as the Lansil month × 12 when they set their own numbers.
- Supplier price check (LG-T09 way A), LIVE since 2026-11-09 (§8). Still open: - The three pricing edges in §6: 20% under the listing when they pay more, no price under 87.5% of the listing, and the team takes unreadable links. - The sourcing team's two-day text promise, still a default on the inputs card (
lg-tripwires-inputs). - Who gets notified for the tagprice-check(a GoHighLevel workflow on Lansil's side). - Apify: stay on the FREE $5 a month or move to a paid plan. Lookups have run on Lansil's own account since 2026-11-11 (§4). Since 2026-11-09 a lookup is about $0.04 (name and photo), so the free month covers about 125 lookups, then the page falls back to the team. FREE also runs only 5 jobs at once: two visitors at the same moment are fine, and a third waits up to 20 s. - The headline and sub: DECIDED in chat on 2026-11-09. The headline is "Pay up to 20% less than your supplier." (the evening's ruling, replacing the afternoon's 21% line) and the sub is B1's. The tool prices up to 20% under. Asklg-copychief-price-check-hero-r1is answered. - Copy chief round 1: the review is written and posted as 13 rewrites, waiting on David's APPROVE / DENY / FEEDBACK (asklg-copychief-price-check-r1on Copy Sheet HQ). Approved rows get applied in one draft deploy. The top three: a reason why plus the guarantees at the number (S1, S3), the year figure (S2), and an after-block that names the landed cost with a WhatsApp button that names a specialist (S6, S7). The input note's "Alibaba / AliExpress" is answered: removed 2026-11-09. - Known limits: branded devices (an Apple AirTag) match accessories, and bundles and kits ($99.99 "Prime Face System") match nothing plausible and go to the team. A name that hides what the product is gets the store's own category and description words. When those are wrong too, the visitor can tap another match. - Home page design lanes: round one ANSWERED in chat 2026-11-03 (Route and Depth, see §6; ask
lg-home-lanes). Round two's four lanes are on their board at /docs/lansilglobal/home-lanes/r2/, waiting on David's picks (asklg-home-lanes-r2). For each section, either the best version or a whole lane. - What happens next: the picks are rebuilt in one look on staging (lansil-site.netlify.app) throughproducts/lansilglobal/deploy.sh. - Type: the site's typeface must be settled before that ships: Kanit, a licensed PP Neue Montreal, or the winning lane's Google font. - Tripwire mock-ups (2026-10-26): pick one way per tool and leave feedback on the mock-up board (pointer
lg-tripwire-mocks); then the picks get built in the plan's order, the supplier price check first. Needed for the build: a server-side Alibaba product lookup (David named the Alibaba MCP; none is connected here), and the answers onlg-tripwires-inputs(including whether the Google ad's 30% moves to 20%). Added the same day with David's two notes:T07-postage(Lansil's DHL eCommerce rates by zone and weight; until they come the live map shows zones only) andT09-store(the store-link way reads the store's product list on the server; NO = the visitor lists products by hand). - Tripwires (2026-10-25): PICKED — BUILD on T02 invoice teardown, T09 supplier price check, T07 delivery-days map, T03 credit-terms check, T05 freight estimate (the other fifteen unmarked = not this round). What the five need:
lg-tripwires-inputson Needs You, a default per item — T02 answered within one business day; T09 a factory price and the landed cost within two business days; T07 distance zones from NV and PA (public USPS zone data) beside "2–4 days anywhere in the US", or Lansil's own days by zone; T03 "up to $1M, sized to your order volume on a call" until Alan gives the rule; T05 a public market range by sea and air with Lansil's exact quote by text; every request taggedtripwire-lead+ its tool with Dwight's instant text. Still open onlg-tripwires-decisions: D1 the website after the unlock, D2 the inventory-loan group (T03 needs it), D3 drop the 34 no-fit round-one keywords, D4 the round-two page swap, D5 who runs the human tripwires. The inventory-loan volumes need the Google login renewed (invalid_raptagain 2026-10-25); seeds intripwires.json → new_groups. - The diagram for Alan: ANSWERED 2026-10-25 — A (in chat; recorded on the hub ask and starred on the board). Live at /amplify/ on the client board; the media-plan row 4d published the same day on David's word.
- A catchier name for the 60-day-terms method (David 2026-10-12 on "the Negative Cash Conversion Cycle method": "This is great, but need a better and catchier name") — wanted before the credit-terms check or the /financing/ page carries it.
- ANSWERED 2026-11-08 in chat (§6), both asks recorded answered:
lg-google-media-plan-goandlg-copychief-google-ads-r1. New, for David: the house Google Ads developer token is Explorer access (2,880 operations a day across every account it touches, reads included); applying for Basic access (15,000 a day) in the API Center of the manager account that owns the token stops a build like round one being split over two days. - Google Ads media plan — the go (2026-10-21, ask
lg-google-media-plan-go, five YES/NO/FEEDBACK rows): (1) the go + the start date (Nov 3 assumed; the campaigns go in PAUSED viabuild_round1.py --goand switch on at 00:00 PT); (2) $100/day as two $50 campaigns or another cap; (3) US only in round one, Australia waits; (4) demote the old page-view "Lead" conversion in ECOM LG FZE to secondary before day one (the §9 item from the 2026-10-14 read, still open); (5) the Mon/Thu reads with verdicts on the lane. Pluslg-copychief-google-ads-r1: APPROVE / DENY / FEEDBACK per ad group on the Copy-Chief tab (APPROVE = the lines go in as written; DENY = the group does not run). After the go: the client board row 4d deploys (budget + date in the client's words), the Google refresh token must be fresh on the build day (invalid_rapt~daily). - Kickoff next steps (2026-10-09): Dwight → Figma files (current site + the China team's proposal), Google Ads Standard access on both accounts (the 627-986-4143 invite is email-only), GTM + GA4 admin, Meta Business Manager admin or whatever access exists, GoHighLevel user, raw photos + video (Google Drive), the call recording. David → audit both Google Ads accounts + GoHighLevel once access lands; the launch board (done 2026-10-09). Login to the David@lansilglobal account failed — Dwight re-sent credentials to David's personal email. - Status 2026-10-13: photos + video → RECEIVED (the Drive bank; catalogued, mirroring). Meta → ad account + pixel + CAPI token received; the Page pending; the partner share to AMV LLC
113824746806442still to ask (the token cannot read). GA4 + GTM → granted to the Lansil login; david@davidsanghera.com still to be added. GoHighLevel → pending. Google Ads Standard access → pending. Figma files + the call recording → not yet. - DNS cutover checklist (David 2026-10-13: "keep that in mind once the Netlify site's done and we have to do the deploy"; mirrored in the header ofproducts/lansilglobal/deploy.sh, which also guards step 1): (1) the blog stays on WordPress — Dwight gives the install its own hostname at Hostinger (e.g. wp.lansilglobal.com), WordPress serves it without a canonical redirect, the/blogproxy target innetlify.tomlchanges to it, twins regenerated (the deploy script refuses a proxy that loops back into Netlify); (2) drop the stagingX-Robots-Tag: noindexheader fromnetlify.toml, regenerate, flip the script's noindex check to "must NOT ship"; (3) lansilglobal.com + www as the Netlify custom domain, HTTPS issued, then the DNS forwarded at Hostinger; (4) smoke after the switch: home, every sitemap page,/blog+ five post URLs, the quote form, Lansil Pixel 2.0 + GA4 + GTM firing on the new site, Search Console on the new site; (5) the old WordPress pages that are not on the new sitemap get 301s to their nearest new page (map them from the WordPress sitemap before the switch). - New asks from the ingest (2026-10-13): (1) DECIDED 2026-10-13 — the blog stays on WordPress as it is and the new site links to it (/blogproxied); what the decision now needs from Dwight before the DNS switch: a hostname for the WordPress install at Hostinger (e.g. wp.lansilglobal.com) and WordPress accepting it without a canonical redirect, then the proxy target inproducts/lansilglobal/netlify.tomlchanges to that host — add it to the launch checks; (2) PP Neue Montreal in the brand kit is the free-for-personal-use cut — buy the web licence or set the site typeface from Kanit / a licensed face before the design lanes; (3) the Meta Page share (still pending) and a CAPI token generated for Pixel 2.0 when the relay is wired; (4) the ad account's Asia/Manila timezone — leave it, report in Manila days. - From the CRM read (2026-10-13): (a) wire UTM + click-id capture into every GoHighLevel form on the new site (hidden fields: utm_source / medium / campaign / content / term, gclid, fbclid, landing page) — nothing is attributed today; (b) stage hygiene with Dwight: what “No Response” (377), “Bad Fit” (318) and “Future Opportunity” (213) mean, an SLA for each, and whether the two “Signed Up” stages (Sales 34 won, Close 22 won) count the same customers; (c) who Robert is (the closer's calendar and the “Close” pipeline) and whether the setter hire replaces or feeds him; (d) the 0 calendar bookings in 12 months against 32 open “Post-Discovery Call” cards — how calls are booked today. - For the calculator's lead (2026-10-14): Dwight addscontacts.writeandlocations/customFields.writeto the Private Integration (thenpython3 clients/lansilglobal/ingest/ghl_fields.py ensurecreates the tencalc_*fields, and the token + location go onto the Netlify site asLANSIL_GHL_TOKEN/LANSIL_GHL_LOCATION—netlify env:set, never the repo), and builds the workflow on thecalculator-leadtag: the instant text, the setter's call, the pipeline stage. Until then the lanes unlock without storing anything. - For the calculator subdomain (2026-10-18): Dwight adds at HostingerratesCNAMElansil-site.netlify.app(Netlify issues the certificate itself once it resolves) — thenrates.lansilglobal.comis the link for ads and the client board; and thecalculator-leadworkflow (still open) is what turns those opt-ins into calls. - From the Google Ads read (2026-10-14): (a) the primary “Lead” conversion is a page view — demote it and make the two form submits the only primaries before anything relaunches, then rebuild the conversion set on the new site (quote form, calculator unlock, call); (b) the country split: Australia converted 11 times on $1.4K while the US converted 4 times on $3.1K — ask Dwight/Alan whether AU leads were real customers (the AU campaign used US terms) before the relaunch decides geos; (c) the $2,400 on terms that never converted and the competitor-name clicks → the Keyword Studio's NEG-NAMES + NEG-OFF lists become the campaign-level negatives on day one; (d) the same keywords across six country campaigns and 1,991 ad-group negatives → the relaunch is the Keyword Studio plan (one campaign per service line × BID groups, exact + phrase, campaign-level negatives); (e) the ship.lansilglobal.com funnel pages and their doubled UTM templates → the new site's pages replace them and one clean tracking template carries campaign / ad group / keyword / gclid; (f) desktop-only delivery — confirm it was a device bid adjustment, not an audience accident; (g) Dwight's 616 edits in a paused month — what is he changing, and should it stop until the plan. - ANSWERED 2026-10-20 by David (see §6 and
truth-gate.mdf11–f23): the offer numbers, the operations facts, customers on camera (→ the casting program), the QC reject rate (not available). Still open from the lists below: the company numbers (Q18), the sales-process / economics / content-machine questions 24–34 — Q26 half-answered 2026-10-20 (the stage glossary,crm/STAGES.md; no SLA per stage, the two Signed Up stages unreconciled). - Angle Studio round-2 questions for Alan (2026-10-06): the offer numbers David wants — o1 sourced cost ~20% under the Alibaba price, o2 free storage, o3 fulfillment ~10% under the US average, o4 do the 60-day terms cover freight and fulfillment too; does Lansil handle US import clearance (A04's "import agent"); can one SKU run in both lanes at once, Shenzhen-direct and a US warehouse (A03, A17); stock transfers between Shenzhen, NV and PA (A15); the smallest first run for a new product (A17); real customers willing to go on camera (the rivals' longest-running videos are customer testimonials; x7 bans invented ones).
- Keyword Studio questions for Alan (2026-10-04): do NV/PA ship Seller Fulfilled Prime / FBM orders (US-AMAZON, FBA-SFP)? minimums for small dropship accounts (CN-DROPCN)? is custom packaging a service (MFG-PACK)? freight destinations beyond the US (FR-ABROAD)? the client industry mix (US-VERT)? the Google Ads account itself — match types and search-terms report (NEG-NAMES says broad match is leaking).
- Alan (kickoff): the real rate card (A10), the Shenzhen QC checklist / reject rate (A03), staff names + consent for the 'who checks your pallet' hooks (A12), financing qualification rules (A01/A07) — and the Trustpilot reviews (18) as a service-quality read.
1. Account economics — avg monthly billing per client, gross margin by service line, retention, quote → first-order time → CAC ceiling per signed account → ANSWERED 2026-10-12 (pulled 2026-10-25): "We need to think of this as Cost per Lead. Because these clients once they are onboarded have an insane LTV. So the idea is getting volume of leads to qualify and nurture and get on phone calls with so we can understnad if they high volume orders or low volume orders." — the metric is cost per lead; order volume is the qualifier; the margin and CAC questions stay open
2. Sales process — quote form → who follows up (Robert?) → response SLA → close rate → what kills deals
3. Capacity + priority service line — accounts/month that onboarding + NV/PA can absorb; fill US warehouses vs sourcing vs FBA prep → ANSWERED 2026-10-12 (pulled 2026-10-25): USA Order Fulfillment (fill NV/PA) is the 90-day priority; capacity stays open
4. Ideal client + floors — orders/month minimum, categories wanted/refused (DG, liquids), geographies
5. Rate card + guarantees — what goes on the site; which of "no hidden fees / no exit fees / no postage markups / free rework on QC failure" Alan will actually guarantee; credit qualification
6. Access — Google Ads (history?), Meta, LinkedIn, GA4/GTM, CRM, email platform, CMS/host, registrar, socials
7. Proof — usable client names/logos, testimonials, case studies with numbers, current facility sq ft, footage; reconcile 20M/15M orders, 1,492/2,000 factories, 74/100 staff
8. Alan on camera / LinkedIn voice? (founder-led content is the category's acquisition pattern) → ANSWERED 2026-10-12 (pulled 2026-10-25): yes — on-camera video and LinkedIn in his voice
9. Shoots — NV + PA access + dates + on-site leads; Shenzhen shoot owner; staff consent; subtitles
10. Hiring — budget, contractor vs payroll, ours vs theirs: short-form editor, LinkedIn ghostwriter, day-rate videographers → ANSWERED 2026-10-12 (pulled 2026-10-25): "We can be hiring all those people once we start getting leads in."
11. Referral program — client-refers-client credit vs partner/affiliate (Shopify agencies, sourcing consultants); existing partnerships
12. Email — list size + segments, platform, goals
13. Approval flow — who signs off on copy, claims, design → ANSWERED 2026-10-12 (pulled 2026-10-25): David decides, Alan reviews weekly
14. Consent — may David's own brands be used as customer-discovery / case-study material? → ANSWERED 2026-10-12 (pulled 2026-10-25): no — David's own brands are not used as discovery or case-study material
10 · Change log
- 2026-11-15 — Read why Lansil's WhatsApp messages fail in GHL: 34 of 74 outbound in three weeks. 13 hit Meta's marketing limit (11 to US numbers, where Meta delivers no marketing templates), 13 were free text outside the 24-hour window, 7 went to numbers not on WhatsApp, and 1 template was missing its name (§6, §9). Nothing changed.
- 2026-11-13 — Business name = the website on every calculator and price-check lead. The step was missing on the GA route because the "GA Ads Opt-ins" workflow does not copy it.
functions/lead.mjs; draft6af7be7c293925f0f3c4469b, production6af7be8cb35b02866dbaa55c. The one GA-route lead without it was filled and read back (§6, §8). - 2026-11-12 (night) — Tracking copied onto the 6 earlier ad leads in Lansil's CRM (utm_*, Google Ad Click ID, gbraid, heard = Ads; empty fields only, 6 of 6 read back) at David's ask. The first two real leads on the new lead function were checked and are complete. The two blank "google ads" contacts are WhatsApp chats from the calculator's button (§6, §8, §9).
- 2026-11-12 — Calculator leads from a Google ad come in as "GA Form Ads FY26" leads: the form's name as their source, and enrolled in its workflow GA Ads Opt-ins (David: "please use this form for leads").
functions/lead.mjs; draft6af69d3f1f696e5ab0b7622f, production6af69d56ae8eb4859f1c2b08; test-mode create answered on the live site (§6, §8, §9). - 2026-11-12 — Google Ads: the "Lansil Leads" column set rebuilt a second time after David saw Google's default view again. A set named "1.0" (Google's default columns) had replaced it under David's login; no session made it. Checked after a fresh open of the account; "1.0" left in place (§8).
- 2026-11-12 — Google Ads: the column set "Lansil Leads" saved for real after David saw it reset (the first save never took). Verified in the Columns menu and after a reload (§8).
- 2026-11-12 — Google Ads columns (§8): column set "Lansil Leads" on the Campaigns view of ECOM LG FZE. It has four custom columns (Calculator Leads, Cost / Lead, Lead Rate, WhatsApp Clicks) and drops the video and value columns. This is a view change, not a campaign change. Done at David's ask.
- 2026-11-12 — The calculator's Store website box is required (David: "all of these are required").
pricing/calc/ui.jschecks it and names it in the message. Draft6af5785dfd9c7c11931379c6, production6af578731a58ad9d2662e541, checked live at 1440 and 390 in?selftest=1. An empty box or "my store" stays locked and sends nothing; a real address unlocks (§6, §8). - 2026-11-11 (night) — Calculator leads (§6, §8, §9): -
functions/lead.mjs: phones saved with the country code; where they are, from Netlify's geo; Lansil's own fields filled where empty; one note per calculator lead, rewritten as they play; the price check gets the phone, location and fields too. -pricing/calc/ui.js: an optional Store website box on the lock; the payload carries the store, the toggles, which inputs they changed, the seconds before sign-up and the note id. - Draft6af5709d7af26a009a137908, production6af57157bdffffad63a01df2. - 2026-11-11 — The price check's lookups run on Lansil's own Apify account
lansil_global(support@lansilglobal.com, FREE).LANSIL_APIFY_TOKENwas swapped on lansil-site (same scopes and contexts, secret). Then draft6af4ac5b1ccbe6de83966685and production6af4acfaa22ec9bbe8179aa4were deployed. Checked with one real lookup on each: draft $0.58 → $0.47; live $1.00 → $0.80, keep $100 at 500 units. The house account logged no run after 17:56 UTC (§4, §6, §8, §9). - 2026-11-09 (evening) — Price check: the headline and
<title>say "Pay up to 20% less than your supplier." (David: "up to 20% off type language").check.jsprices at 0.80 again, rounded UP to the cent; the floor is unchanged. tripwires.json LG-T09 status and promise follow. Copy chief round 2 isproducts/lansilglobal/pricing/COPYCHIEF-price-check-hero-r2.md. Draft6af2a701344c1395e2cf560c, then production6af2a776143f72b6d21b52ba; checked live ($1.50 quote → $1.20, keep $150 at 500 units, no console errors) (§6, §8, §9). - 2026-11-09 (evening) — Google Ads round one is LIVE: the runner built both campaigns at 19:35 PT, switched them on and added the exact twins (
EXIT 0); A24445030068and B24450332765, 2,360 keywords in review. The recordround1-created.jsonwas copied back into the repo (§8). - 2026-11-09 (afternoon, last) — The calculator's test mode (
?selftest=1) now stores nothing (David: "yes make the calculator test mode save nothing").lead.mjsreturns {ok, mode:"test"} before any GoHighLevel call on the unlock and on every slider update, like the price check. Test unlocks fired no GA4, Meta or Google Ads events before this either (the tracking runtime already skips test mode). - 2026-11-09 (afternoon, last) — Correction: the 4
calculator-leadcontacts are David's and Dwight's tests, not unrouted prospects; the backfill question is dropped (§6, §9). - 2026-11-09 (afternoon, last) — Lead routing:
functions/lead.mjsenrols every new calculator and price-check opt-in in Lansil's "1st Contact Form Opt-ins" workflow, which handles New Lead, Dwight and follow-ups (§6, §9). Production deploy follows. - 2026-11-09 (afternoon, later still) — The supplier price check is LIVE at https://rates.lansilglobal.com/price-check/:
LANSIL_PRICE_CHECK_LIVE=1was set for production, then production deploy6af222252e2bda4aaa0f99e0. Checked on the live domain (§8). tripwires.json LG-T09 status is LIVE (§6, §8, §9). - 2026-11-09 (afternoon, later) — Price check at 21%: the headline and
<title>say "Pay 21% less than your supplier.";check.jsprices at 0.79, rounded down to the cent, with the 70%-of-quote floor kept; tripwires.json promise line updated. Draft6af21e8763e2420b32adee6fchecked end to end ($1.50 quote → $1.18, keep $160 at 500 units). Hero ask answered (§6, §9). - 2026-11-09 (afternoon) — Price check: the hero sub is now "Your supplier might not be the factory, and you pay their markup on every unit. Paste a product link and see the factory price in about a minute." (David approved B1's sub in chat). Changed
pricing/price-check/index.htmland the tripwires.json promise line; draft6af21e8763e2420b32adee6fdeployed (§6, §9). - 2026-11-09 (early afternoon) — Price check, the headline and sub's copy-chief round:
products/lansilglobal/pricing/COPYCHIEF-price-check-hero-r1.md,copychief-price-check-hero-r1-items.jsonand the hub doctools/brand-hub/docs/lansilglobal/copychief/copychief-price-check-hero-r1.html. Asklg-copychief-price-check-hero-r1is posted (28 rows), and Copy Sheet HQ is deployed (§6, §9). - 2026-11-09 (midday) — Price check: the input note no longer names Alibaba or AliExpress. Copy-chief round 1 added:
products/lansilglobal/pricing/COPYCHIEF-price-check-r1.md,copychief-price-check-r1-items.json, and the hub doctools/brand-hub/docs/lansilglobal/copychief/copychief-price-check-r1.html. Asklg-copychief-price-check-r1is posted to the main checkout's lane (§6, §9). - 2026-11-09 (late morning) — Price check: the progress bar after the sign-up; the lead is sent at sign-up with the result as a second note; no "Alibaba" on the result page (§6, §8, §9). Changed
pricing/price-check/{check.js,index.html}andfunctions/lead.mjs(update: "result", the "still matching" note). Draft redeployed, production untouched. - 2026-11-09 (morning) — The supplier price check searches Alibaba by name and photo and shows the matches (§6, §8, §9). Changed: -
functions/price-check.mjs: the search phrase, two searches, pooled ranking, retry on a refused start, the matches sent to the page, cache v5, and a fix so a listing without price tiers keeps its listed price instead of going to the team. -functions/lead.mjs: the CRM note names the search and the pick;update: "pick"adds a short note. -pricing/price-check/{index.html,check.js}: the matches list and the pick. - Draft redeployed, production untouched. - 2026-11-08 (night, later) — The supplier price check, way A, built and draft-deployed (§6, §8, §9). New:
products/lansilglobal/pricing/price-check/{index.html,check.js}andfunctions/price-check.mjs. Changed:lead.mjs(thetripwirekind),netlify.toml+ regenerated_redirects,briefs/lansilglobal/site/tripwires.json(LG-T09). Netlify envLANSIL_APIFY_TOKENwas set. Production is untouched. - 2026-11-08 (night) — Google Ads round one, the go (§6): first resources made in ECOM LG FZE (goal, 12 assets, the shared negative list); the two campaigns build Monday Nov 9 from 19:35 PT.
build_round1.pygained--preflight,--status,--enable,--add-exact,--phrase-first,--start now|YYYY-MM-DD, the custom goal, keyword cleaning, the shared negative list, all-or-nothing requests and resume; askslg-google-media-plan-go+lg-copychief-google-ads-r1answered. A validate-only pass (185 operations, nothing created) cleared the campaign settings, all 15 ads and the punctuated keywords, and caught one keyword Google refuses ("shenzhen chuanhai int''l forwarder", a doubled apostrophe: one now stays); Monday's runner is a frozen copy in~/.lansil/round1-build/. Client board row 4d deployed (clients/lansilglobal/deploy.sh). BRAND.md §6, §8, §9. - 2026-11-04 — The home page's round two: four hybrid lanes and their board, after David picked Route and Depth. - Lane pages:
tools/brand-hub/docs/lansilglobal/home-lanes/{k-route-people,l-depth-line,m-campaign,n-journey}/. Board:home-lanes/r2/(index.html+board/). Assets added:assets/cut/(40 people, 5 MB) andassets/photos/team-{anniversary,beach,reception}. -tools/design-lanes/gainedmake_cutouts.py+cutout.swift,density.py,copy_check.py,photo_contrast.py,frames.py,R2-BRIEF.md,lansil-home-lanes-r2.json;build_home_board.pylearned later rounds, andsection_shots.pylearned--film. - Each lane had an independent interface review and a fix pass: K 6 must-fix and 8 should-fix, L 11 and 8, M 6 and 15, N 8 and 12. - Ask posted:lg-home-lanes-r2. BRAND.md updated in §6, §8 and §9. - 2026-11-03 — The home page's ten design lanes and their judging board. - Lane pages:
tools/brand-hub/docs/lansilglobal/home-lanes/. The folder holdstoday/, ten lane folders,assets/, the board's per-section shots inboard/, andindex.html, the board itself. - Board builder:tools/design-lanes/. It holds the lane brief, the checker, the section shots, the board build and the review brief. - The six loops are on the UGC Studio. - Ask posted:lg-home-lanes. - BRAND.md updated in §6, §8 and §9. - 2026-11-03 — Media Plan HQ:
tools/mediaplan-hq/(model.py · template.html · build.py · netlify.toml + _headers/_redirects) +tools/deploy-mediaplan-hq.py(gated: freshness, hygiene --internal, no other brand on a plan page, smoke); https://mediaplan-hq.netlify.app/lansilglobal/media-plan-google-2026-11/ is this plan. Launch Center: /docs/lansilglobal-media-plan-google-2026-11.html → 301 to the HQ (build.py write_redirects); the Copy Sheet HQ Media plan tab and mediaplan.py removed. - 2026-11-02 (night) — Media plans as sheets:
tools/copy-sheet-hq/mediaplan.py(new),grid.py(the Media plan tab, the read-only mode, the ads card),build.py,deploy-copy-sheet-hq.py(nine data files),netlify.toml+_redirects;tools/brand-hub/media_plan_page.pyrewritten to write the read-only sheet; the hub's Copy Sheet links gain Media plan. §6, §8. - 2026-11-02 (night) — The bill screen's per-order figures moved to the top:
products/lansilglobal/pricing/lanes/bill/index.html(.uniton each bill,.unitson the phone receipt, the Per order rows removed from the footers). Copy Studio re-extracted. §6. - 2026-11-02 (night) — The job post is live on the new site: -
products/lansilglobal/careers/community-growth-lead/index.htmlandassets/careers-community-growth-lead.png. - Site map:sitemap.jsonLGS-23 (no_ads,live_url) and its line inSITEMAP.md. The Site Studio board was rebuilt. - §6 and §8 updated. - 2026-11-02 (later) — David's answers folded into the job description: Lansil hires, the hire reports to David, no base pay, the LinkedIn messages go out from Alan's profile. §6 updated.
- 2026-11-02 — The community lead's job description,
clients/lansilglobal/hiring/community-growth-lead.md, plus the §6 entry (the role, the EcomFuel finding, the open calls). - 2026-10-26 (night) — David's screenshot: the mock-up sign-up refused "www.herliaison.com" ("Please enter a URL."). The "Store website" box in
kit.js(11 of 12 mock-ups) and the link boxes on t09/a and t09/c were browser url fields, which refuse anything without https://. They are now text boxes with the url keyboard and take a link however it is typed. Tested at 1280 and 390 wide with five ways of typing it: 0 refused, against 88 on the old kit. Build rule for the live tools: every website or link box takes it as typed and we clean it up on our side. - 2026-10-26 — David's two mock-up notes:
map.jsgains sample postage by zone and weight (rate,postage,weightChips); t07/a, b, c show postage both coasts against one; t09/a, b, c open on "Pay 20% less than your supplier" with the middleman line; t09/c rebuilt as "your store link" (44px tick boxes); board text, cards re-shot and re-audited at 1280/390;tripwires.jsonT07/T09 (davids_note,copy_rule, needs) + status; inputs card gainsT07-postageandT09-store. - 2026-10-26 — twelve tripwire mock-ups + board kind
mocks(tools/brand-hub/mock_board.py, dispatch + boards label in build.py),lansilglobal--tripwire-mocks.json, lane pointerlg-tripwire-mocks, inputs card updated, T02 parked on the tripwire board (chat),tripwires.json(T09 at 20% under Alibaba,picked= four,mockups),TRIPWIRES.md"The four, in build order". - 2026-10-25 (evening) — row 4d live on the client board (two wording fixes); tripwire picks folded (
tripwires.jsonverdicts +picked,TRIPWIRES.md"The five, in build order", the board note),lg-tripwires-boardpointer withdrawn,lg-tripwires-inputsposted. - 2026-10-25 (later) — Lane A on the client board + the tripwire board:
clients/lansilglobal/launch-board/amplify/index.html+ the row on the board (4d held in a stripped comment), deployed;tools/brand-hub/tripwire_board.py(new board kind),tools/site-studio/tripwires.py(writestools/brand-hub/boards/lansilglobal--tripwires.json,verdictsfolds the marks),tripwires.json(plain group labels,board_sections, short no-fit reasons),build.py(dispatch + boards label + Site Studio pointer + aboardpointer surface withpointer_url), lane:lg-tripwires-r1withdrawn,lg-tripwires-decisionsto Needs You,lg-tripwires-boardpointer, Quick link re-pointed;TRIPWIRES.mdpoints at the board. - 2026-10-25 — Tripwires per keyword + the diagram for Alan + nine old hub answers folded (§6, §8, §9):
briefs/lansilglobal/site/tripwires.json+TRIPWIRES.md,tools/site-studio/tripwires.py(new: maps every media-plan keyword to a tripwire, writes the CSV, the map, the filterable keyword page and the plan doc), the Site Studio tab's tripwire section (render_tripwiresintools/brand-hub/build.py; asks withsurface: "sitestudio"render their one interactive card in the tab and a pointer on Needs You, the Copy-Chief pattern generalised throughSURFACE_REF+ask_card), askslg-tripwires-r1+lg-tripwires-decisions+lg-amplify-diagram, Quick link; the three diagram lanes intools/brand-hub/docs/lansilglobal/amplify/{a,b,c}/, boardlansilglobal--amplify-lanes(cards indocs/lansilglobal/boards/amplify-lanes/), chief pass next to the client board;truth-gate.mdgains the 2026-10-12 verdict table; six kickoff questions in §9 marked answered.
- 2026-10-21 (night) — Google Ads media plan for the calculator built, waiting on the go (§6, §8, §9):
briefs/lansilglobal/media-plan-google-2026-11.md+.json(channel google-search; the twin shape = campaigns of ad groups on keyword groups, own headlines + shared spine, sitelinks/callouts/snippet, rounds, rules, decisions, preflight),clients/lansilglobal/ads/google/build_round1.py(dry run 1,181 keywords / 932 negatives / 15 RSAs, all lengths OK;--go= paused creation in ECOM LG FZE on David's word),COPYCHIEF-google-ads-r1.md+copychief-google-ads-r1-items.json(19 rows → asklg-copychief-google-ads-r1, Copy-Chief tab), asklg-google-media-plan-go(five decisions) + Quick link on the lane, plan page viamedia_plan_page.py(newcells_searchadapter + text-ad mock + headline/description lengths + keyword sample), the bill screen's phone shot atdocs/lansilglobal/media-plan/pages/LG-C1.jpg, the rivals' Google previews decoded intocompetitors/rival-ads-corpus-2026-10-21.{md,json}(Demand Gen copy only; RSA previews are iframes), client board row 4d (not deployed). - 2026-10-13 — Access ingest, day 1. The Drive Photo & Video Bank catalogued (6,640 files; 3,821 unique, 39.7 GB) and mirrored into the main checkout (git-ignored);
clients/lansilglobal/ingest/born —drive_bank.py,ga4_pull.py,gtm_pull.py,meta_pull.py,ghl_pull.py,_google.py; the Meta ad account, pixel and CAPI token recorded (token in~/.lansil/meta.env; probe shows it is CAPI-only, no reads); GA4 + GTM found granted to the Lansil login, not david@davidsanghera.com; the Drive API enabled on the analytics-mcp Cloud project through the Service Usage API;.gitignorefor the bank bytes; the client launch board updated (Last updated Oct 13). - 2026-10-13 (pm) — Meta read with our own system user: the ad account “Lansil Global V2” is shared to AMV LLC, “Lansil Pixel 2.0” lives inside AMV LLC (never fired), the old pixel fires PageView only; history pulled to
clients/lansilglobal/ads/meta/2026-10-13/. Two more Drive folders catalogued + mirrored: Brand Assets (73 unique files, 4.1 GB) and the Content Archive (43 docs →content-archive/Markdown). Bank mirror complete (3,340 fetched, 2,174 twin paths linked); 640px thumbnails made for the shot-selection board;drive_bank.pygains--set bank|brand|content,thumbsandtext. Asset Bank shipped the same evening: hub tab + board kindbank(tools/brand-hub/bank_board.py,render_assetbank), thumbnails onlansil-bank,skills/asset-bank/SKILL.md. - 2026-10-13 (late) — GoHighLevel pulled (
clients/lansilglobal/ingest/ghl_pull.py, browser-style User-Agent for Cloudflare, per-calendar bookings) →clients/lansilglobal/crm/2026-10-13/; §4 CRM row rewritten with the read; §9 asks (UTM capture, stage hygiene, Robert, bookings). Calculator numbers confirmed by David → Truth Gate p1–p5 + o4, §3, the spec's “what it needs from Alan” column resolved. - 2026-10-13 (evening) — Calculator lanes built on the engine (David: “whats next on the calculator”):
pricing/calc/+ four lane prototypes on staging, Playwright shots (desktop + phone, 0 console errors), the judging boardlansilglobal--calculator-lanes(kind verdict, 4 winners = rank), hub pick-board cards gain an “Open the page ↗” link (c.url). - 2026-10-14 — Google Ads read (David: Standard access on “Lansil Global” 873-360-9807 — the manager; 627-986-4143 under it holds the campaigns):
clients/lansilglobal/ingest/gads_pull.pyon the MGI puller (API v24 — v21 answers an HTML 404, sunset;clients/mgi/audit/pull.pybumped too), 24 GAQL queries →ads/google/6279864143/2026-10-14/. §4 rewritten with the read; the first read of the manager alone (0 campaigns) was the manager, not an empty account. - 2026-10-18 — Live bar + footer on the four screens, the home page bar, the
rates.lansilglobal.comcustom domain + host rules, the door's relative redirect (so the subdomain root works), lock small print reworded;custom_domainset through the Netlify API (an alias needs a primary first); staging verified (logo 200, pages 200, 0 console errors). - 2026-10-18 — Calculator split test built:
pricing/index.html(the door),lead.mjscounts views / opt-ins per screen in Blobs (+?stats=1, CORS for the hub),ui.jsposts one “view” per browser per screen (?selftest=1excluded),functions/package.jsonfor@netlify/blobs, the Site Studio tab's live card. Pair = bill + wizard. Verified on staging (door random + sticky, one view per browser per screen, the opt-in branch withtest:trueexcluded, counters zeroed with the reset:POST {kind:"reset", key}againstLANSIL_CALC_RESETon the site env; the secret lives in~/.lansil/calc.env, 600 — zero the counters again the day real traffic starts). - 2026-10-18 — Bill lane on phones: the two bills side by side (David: "possible to make the two bills side by side on mobile? somehow"). Under 620 px the two paper invoices give way to ONE receipt that shares the line labels: label left, Lansil's column (the $0 stamps) and the typical 3PL's column side by side, both totals, per order, per day, the one-time onboarding line and the green “you keep” line. Desktop keeps the two paper bills. Measured at 390 px: no overflow, the receipt 324 px wide, cells 13–15 px (amounts in the mono face). The board card re-shot.
- 2026-10-15 — CRM write scope landed →
ghl_fields.py ensure(10 fields; GoHighLevel derives the key from the name:calc_orders_range_explored,lead.mjsaligned), Netlify env via the API (env:setanswers Not Found for an unlinked site; a secret needs explicit contexts, not “all”), staging redeployed, the lock test created + updated a real contact, deleted after. Client board row 4. - 2026-10-15 — Dwight opens two Google Ads accounts (ECOM LG FZE 218-766-7313, Lansil-owned; 627-986-4143, agency-owned); David picks ECOM LG FZE for the relaunch.
gads_pull.pygains--login+ an automatic sign-in as the account itself for direct grants. The ECOM LG FZE read is blocked on the Google login re-auth (invalid_raptagain) — David re-runs the gcloud login, then the pull runs. - 2026-10-14 — Phone pass on the four lanes (David: "these should be optimized for a mobile view right?"). Measured at 390 px with Playwright before and after: no sideways scroll on any lane; the lock form was overflowing its blurred block by ~190 px on the slider lane (now in the flow above the block on phones); the dashboard lane's no-wrap ledger was forcing a 414–664 px layout (grid
minmax(0,1fr)+ wrapping cells); copy raised to 16 px, labels and eyebrows to 13 px, footnotes to 14 px; buttons full-width at 48 px, the range slider 44 px. The floors applied: copy 16 px · labels 13 px · footnotes 14 px · targets 44 pt (feedback_cro_usability_floors). - 2026-10-14 — The lock shipped on every lane (David: blur everything):
LansilUI.lock+lock.css, thelead.mjsNetlify function (create / update → GoHighLevel upsert + PUT withcalc_*custom fields resolved by key; prototype mode without env),products/lansilglobal/deploy.shnow ships--functionsand checks the function answers;ghl_fields.py ensure; end-to-end test on staging; lanes re-shot; spec v1.3. - 2026-10-14 — Competitor names off the calculator (David):
rivals.jsondown to the typical-US-3PL preset, the “Compare with” controls removed from the slider / bill / dashboard lanes, the wizard's preset buttons replaced by generic “who ships today” answers; lanes redeployed and re-shot, the board note and the spec (v1.2) updated. - 2026-10-12 — sitemap verdicts folded (21 KEEP); the new site's home page written, chiefed and live on staging (lansil-site.netlify.app) + in the Copy Studio;
products/lansilglobal/created (page, deploy.sh, netlify.toml + twins). - 2026-10-09 — kickoff call with Dwight (decisions in §6, accounts in §4, asks in §9); the client-facing launch board built and deployed (
clients/lansilglobal/launch-board/→ lansil-launch-board.netlify.app); Google Ads account 627-986-4143 identified (email-only invite). - 2026-10-08 — new-site decision recorded; sitemap v2 drafted (
briefs/lansilglobal/site/SITEMAP.md). Found on the current site:/direct-china-fulfillment301s to the beauty-brands page, and the keyword plan's China landing page was that URL — the new sitemap's pages replace every landing page in PLAN.md once approved. - 2026-10-06 — Angle Studio round 2: round-1 verdicts folded (GO 3 · KILL 4 · 20 good hooks); A03 / A04 / A10 rewritten from David's notes; A14 / A15 (his A13 note) and A16 / A17 (66 competitor videos transcribed) added; Truth Gate PENDING o1–o4; hooks batch 2 through the Copy-Chief gate (55 boarded, 14 withdrawn incl. A02's de-minimis lines); board round 2 deployed.
- 2026-10-06 — Keyword Studio verdicts folded (BID 16 · TEST 5 · SKIP 22, 4 notes) → PLAN.md + plan page on the hub; the tab shows the plan.
- 2026-10-04 — Keyword Studio tab live on the hub (every brand's nav; Lansil populated: 43 groups / 2,345 keywords, board
lansilglobal--keywords, competitor keywords inside the tab); keyword taxonomy fix (176 keywords out of off-target) + keyword pages rebuilt. - 2026-10-04 — keyword research pages live on the hub (8 competitors + Lansil baseline + index, SpyFu re-export 2026-10-04); profiles, competitor index and docs index link to them; staging markdown moved out of
docs/(it was being served publicly). - 2026-09-21 — Angle Studio: Hook Studio screen live (
lansilglobal--hooks, 65 hooks, batch 1); hooks judged there feed TASTE.md and the next batch (angles.py hooks --batch 2) lands flagged NEW. - 2026-09-16 — Angle Studio round 1: 13 angles + 65 hooks boarded (
lansilglobal--angles), competitor program converted into evidence, Deep Research VoC banked, ask lg-angle-board-r1. - 2026-09-12 — competitor intel program: Meta Ad Library, Google Ads Transparency Center, LinkedIn Ad Library, SpyFu (8 competitors + lansilglobal.com baseline) and Gemini DR profiles; findings: Google search is ShipMonk/ShipBob's; Portless owns the cash-conversion angle on Meta + LinkedIn; Lansil's own est. CPC ~$20 on ~$6.6K/mo (verify in-account).
- 2026-09-10 — client folder + Obsidian hub created; Alan's notes filed verbatim; DR pipeline staged and launched.
- 2026-09-11 — DR complete 7/7; MANIFEST + Truth Gate; distilled notes 03/04/05/09; founder interview verified; BRAND.md created; Launch Center workspace spun up (lane, asks, truth-gate, benchmark board, docs).
Appendix A · Research digest (Gemini DR 3.1 Pro, 2026-09-10/11)
- 01 Market Awareness — majority Problem Aware (45%), Unaware 35%, Solution Aware 12%, Product/Category Aware 6%, Most Aware 2% → 80% don't know an integrated partner exists. Dominant pains: tariff bills (de minimis), cash locked 90–120 days, vendor hand-off failures. TAM ≈ 1.44M viable stores (top 30% of ~4.8M Shopify storefronts; US ≈ 54%); spend estimates $45–65B (US brands) to $151B (Shopify global) — order-of-magnitude only.
- 02 Competitors — see Appendix B. Gaps to own, ranked: (1) tariff-proof hybrid, (2) FinTech logistics (credit), (3) anti-ShipBob/ShipMonk safe haven, (4) incubation-to-enterprise. "Hip, young, direct" decoded: interactive supply-chain timeline, self-serve calculators, dashboard-as-hero; founder-led content is the acquisition pattern; DayOne's uniformed-staff warehouse tours are the proven "real footage" format.
- 03 Psychographics — battle-hardened, cynical operator; trusts peer horror stories over vendor marketing; three prejudices (VC "Big 3PL", Instagram sourcing agent, Amazon-as-landlord); pre-installed villain story (the 2014–20 golden age taken by de minimis repeal, Amazon fees, RBF debt). Wants operational truth, never "automated fulfillment". Lansil's four triggers: founder's own QC loss → in-house QC; private + human vs the VC trap; 60-day terms vs the cash-flow death spiral (strongest); the post–de minimis hybrid.
- 04 → 06 Mechanisms (ranked) — see Appendix C.
- 05 Unified brief — five objections with answers (faceless VC / hidden fees / migration risk / comprador agents / jack-of-all-trades) and a promise-per-pain table; language to use = operator-grade finance vocabulary + proof words; avoid = "effortless", "AI routing", leading with baseline features.
- 07 Copy brief — 25/25 answered; spokesperson Alan; nicknames "the Negative Cash Conversion Cycle method" · "the Legal De Minimis Bypass" · "the Golden Sample Shield"; three offer architectures (China-direct / US omnichannel / fully integrated hybrid); lead hooks: The 2026 De Minimis Death Trap · The Cash Flow Death Spiral · The Big-3PL Betrayal (the SaaS illusion) · The 19th-Century Comprador Trap · The Amazon Algorithmic Tax. Truth Gate applies (Appendix D).
Appendix B · Competitor battlecards (DR 02, 2026-09-11)
| Competitor | Model + footprint | Acquisition | Hero / hooks | Pricing | Weak spot | China + US? |
| Bestfulfil | China-only private agent | SEO/blog, sales-led | "one-stop service platform for global sellers"; 1-1 agent; custom packaging | quote-only | manual, no dashboard | No |
| EcommOps | Shenzhen, China-direct only | founder podcast/keynotes, Clutch 5.0 | "The Right Way to Fulfill From China"; 5–8 d; 90% less cash in inventory | quote; min 30–35 orders/day or 500 units | gatekeeps small brands; exposed to de minimis end | No |
| DayOne | Ningbo + Dongguan, 500+ staff, 120K orders/day | "DayOne University" SEO, warehouse-tour videos | "We simplify dropshipping and sourcing from China"; discreet tracking | quote, prepaid | prepay friction | No |
| Ecomflow | HK/China, ~13 staff, ~$1.6M | founder LinkedIn, technical content | "Streamline your D2C supply chain… from China to the world"; origin masking | published rate list | tiny; volumetric shock | No |
| Portless | Shenzhen 170K sq ft; $18M Series A 2025 | PR, CEO podcasts, investor intros, ROI calculators | "Ship from Asia straight to your customer's door"; Shein blueprint; 3× cash flow | transparent all-in | heavy items; exposed to de minimis end; doesn't source | No |
| ShipMonk | 12+ US/global facilities | paid search, integrations | "Tech-enabled fulfilment network" | published + monthly minimums | Reddit: hidden fees, markups | No (no sourcing) |
| ShipBob | 50+ facilities worldwide | massive paid + CRO, Shopify app | "Global Omnichannel Fulfillment" → Request a Demo | opaque | Reddit: delays, lost inventory, ghosting, double billing | No (no sourcing) |
| NextSmartShip | Shenzhen 4PL + own US/UK/DE/AU warehouses + sourcing division | tariff content, calculators | "Fulfillment Simplified!"; free onboarding, 60 d free storage, no minimums | published $4.10/0.9 kg | thin proof (39 brands) | YES |
Appendix C · Mechanisms, ranked (DR 04 + 06)
| Rank | Problem mechanism → Solution mechanism | One line | Score |
| 1 | The DPO Deficit → Capital-Integrated / Liquidity-Matched Fulfillment (60-day terms) | "You pay the factory after the product has already sold." | 9.5 |
| 2 | The Single-Parcel Tariff Trap → Omni-Shore Tariff Dilution (bulk entry from NV/PA) | "One customs entry, not ten thousand." | 9.0 |
| 3 | Post-Freight QC Liability → Point-of-Origin Verification | "Find the defect in Shenzhen, not in Pennsylvania." | 8.5 |
| 4 | Siloed Stock Syndrome → Dynamic Omnichannel Allocation | "One pool of stock, every channel." | 8.0 |
| 5 | Multi-Vendor Bullwhip → Single-Custody Architecture | "No hand-offs, no bullwhip." (must never sound like SaaS) | 7.5 |
Paradigm to land: your supply-chain architecture is obsolete, not your 3PL is bad. Framework mapping hypothesis: (3) trade editorial ← #2 for cold traffic · (1) founder + customer ← #3 · (4) Top-5 ranking ← #1 + #2 as scorecard criteria for Solution-Aware search/retargeting · (2) investigation — park.
Appendix D · Truth Gate — do NOT use before Alan confirms (full file: truth-gate.md)
- Invented guarantees (07 Q18/Q20): no hidden receiving fees · no carrier markups · no exit fees · free factory rework on QC failure.
- Invented scarcity (07 Q22): "caps new merchant integrations per month", "refuses to outsource labor".
- Anonymous "UK electronics brand pivot" case study (07 Q15) — treat as fabricated until a real client is supplied.
- Conflicting facility sizes (see §2); conflicting company numbers (see §0).
- Founder-story details from press not yet re-verified: Cobh, 2010 Shanghai vs 2011, age 27, "built from a backpack", fluent Chinese, "flew to Nevada, secured the warehouse and hired staff in a week".
- Tariff specifics ("June 24, 2026 suspension", Entry Type 86/11/01, "12.5% forced-labor duty", "$4–25 per-parcel broker fee") — primary-source before use.