Proposed -- needs human approval

This tower does not push to git. To approve, a human makes this exact edit and commits it:

1. Open docs/marketing/journal/005-analytics-ga4-datalayer-only.md in the docs repo.
2. Change the line "- **Status:** proposed" to "- **Status:** active".
3. Commit and push. The journal's git history is the audit trail; this tower does not commit on your behalf.

005: Behavior analytics is GA4 through the existing GTM, dataLayer-only; no new vendors

Date: 2026-08-30
Decider: engine proposal (Claude), pending Daud approval of the GTM container config step
Status: proposed
Review by: 2026-10-15

Decision

Full-funnel behavior instrumentation (view_item, view_item_list, select_item, begin_checkout, add_shipping_info, add_payment_info, purchase, plus voucher_applied and newsletter_subscribe) ships as typed dataLayer pushes through the existing lib/analytics.ts bridge into the GTM container already on the page. GA4 standard event names are used so built-in reports work without custom setup. No second analytics vendor, no new scripts, no client-side ad pixels. Full spec: docs/marketing/specs/analytics-tracking-plan.md.

Evidence

Recon (2026-08-30) found near-zero event coverage: the one ecommerce-shaped call (view_item) was wired to a documented no-op stub (lib/utils/legacy-stubs.ts:20-22) and the entire checkout path had zero analytics calls. GA4 could see pageviews at best; behavior could not be monitored, which is the gap this closes.
A dataLayer push is an in-memory array append; the GTM container script is already loading today, so instrumentation adds no network requests and no script weight. This satisfies the measurement spine's speed rule by construction.
The site just completed a CWV overhaul; adding any new vendor script (PostHog, Clarity, etc.) would spend that budget for capabilities GA4 plus growth.conversion_log already provide. Session replay is the only real gap and stays a journal-gated future experiment with a before/after CWV read.
Revenue truth remains Stripe + growth (blueprint first principle 1); GA4 purchase.transaction_id equals the PaymentIntent id so the layers reconcile without GA4 ever becoming the reward signal.

Predicted effect

Within 7 days of the GTM tags being configured (the human runbook step), GA4 shows a complete tour-to-purchase funnel with 95 percent or more of Stripe-confirmed bookings appearing as purchase events (client-side loss from blockers is expected; the growth server feed covers it for ads). CWV (LCP, INP, CLS) in CrUX is unchanged over the following 28 days.

Measured effect

(pending)