GA4 event architecture
An event and parameter model designed around your funnel, not the default recommended list — so reports answer questions you actually have.
The unglamorous work that decides whether every other marketing decision is based on something real.
Almost every account I audit is optimising toward a number that isn't true. Conversions firing on page load. Thank-you pages counted on refresh. Phone calls — often the majority of leads — invisible to the ad platforms entirely. Purchases in GA4 that don't reconcile with what the payment processor says.
None of that is exotic. It is just nobody's job until it is, and the cost compounds quietly: bid strategies trained on bad data spend real money learning the wrong lesson.
This engagement fixes the measurement layer end to end and hands you a documented map of what fires, when, and where it lands.
01 — What's included
An event and parameter model designed around your funnel, not the default recommended list — so reports answer questions you actually have.
Clean container, versioned changes, no orphaned tags. Triggers documented so the next person can follow them.
Google Ads and Meta conversions wired to the events that represent real money, including offline uploads when the deal closes by phone or in person.
Server-side events with deduplication, so post-iOS attribution holds up without inflating your numbers.
CallRail or CallTrackingMetrics with dynamic number insertion, so phone leads get attributed to the campaign that produced them.
Firebase events, SKAdNetwork configuration, and App Store Connect reconciliation for iOS portfolios.
02 — Proof
On a two-app portfolio the media looked broken — installs were cheap, subscriptions weren't coming. Correct funnel measurement showed the drop was at the paywall, not the ads. That fix cost nothing in media spend.
03 — How it runs
Buying against a broken signal just spends the budget faster, so the measurement work always comes first.
Get a free account review →Every tag, trigger, and conversion action currently firing, plus what each one actually measures versus what it claims to.
Decide what a conversion means in your business before deciding how to track it. This is the step most implementations skip.
Implement in GTM, then test every path in preview mode and DebugView — including the ones that only fire on mobile.
Match platform numbers against your source of truth, then hand over a written map of the whole setup.
04 — Questions
A standard site is usually one to two weeks from access to handover. Multi-domain setups, apps with Firebase, or anything with offline conversion imports take longer because validation takes longer.
Usually it can be fixed. A rebuild is only worth it when the historical data is already unreliable enough that you wouldn't trust year-on-year comparisons anyway — in which case a clean start costs less than perpetual caveats.
Most of it is Tag Manager and doesn't need a developer. For dataLayer pushes on custom checkouts or app SDK work, I write the spec and work alongside your engineers.
Send over the account, the vertical, and what's not working. You'll get a reply within two working days and the first two things I'd change — free, whether or not you hire AQM.