Service

GA4 that answers business questions

An event taxonomy designed around your funnel, implemented in GTM, verified against the platforms, and documented so it stays true after the next release.

Most GA4 properties were migrated, not designed. The events came across from Universal Analytics, the ecommerce schema was half-filled, and nobody wrote down what any of it means — so the reports are technically populated and practically unusable. This rebuilds the measurement from the questions you actually need answered.

You probably need this if

Nobody can say with confidence what a given event in your property means
Ecommerce reports are missing items, revenue, or both
Key events fire more than once per conversion, or not at all
Your funnel reports break every time the product ships
The property was migrated from Universal Analytics and never revisited

What gets delivered

An event taxonomy: Named around your funnel rather than the GA4 template. Two hundred events is not a taxonomy; the thirty-five that drive decisions are.
A dataLayer specification: What the front end must push, when, and with which parameters. The document your developers work from.
GTM implementation: Tags, triggers and variables built and versioned in your container, with a QA pass against GA4 DebugView.
Ecommerce and key events: The full purchase funnel where it applies, with the item schema populated properly so revenue reconciles.
BigQuery export configured: Because the GA4 interface will eventually stop being enough, and the export only backfills from the day you turn it on.

How it runs

Questions firstWhat decisions should this data support? Everything downstream is derived from that list, which is why it is written before any tag is touched.
Taxonomy and specEvents and parameters designed, reviewed with you, and written up as a dataLayer spec for the front end.
Implement and QABuilt in GTM, verified event by event in DebugView, checked against the ad platforms.
Document and hand overNaming conventions, the spec, and a walkthrough so the next feature ships with tracking instead of a follow-up ticket.

Scoped projectScoped4–8 weeks · fixed scope agreed before it starts

If you mainly need to know whether the current property is trustworthy, the $3,000 attribution audit answers that first.

Written up in detail

GA4 vs Amplitude for a Superapp: Cost and TaxonomyGA4 is free until superapp volume turns its quotas into your architecture. A comparison grounded in a 200-event cleanup, a vendor-selection memo, and the published limits of both tools.GA4 AI Assistant Channel: What It Does and MissesGoogle shipped a native AI Assistant channel in May 2026. It classifies on the referrer header - and roughly 70% of AI traffic arrives without one. What the native channel covers, the custom channel group that fills the gap, and how to size the part that lands in Direct forever.Amplitude Event Taxonomy for a Superapp: 200 to 35Two hundred events, eighteen prefixes, and nobody could answer how many people bought a policy last month. How a superapp event taxonomy collapses to thirty-five with one required parameter, why the money event has to leave the browser, and the migration step everyone skips.First-Party Data and DWH: Why Build Your OwnGDPR, CCPA, and the death of third-party cookies make dependence on Meta and Google dangerous. We look at how enterprise companies are saving marketing with the Modern Data Stack and Reverse ETL.

Questions I get asked

Can you fix our existing property instead of rebuilding?

Often, yes. The audit establishes which it should be. Rebuilding a taxonomy is disruptive and I would rather not recommend it when a correction will do.

Do we need BigQuery?

Not on day one, but turn the export on regardless. It only collects from the moment it is enabled, so switching it on late costs you history you cannot recover.

Should we be on GA4 at all, or something like Amplitude?

Depends on whether your questions are marketing questions or product questions. I have written a full comparison at superapp scale rather than repeating it here.

Who writes the dataLayer code?

Your developers, from the spec. If you would rather I did it, that is scoped as part of the project.

Find out what is wrong first

The attribution audit establishes whether your property needs a correction or a rebuild, before either is paid for.

Book the audit call