Web Analytics · Google Analytics & GA4
GA4 Event & Conversion Tracking
One agreed name per action, one agreed rule for what counts as a key event.
When teams use different names for the same action, GA4 stops giving them a shared view of performance. We rebuild the event and conversion setup so each number has one agreed meaning.
An event and key-event setup your team can read, test, and keep tidy long after we leave.
Some of the 500+ brands we've worked with
See all referencesHow we run it
From disputed definitions to an agreed event model
This work stays inside GA4, while deployment mechanics remain with GTM. Each step resolves a specific decision and produces an artifact your team can review.
How we hold ourselves to it
- Business actions, turned into one agreed event model — We identify the actions that matter, including signups, checkouts, leads, and key page views, and turn them into a clear GA4 event model. Each event gets one agreed meaning.
- One naming convention, applied to every parameter and key event — We write the naming convention, the parameters each event carries, and the criteria for marking an event as a key event. No more guessing which of five similar events feeds a report.
- Where GA4 configuration ends and GTM begins — We review or set up the event and key-event configuration inside GA4 and cross-check where the GTM boundary sits, so nothing is configured twice or left in no man's land.
- What DebugView confirms and what it cannot see — We replay representative journeys in DebugView and standard reports, record what GA4 cannot observe, and document the exceptions so they are not mistaken for defects later.
Align on actions and definitions
We agree on the list of business actions, what each one is called, and which ones deserve key-event status. This is where disputes about meaning get settled, before they turn up in a report six months later.
Agreed action list
- AI assist
- Scans existing GA4 properties to extract active event schemas and unmapped parameters.
- Human gate
- Measurement specialist confirms business event definitions with product and marketing leads.
- Owners
- Measurement Analyst, Business Stakeholder


Map the implementation
We compare the agreed model with what GA4 currently collects and map which events should exist, which parameters they should carry, and where GTM takes over.
Event mapping sheet
- AI assist
- Compares agreed events against what GA4 already collects.
- Human gate
- Analyst confirms the target-state mapping is buildable.
- Owners
- Measurement Analyst, GTM Owner


Test the journeys
We replay the journeys the business relies on, such as signup, purchase, and lead submission, in DebugView and confirm that events and parameters arrive as designed.
Journey test notes
- AI assist
- Flags events or parameters missing in DebugView.
- Human gate
- Analyst decides whether a flagged gap is a real defect or just noise.
- Owners
- Measurement Analyst, Engineering Contact


Reconcile and report
We compare GA4 numbers against the sources you trust, explain the gaps we can explain, and flag the ones that come from consent, blockers, or platform behaviour.
Reconciliation summary
- AI assist
- Calculates variance between GA4 and trusted sources.
- Human gate
- Analyst attributes each gap to consent, blockers, or platform.
- Owners
- Measurement Analyst


Hand over and set recheck triggers
We review the documentation with your team and agree on the changes that should trigger another check, such as a site relaunch, consent update, or new platform feature.
Handover session
- AI assist
- Drafts recheck triggers from the events most likely to drift.
- Human gate
- Team agrees which triggers actually warrant a re-check.
- Owners
- Measurement Analyst, GTM Owner


Four parts of GA4 configuration, with deployment kept in GTM.
Automation scans the property for active event schemas and unmapped parameters, compares the agreed events against what GA4 already collects, flags what DebugView cannot see, and calculates the variance against trusted sources. Meaning stays with people: what a business action is, whether a gap is a defect or noise, and what to attribute the remainder to.
What you get
An event dictionary that answers the question before someone has to ask.
Every document here is meant for daily use.
Working document
Event dictionary
Every event with its agreed name, its meaning, and the business action behind it. The single place to check what anything in GA4 is supposed to be.
Accepted when
Every event name in the property appears in the dictionary with the business action behind it.
Cadence: Extended with every new event
Configuration specification
Parameter and key-event specification
The parameters each event carries, their expected values, and the exact criteria that make an event a key event in your property.
Accepted when
Each key event states the criteria that make it one, not just the checkbox.
Cadence: Revised when a key event changes
QA notes
Validation notes
What we tested, what passed, what we could not observe, and why. Includes the DebugView and report evidence from the journey tests.
Accepted when
What could not be observed is written down as unobserved, with the reason.
Cadence: Captured per validation pass
Runbook
Maintenance and handover guide
How to add a new event without breaking the convention, who owns what, and when the setup needs a fresh look.
Accepted when
A new owner can add an event from the guide without breaking the naming convention.
Cadence: Updated at handover
We call it done when: the journeys replay cleanly in DebugView, the counts reconcile against a trusted source inside an agreed variance, and the negative cases produce nothing.
Fit and readiness
The team needs one definition of what counts as a conversion.
Here is how to tell whether this work fits the problem.
A good fit when
- Reports show the same action under several event names, or nobody is sure which event counts as a conversion.
- A new site, app, or consent setup is coming and the current event model will not survive it unchanged.
- Marketing, product, and analytics use different key-event definitions, so the same business action changes meaning from one report to another.
- A business owner can say what each action means, but the GA4 event model has never captured those definitions.
Better handled as other work when
- You need tags, triggers, or containers built or repaired. That is GTM work, and we will direct you to the appropriate task.
- There is no one available who can confirm what the business actions actually are, so events would be named in a vacuum.
If one of these is closer to your situation, start here instead: All Google Analytics & GA4 Consulting tasks
We call it done when: product and marketing have agreed what each business action means before a single event name is written.
People who build your measurement system
Zeo designs measurement systems that connect a business decision to governed collection and reporting you can check. The people shown here work on the part of that system this page covers.

Zafer Yıldız
Web Analytics Manager

Abdullah Tanıdır
Performance Marketing Team Lead

İlker Emir
Senior Performance Marketing Executive

Sevda Yurtvermez
Performance Marketing Team Lead

İpek Ezer
Performance Marketing Executive

Serap Yurtvermez
Performance Marketing Team Lead

Onur Durdağı
Performance Marketing Executive

Deniz Çağın Demirci
Frontend Developer

Mirzamin Aghazada
UI/UX Designer

Yağmur Bayram
Sr. SEO Analyst

Gülşah Şahin Özkan
Senior SEO Analyst
Tools we use
Tools behind this work
Google AnalyticsDebugView confirms an event arrived with the right parameters, not just that it fired
Google Tag Managerwhere every mapped event actually gets deployed, kept apart from the GA4 settings
Google Tag Assistantwalks the test journeys step by step to confirm each tag actually fired
Next step
Resolve the event model before relying on the next report


Before we start
Questions teams ask before booking
No. Tags, triggers, and containers belong to our GTM work, and we keep that boundary explicit. Here we own how events and key events behave inside GA4, and we coordinate with whoever handles deployment so nothing is configured twice.























