GA4 setup and audit
We first write down which decision needs which data; then we set up GA4 to that plan, or audit the existing setup item by item. The job is finished when the numbers have been checked against real order and form records.
What is a GA4 setup?
A GA4 setup opens a Google Analytics 4 property in the business’s account and measures purchases, form submissions and similar actions as events under a written measurement plan. Adding the code takes minutes; the real work is events firing once at the right moment, key events, consent mode, the Google Ads link and testing. An audit checks the same items.
What it covers
Measurement plan
The actions that make the business money, the name and parameters of each event, and which event will feed ad bidding are written in one document.
Property and core settings
The property is opened in your account. Time zone and currency, a single web data stream, data retention, the internal traffic filter and unwanted referrals are set.
Events and key events
Events fire from one source, Google Tag Manager: a successful submission on forms instead of a button click, and on purchases a unique transaction ID, value, currency and the list of items.
Consent mode
We test that the choice made in the cookie banner really reaches the Google tags. The choice between basic and advanced implementation is made with your legal adviser’s view.
Google Ads link and server-side tagging
GA4 and Google Ads are linked, and we settle which conversion is primary so that the same sale is not counted twice. Server-side tagging is set up where there is a case for it.
Testing and documentation
Every event is checked with Tag Assistant and DebugView, and a real test transaction is made. What each event counts is written into an event dictionary.
How it runs
- Inventory and audit.
In the first week we list the existing property, the tags on the site, the Tag Manager container and the conversions in the ad accounts. Tags that load twice and missing parameters are written up as findings.
- Measurement plan.
Which decision needs which data is written down and agreed with you. Whether server-side tagging is needed is also settled as a written decision in this document.
- Setup.
Tags are built in Tag Manager, tried in preview mode and published as a version. Old tags added straight into the site’s code are cleared out.
- Testing and comparison.
A form is filled in and a test order placed. We write down how many times each conversion appears in GA4, in the ad platform and in the CRM, and compare the result with real records.
- Handover and monitoring.
The setup document and the event dictionary are handed over. If the work continues, changes to measurement and any notable gaps are in the Monday report.
What you hold at the end
- A written measurement plan
- A GA4 property and Tag Manager container set up in your account
- A list of audit findings
- Consent mode set up, with a test record
- The Google Ads link and conversion settings
- A setup document that includes the event dictionary
When this is not the right thing to buy
- If you expect GA4 and the ad platforms to show the same number, we cannot deliver that: the tools count differently. What we can do is make the gap explainable and stable from week to week.
- If access to the site’s code or to Tag Manager cannot be given and we cannot reach a developer, the e-commerce events will be left half done.
- If all you want is the measurement code added to the site, that takes minutes and you do not need an agency for it.
Questions
The questions we hear most about this work. If yours is not here, write to us and we will answer within one working day.
How long does a GA4 setup take?
Adding the code takes minutes; most of the time goes on the measurement plan, the e-commerce events and testing. The first week is the inventory and audit, then the plan is agreed and the setup moves forward in weekly cycles. The total depends on the site’s platform, the number of events to measure, how quickly a developer can be reached and whether server-side tagging is needed. That is why it would be wrong to name one timescale before seeing the inventory.
What do you need from us for a GA4 setup?
Administrator access to your GA4 account, access to the Google Tag Manager container and the ad accounts, access to your cookie banner tool, and a line to a developer who can change the site’s code. For the comparison we also need to look at the system that holds your order or form records. If you have no account, we open the property and the container in your name and add ourselves only as a user.
What determines the fee for a GA4 setup and audit?
Whether the work is a setup from scratch or an audit and repair of an existing one, the number of events to measure, the scope of the e-commerce events, the number of sites and domains, the ad platforms to be linked and whether server-side tagging is needed. With server-side tagging, hosting is a separate item, paid to the provider from your own account. The first meeting is free; 2 to 3 days later we send a proposal that lists each item.
Will you rebuild our existing GA4 setup or repair it?
Most of the time we repair it. Deleting the property and opening a new one leaves the historical data behind, so we first audit it and write down what works, what counts twice and what is missing. We bring the tags together under one source in Tag Manager and make the event names consistent. When the definition of an event changes, we note the date in the setup document, because the numbers before and after that date do not count the same thing.
Will GA4 and Google Ads numbers match after the setup?
Not exactly, and they should not be expected to. The two tools credit a conversion under different attribution rules and to different dates, and visitors who decline cookies widen the gap. The measure of a sound setup is that the gap is stable from week to week and can be explained. A sudden jump usually means a tag, a consent setting or a redirect in the payment step has changed.
Do GA4 numbers drop once consent mode is set up?
Yes, they can. Events from visitors who decline cookies are no longer measured with cookies, so the observed numbers may fall. In the advanced implementation Google models part of that gap as an estimate; an estimate is not a real count and should not stand in for the figures in your till. The choice between basic and advanced implementation also has a legal side, so we make it with your legal adviser’s view.
Do we need server-side tagging?
Not every site does. First the Tag Manager structure, conversions fired from a single source and consent mode have to be clean. Server-side tagging is worth considering if the ad budget is large, if part of the sale is completed on the payment provider’s page or away from the site, if Meta ads carry a significant share of sales, or if sales close on the phone. We write the decision, with its reasons, into the measurement plan.
Once the setup is fixed, is our historical data fixed too?
No, it is not. GA4 does not reprocess data it has already collected under new rules; a fix applies from the day it is made. Data deleted while the retention period was short does not come back either. That is why we write the date of every change into the setup document. When you need to compare with an earlier period, relying on order and CRM records is sounder than trusting old GA4 numbers.
Further reading
- GA4 setup checklist: from property to DebugView testsFrom the property to consent mode, from internal traffic to testing: the items to work through in order to set up GA4 properly from day one.
- GA4 vs Google Ads: why the conversion numbers differWe separate, step by step, why the numbers drift apart, which gaps are normal and which point to a mistake in the setup.
- Measure before you raise the ad budget: four questionsFour questions to ask before you put more money into an ad account.
From the same service
- Data and CROThe whole service: measurement, not guesswork
- Conversion rate optimisation (CRO)We first find where people get stuck on the site, using data and session recordings; then we decide by testing changes. If your traffic is not enough for an A/B test, we say so at the start and suggest another route.