Skip to content

Tracking

A GA4 audit checklist for UK ecommerce sites.

By Felix Grant5 min read

A useful GA4 audit checks four things in order: the property is set up correctly, consent is handled lawfully, ecommerce events fire once with the right values, and the reports people use are built on those events. This is the list we work through on every ecommerce audit, written so you can run it yourself.

Prepare

Before you start

You need Editor access to the GA4 property, access to Google Tag Manager if you use it, an order export from your ecommerce platform for the last full month, and a test card or test mode on your checkout. Set aside a day for a small store.

Keep a simple sheet as you go with three columns: what you checked, what you found, and how much it matters. That sheet becomes your fix list at the end.

Property

Check the property settings

These take minutes to check and cause long running problems when they are wrong. Start with the time zone and currency. They should match how Finance reports, because a property set to US Pacific time will split UK trading days in the wrong place.

Set data retention to 14 months rather than the default of 2. Explorations can only look back as far as retention allows, and you will want year on year comparisons later.

Check that internal and developer traffic is filtered, and that the filter is active rather than left in testing. Add your payment providers, such as PayPal, Klarna and your card gateway, to the list of unwanted referrals so returning buyers are not credited to them.

If checkout or booking sits on another domain, set up cross domain measurement. Link Google Ads and Search Console so paid and organic data can be read alongside site behaviour. Finally, make sure there is one web data stream per site. Duplicate streams on the same site double count.

Events

Walk the journey in DebugView

GA4 collects some events automatically, but the ones that matter for ecommerce usually need setting up. Open DebugView and walk through a full journey: land, browse a category, view a product, add to basket, begin checkout and pay.

Each step should send the recommended event name, such as view_item, add_to_cart, begin_checkout and purchase, and each should carry the items with ID, name, price and quantity.

Look for events firing twice. The most common cause is a platform app and a hand built tag both sending the same event. Then check that enquiries, sign ups and other actions that matter to the business are tracked as their own events.

Mark the events that matter as key events, and only those. A list of twenty key events tells you nothing.

Purchase

Give the purchase event its own check

This is where most revenue problems live. The purchase event should carry a transaction ID that matches the order number in your platform, a currency on every purchase, and a value that is calculated the same way on every checkout route. Decide whether value includes VAT and shipping, and make every route do the same.

Reload the order confirmation page and check that GA4 does not record a second purchase. Then test every payment method, including those that send the customer to another site to pay and back again.

Finally, match one month of GA4 transaction IDs to your order export. Count the orders missing from GA4 and group them by payment method. Missing purchases are rarely random, and the grouping usually points straight at the cause. Check how refunds are handled too: either send refund events or note that GA4 revenue is gross of refunds.

Missing purchases are rarely random. Group them and the cause shows itself.
What the order match usually reveals

Attribution

Check where credit goes

Every paid campaign link should carry UTM parameters, written the same way each time. GA4 treats facebook and Facebook as two different sources. Google Ads auto tagging should be on, with the accounts linked.

Look at the Unassigned channel. A large share usually means broken UTMs, missing referral exclusions or tags firing before the page knows where the visitor came from. A high share of revenue under Direct often hides email, app or payment provider traffic that lost its source. Self referrals, where your own domain appears as a referrer, point to cross domain or subdomain problems.

Reporting

Check what people actually use

Tracking that is right but unreadable still fails. Ask who opens GA4 each week, and which report, and build for those people.

Check custom dimensions and metrics are named clearly and still used. If you report in Looker Studio, check each chart's date range and filters. Confirm the revenue figure in the main dashboard matches a figure Finance recognises, with any difference written down. And switch on the BigQuery export if you will want raw data later, because it only collects from the day it is turned on.

Prioritise

Sort what you find

An audit usually turns up more than you can fix at once. Sort findings into three groups. Fix now covers anything that makes revenue or consent wrong, such as a purchase event firing twice or analytics cookies set before consent. Fix next covers anything that makes channel credit wrong, such as missing referral exclusions or inconsistent UTMs. Tidy later covers anything that makes reports harder to read.

Fix the first group before you change budgets based on GA4. Fixing the second group often changes which channels look profitable, so expect some uncomfortable conversations.

When we run this audit for a client, we also match a month of orders to GA4 transaction by transaction and write the reconciliation up line by line. That is the part most teams do not have time for, and it is what turns the numbers feel wrong into a list of fixes. Ours is part of the Discover audit at £1,950.

Key takeaways

  • Check time zone, currency, retention, filters and referral exclusions first
  • Test consent with a private window and your browser's developer tools
  • Make sure the purchase event fires once, with an order ID and a consistent value
  • Match a month of orders to GA4 and group the missing ones by payment method
  • Fix anything that makes revenue or consent wrong before changing budgets
Felix Grant

Written by

Felix Grant

Founder of Kitebox, a business technology and AI consultancy that builds what it recommends. Based in Mallorca, working with UK businesses.

About Felix →

Want this set up for your team?

Book a call and we'll show you how the queue, veto window and logs would work for your content.