Every SEO decision you make rests on your analytics being correct. In audits, it usually is not.
We audit analytics before we audit anything else, because there is no point recommending changes measured against numbers that are wrong. In the majority of properties we review, at least one of the following is true.
1. Payment gateways missing from referral exclusions
A customer goes to your checkout, is redirected to a payment provider, and comes back. Without a referral exclusion, GA4 treats the return as a brand new session sourced from the payment provider. The result is that organic search gets credit for the visit but not the purchase, and your most valuable channel looks like your least valuable one.
2. The tag deployed twice
Usually one instance hard-coded in the theme and another added later through Tag Manager by someone who did not know. Every pageview fires twice, sessions inflate, and engagement rate collapses because the duplicate hit registers as immediate activity. Check by looking for two config events on a single page load in the network tab.
3. Conversions firing on page load
A “thank you” page conversion that fires whenever the page loads counts every refresh and every back-button return. We have seen conversion counts inflated by 40 percent this way. Fire on the actual form submission event, or deduplicate on a transaction identifier.
4. Data retention left at two months
GA4 defaults event-level data retention to two months. Most teams never change it, then discover during their annual review that year-on-year comparison is impossible because the data no longer exists. Set it to 14 months on day one. It is one dropdown, and it cannot be applied retroactively.
5. Internal traffic never excluded
Your own team is usually your most active set of visitors. Without an internal traffic filter, staff sessions inflate engagement on exactly the pages your team works on most, which are rarely the pages your customers care about.
6. Search Console not linked
Linking Search Console to GA4 takes two minutes and gives you query data alongside behaviour data in one place. It is free, and the majority of properties we audit have never done it.
Validate before you trust
Pick a single day. Compare GA4 transactions against your actual order records, and GA4 organic sessions against Search Console clicks for the same period. They will never match exactly — different measurement models — but a variance above roughly 10 percent means something is wrong, and it is worth finding out what before you build a strategy on top of it.