Why GA4 Numbers Don't Match Google Ads or Other Tools

Put GA4 next to Google Ads, a second analytics tool, or your own database, and the numbers disagree. They always disagree. The productive question is not how to make them match, which is impossible, but understanding each gap well enough to know which number to trust for which decision.

Different tools count different things

Most "discrepancies" dissolve once the units are stated precisely. Google Ads counts clicks; GA4 counts sessions. A visitor can click twice within one session, or click and leave before your page and its tags finish loading, so clicks reliably exceed ad-attributed sessions.

Users are worse. GA4's user count depends on its reporting identity: cookies, possibly blended with Google signals and modeling. Another tool with another identity model counts a different population. Even within GA4, users is an estimate, subject to cookie deletion, browser privacy features that limit cookie lifetime, and cross-device visits counted as separate people.

Your backend, meanwhile, counts facts: orders that exist, signups that exist. No tag-based tool will match it exactly, because tags can be blocked, race the page unload, or fire twice. When totals matter, the backend wins by definition.

Attribution: the same conversion, credited differently

The Google Ads gap deserves its own section because it burns the most ad budget on confusion. Google Ads reports a conversion when someone who clicked an ad converts within the Ads attribution window, and it books that conversion on the date of the click. GA4 books key events on the date they happen and, in its attribution reports, distributes credit across channels using its attribution model rather than giving the ad full credit.

So a click on the 28th converting on the 3rd appears in different months in the two tools, and a conversion that touched search, email, and an ad may count as one full conversion in Ads while GA4's reports assign it fractionally or to another channel entirely. Add differing conversion definitions, view-through conversions that Ads can count and GA4 cannot see, and imported versus tagged conversions, and a 20 to 40 percent gap between the two is unremarkable. Neither tool is lying; they answer different questions. Ads answers "what did my ads cause, by Ads' rules". GA4 answers "what happened on the site, credited by GA4's rules".

The missing visitors

Beyond definitions, GA4 genuinely fails to observe part of your audience. Ad blockers and browser tracking protection block the GA4 script outright for a meaningful share of visitors, higher among technical audiences. In the EU, consent rejection removes visitors from observed data, and if consent mode's modeling is active, estimates are blended back in, which makes GA4's totals partly synthetic and even harder to reconcile against a tool that observes directly. A server-side or cookieless tool will typically report more visitors than GA4 for the same site for exactly these reasons.

Then come the mundane causes that account for more discrepancies than anything exotic: filters excluding internal traffic in one tool but not another, timezone differences shifting a day's traffic across midnight, one tool counting bot traffic the other excludes, cross-domain tracking configured in one place only, and thresholding or sampling quietly reshaping GA4's own reports.

A workable reconciliation approach

Pick a source of truth per metric and write it down. Revenue and signups: the backend. Ad spend efficiency: the ad platform, with its window stated. Site behavior and relative trends: your analytics tool. Then compare tools only on percentage terms and direction, not absolutes; if both show conversion rate up 15 percent after a launch, the launch worked, whatever the raw counts say.

When a gap changes suddenly, that is the real signal. Stable disagreement is definitional; a moving one means a tag broke, a filter changed, or a consent banner shipped.

Where Analyse fits in this picture

Analyse does not claim to match GA4 or Google Ads, because no honest tool can. What it offers is a simpler number to reason about: every visit is recorded server-side without consent loss or modeled infill, so counts are observed rather than partly estimated, and one stated definition of a visit applies everywhere in the dashboard. When you reconcile against your backend, you compare real events to real events. The remaining differences, ad blockers aside, tend to be explainable in a sentence. The GA4 comparison goes through the definitional differences one by one.