Why GA4 and your ad platform will never agree

Every account we take over arrives with the same question somewhere in the first fortnight. The ad platform says it drove 180 conversions last month. GA4 says 119. Which one is lying?

Neither. They are answering different questions, and they were built to. The gap is not a bug to be closed — it is a property of the measurement, and the useful skill is knowing how big a gap is normal for your setup and what it means when the size of it changes.

The four reasons the numbers differ

1. They credit different clicks

An ad platform is, structurally, arguing for itself. It reports a conversion when it can connect that conversion to an interaction with its own ads. GA4, by default, distributes credit across the whole path it can observe — paid, organic, direct, referral.

So a customer who clicks a paid ad on Monday, searches your brand on Thursday and buys on Friday is one conversion to the ad platform and some fraction of a conversion to GA4. Both statements are true. They are about different things.

2. They count over different windows, anchored to different days

Ad platforms credit a conversion back to the day of the click, not the day of the purchase. GA4 reports it on the day it happened. A purchase on the 3rd from a click on the 28th appears in different months in the two tools.

This alone explains most month-boundary disputes, and it is why comparing a single month in isolation is close to meaningless. Compare rolling periods instead.

3. View-through conversions

Some platforms count a conversion when the ad was served and not clicked. GA4 has no way to know an impression happened, so it never counts these. If your reporting includes view-through, you are comparing a number that contains something the other number structurally cannot.

Check whether view-through is included before you compare anything. It is often on by default.

4. Consent, blocking and modelling

A meaningful share of sessions never reach analytics at all — consent declined, tracking prevention, ad blockers, privacy browsers. Both tools respond by modelling the gap, and they model it differently, with different inputs and different thresholds.

This is the part that has grown fastest in the last few years, and it is why gaps that were stable in 2020 are not stable now.

Which number should you actually use

Use them for the jobs they are good at.

1. Ad platform data for in-platform decisions. Bids, budgets, creative rotation, which campaign to pause. The platform optimises against its own signal, so this is the number its algorithm is acting on.
2. GA4 for cross-channel comparison. Which channels contribute, how they interact, what the assisted paths look like. One tool that sees everything beats several that each see their own slice.
3. Your own backend for anything financial. Revenue, refunds, margin. Neither analytics tool knows what a customer cost you to serve or whether they returned the product six weeks later.

The mistake is not using both. The mistake is putting two numbers that answer different questions in adjacent columns of one report and asking a client to reconcile them.

When the gap is a real problem

A stable gap is fine. A changing gap is a signal, and worth investigating:

• The gap widens suddenly. Something broke — a tag firing on the wrong trigger, a consent banner change, a checkout deploy that moved the confirmation step.
• The gap narrows to nothing. Usually double counting somewhere, or a conversion action recording an intermediate step rather than the purchase.
• One tool moves and the other does not. Almost always a configuration change in the one that moved.

Set a baseline. Know that your normal is, say, a platform-to-GA4 ratio somewhere around 1.4, and then treat 2.1 as a thing to look at rather than a thing to explain away.

A practical reconciliation

When you do need to close the gap for a specific campaign, work through it in this order. Stop as soon as the difference is explained.

1. Align the date basis. Put both tools on conversion date, or accept that you are comparing click-date to conversion-date and say so.
2. Turn off view-through in the platform report.
3. Set the attribution model to last-click in both. This is a diagnostic setting, not how you should run the account.
4. Confirm both are counting the same event. purchase in GA4 and a conversion action pointing at the same page are not automatically the same thing.
5. Check the time zone on both properties. They are set independently and are wrong more often than anyone expects.
6. Compare a period long enough to absorb the lag — at least three times your typical time-to-conversion.

If a material difference survives all six, you have a genuine tracking defect and it is worth the engineering time.

What we do about it

On every account we run, the reporting names its source. A number labelled Google Ads conversions and a number labelled GA4 key events sit in different sections, with the ratio between them tracked as its own metric over time.

That last part matters more than any individual reconciliation. The absolute gap is a fact about how measurement works. The trend in the gap is the thing that tells you when something has actually broken.