Why GA4 and Google Ads show different numbers

GA4 and Google Ads rarely show the same figures, and Google itself calls that common ^[1]. Google Ads records a conversion on the day of the click and only counts Google Ads; GA4 records it on the day of the conversion and spreads the credit across all channels. A stable difference is therefore normal, a sudden jump is not.

If you report the figures or have them reviewed by a Google Analytics agency, the question to ask is not which number is right, but whether the difference is right. For that, you need to know where it comes from.

Which number do you use for what?

Each number answers a different question. Google Ads only measures conversions from Google Ads sources, after an interaction with an ad from the same account ^{[1], [2]}. GA4 also shows conversions from organic search, email or social ^[1].

In GA4 those actions are called key events; they only become a conversion when you pass them on to Google Ads ^[3]. That step is explained in GA4 key events and conversions: the difference explained.

Why do they count differently?

The best-known example comes from Google itself: someone who clicks on 19 July and buys on 20 July appears in Google Ads on 19 July and in GA4 on 20 July ^[1]. Over a month that largely evens out, per day it doesn't. Google Ads also filters invalid clicks, and can include modelled conversions where Google can't observe everything ^[1].

Both tools still offer two attribution models: data-driven and last click ^{[4], [7]}. If you read elsewhere that Google Ads uses last click by default, that is out of date: data-driven is the default for most conversion actions ^[7]. In GA4, by default only paid Google channels receive credit for the conversions that go to Google Ads, while for the reports in GA4 itself both paid and organic channels do ^[4].

Then there is consent: with basic Consent Mode, nothing goes to GA4 for those who refuse ^[8]. Google Ads can model conversions for those visitors, but for a model based on your own data, Google cites 700 ad clicks over 7 days, per country and domain grouping ^[9]. How to set this up is explained in Consent Mode v2 in practice: basic or advanced?.

How to compare fairly

1. Choose one action. Compare one conversion action in Google Ads with the matching key event, not a total of purchases and forms.
2. Choose the right column. In Google Ads, use the All conv. (by conv. time) column when you compare with a system that records on conversion date, such as GA4 ^{[2], [10]}.
3. Limit GA4 to Google Ads, via the reports under Advertising or a filter on the channel or source.
4. Take a week or a month, not a day, and afterwards wait a few days until processing in GA4 is complete ^[6].
5. Put both on the same time zone. GA4 follows your property's reporting time zone ^[11].
6. Compare both with your own source: the orders in your webshop or the requests in your CRM.

Under Advertising › Conversion management in GA4 you see all conversions shared with Google Ads. Google is also building tools there to detect settings that don't match ^[12].

When is it an error after all?

• A sudden jump in the difference, right after a release of your site, a new cookie banner or a change in Tag Manager.
• A lot of Google Ads traffic as (not set) in GA4, for example because of accounts that aren't linked or auto-tagging that is turned off ^[13].
• Sessions without a source after a change to your banner: incorrect Consent Mode settings can break sessions ^[13].
• A redirect that drops the gclid, for example after a migration to a new URL structure ^[10].
• The same action set to primary twice, via a GA4 import as well as a Google Ads tag.
• Conversion values that are too high, because actions such as a page view or add to cart are set to primary with a value ^[2].

Clicks are not sessions

The tools also differ on traffic: Google Ads counts clicks, GA4 counts sessions. Someone who clicks your ad twice within thirty minutes counts as two clicks and often as one session ^[10]. Conversely, a visitor can come back later via a bookmark and get the last known campaign as the source ^[10].

Google also names technical causes: auto-tagging that is turned off, a redirect that drops the gclid or the UTM parameters, or a landing page without measurement code ^[10]. If your accounts were not yet linked during the period you compare, that data won't appear retroactively ^[10].

Test it by adding ?gclid=test to your landing page URL and checking whether the parameter is still on the page after any redirects ^[10]. In Belgium, consent comes on top: someone who refuses still produces a click, but with basic Consent Mode no session.

If you run Performance Max, that traffic appears in GA4 under Cross-network, not under Paid Search ^[14]. What that means for your attribution is explained in What is Performance Max? How it works, the risks and when it fits.

If you want to make the link between click and conversion more robust, take a look at enhanced conversions in Google Ads.

Not sure whether the difference in your account can be explained? The free tracking audit checks whether GA4 and Google Ads are linked correctly, whether conversions count only once and what happens to visitors who refuse.