A GA4 vs Google Ads discrepancy is the most reported measurement problem I get asked about, and in most accounts it is not a problem at all. The two tools count different things, on different dates, using different rules, and they were never designed to agree.
Three structural differences explain nearly all of it. None of them are settings you have configured wrongly.
This is for you if somebody has asked why your two Google reports disagree and you would like a better answer than a shrug.
Why a GA4 vs Google Ads discrepancy is expected
The two products answer different questions, and the answers were never meant to reconcile.
Google Ads exists to tell you what your advertising bought. It credits conversions to the click that caused them, using its own attribution rules, including activity it can see across devices when somebody is signed in.
GA4 exists to tell you what happened on your website. It credits the session that converted, using its own model, and it treats paid traffic as one source among several.
Ask a different question, get a different number. Neither tool is reporting incorrectly, and no amount of configuration makes two different questions produce one answer.
The first cause of the GA4 vs Google Ads discrepancy: dating
This is the largest single contributor in any account with a sales cycle.
Google Ads books a conversion back to the date of the click. Somebody clicks on the 28th, converts on the 3rd, and Google Ads records that conversion in the previous month.
GA4 books it on the day it happened. Same person, same conversion, different month.
In a monthly report those two allocations diverge substantially, especially near period boundaries and especially where lag is long. Nothing is missing from either tool. The same event has been filed in two different places.
That alone can produce a double-digit percentage difference in a month, and it resolves itself if you look at a long enough window.
The second cause: what counts as a conversion from paid
View-through activity divides the two tools completely.
Google Ads can count somebody who saw an advert, did not click, and converted later within a window. GA4 does not attribute that to paid at all, because from its point of view no paid click ever arrived.
So the same person appears as a paid conversion in one report and a direct or organic conversion in the other. Both are describing what they saw honestly.
This gets larger with display and video activity in the mix, and it is why a GA4 vs Google Ads discrepancy is usually widest on accounts running awareness campaigns alongside search.
The third cause of the GA4 vs Google Ads discrepancy: model and window
The two tools apply different rules to the credit itself.
Google Ads uses its own model across its own interactions. GA4 uses its model across every channel it can see, which means a conversion Google Ads claims entirely may be shared with organic or email inside GA4.
Windows differ too. A conversion inside Google Ads’ window but outside GA4’s session logic gets counted once and not twice.
None of this is fixable in the sense people usually mean. You can align what is alignable, such as conversion definitions and the periods you compare, and a structural residue will remain because the tools genuinely do different jobs.
Why stability matters more than size
There is no published benchmark for how large the gap should be, and chasing a number would be the wrong goal anyway.
What you want is a gap that holds steady. If Google Ads reports 20% more conversions than GA4 every month, that is a known quantity you can work with indefinitely.
A gap that moves is the actual signal. When a steady 15% becomes 45%, something changed: a tag, a consent category, an integration, a site release. That movement is a genuinely good alarm and it only works if somebody wrote down the baseline when things were fine.
So the useful exercise is not reconciliation. It is measuring your normal difference once, then noticing when it stops being normal.
For a sense of how far two honest systems can sit apart, Seer Interactive tested Meta’s incremental attribution setting across $1.05 million of ad spend on six accounts in April 2025. Meta reported 87% of conversions as incremental where GA4 put the same accounts at 67%.
That is Meta rather than Google Ads, and the principle transfers directly. Twenty percentage points, two systems, same events, both working as designed. Six accounts is a small sample and the study says so.
If a gap of that size can exist between correctly configured tools, then a difference between your own two reports is not evidence of a fault. It is evidence that you are running two tools.
What consent changed in 2026
One dated development belongs here because it widened this gap for a lot of accounts.
From 15 June 2026, the advertising storage permission became the sole gate on advertising data reaching Google Ads from GA4. Before that, other signals could carry part of that load.
The effect is that a visitor who accepts analytics but declines advertising now appears in GA4 and not in Google Ads. Plenty of consent banners present those as separate choices, so this is a common pattern rather than an edge case.
This is reported consistently by several consent platform companies and I could not find Google’s own announcement page for it, so attribute it to the platforms rather than to Google directly.
If your GA4 vs Google Ads discrepancy widened around the middle of 2026 and nobody changed anything, this is the first place to look.
How to use the two tools without arguing
Assign each one a job and stop asking it to do the other.
Google Ads answers questions about Google Ads. Which campaign to fund, which keyword is working, whether last week beat the week before. It sees the clicks it bought, in full, which no other tool does.
GA4 answers cross-channel questions. How paid compares with organic and email, what people did on the site, where the journey actually started.
Then pick one system, usually your CRM or order system, as the record of what the business earned. Neither Google tool is that, and treating either as the revenue figure is what turns a technical curiosity into a credibility problem.
What to change this week
Three steps.
Measure your current gap over a full month, on matched date ranges, and write the percentage down. That baseline is worth more than any benchmark you could look up.
Then check whether your consent banner splits analytics and advertising into separate choices, because that split now moves this number directly.
Then agree which tool answers which question, in writing, so the same discussion does not recur every month with different people quoting different figures.
What conversion numbers actually measure covers the counting underneath, platform reported conversions explains why each platform claims what it claims, Google Consent Mode covers the consent mechanism behind the 2026 change, and modelled conversions explains the estimated portion of what you are comparing. On Meta the same structural mismatch exists in a different form, which my Meta Ads work accounts for. To have your own gap measured and explained, book a teardown.