What GA4 tells you that Google Ads does not, and why they never agree

Two reports, same month, same campaign, different conversion numbers. This produces more unnecessary suspicion between businesses and the people running their advertising than almost anything else in this trade, and the cause is usually not misconduct or a broken tag. It is that Google Ads and Google Analytics 4, its analytics product, are answering different questions.

Knowing which one to believe, and when, ends the argument permanently.

Why do the numbers differ?

Three reasons, and all three are by design.

They credit differently. Google Ads reports a conversion against the campaign whose click it saw, and by default it credits the click even if the conversion happened days later. GA4 attributes across the whole journey it can observe, which frequently includes touches Google Ads never saw and does not care about.

They date differently. Google Ads dates a conversion to the day of the click. GA4 dates it to the day it happened. In any account where people take a week to decide, those two calendars will disagree every month, and the disagreement is largest exactly when business is good.

They count differently. Google Ads counts conversions, including more than one per person when you have told it to. GA4 thinks in sessions and users. A single customer filling in two forms is two conversions and one user, and both statements are true.

• Is this campaign worth its spend? Use Google Ads · use this one

  It knows what it paid and what it got. It is the right place to judge a campaign against its own cost.

• What did people actually do on the site? Use GA4 · use this one

  Pages, paths, drop-off, devices, and what happened after the click.

• Which channels contributed to a sale? Use GA4 · use this one

  It sees the touches the ad platform cannot, including the ones you did not pay for.

• Reconciling the two to the same total · wrong tool for it

  They will not match, and time spent forcing them to is time not spent on the account.

• Judging a campaign by GA4's number alone · wrong tool for it

  GA4 spreads credit across the journey, so a campaign that started the journey looks weaker than it was.

Which one should the business trust?

Neither, on its own. Both, against a third thing.

The third thing is your own record: your inbox, your CRM, your booking system, the count of people who actually became customers. That is the only figure with no attribution model in it, and it is the one that decides whether the marketing worked.

The platforms are for allocating credit and steering spend. Your own records are for judging outcomes. Businesses get into trouble when they promote a platform number into an outcome, which is the failure I have written about in Google Ads shows conversions but you are not getting leads.

What should you actually reconcile?

Direction, not totals.

If Google Ads reports conversions rising forty percent and your inbox is flat, something is wrong and it is worth a morning. If Google Ads reports thirty-one and GA4 reports twenty-six, nothing is wrong and chasing the five will teach you nothing.

The rough tolerance I work to is that a gap under about twenty percent between the two platforms is normal in most accounts. A gap that is large, or one that appears suddenly in a month when nothing changed, is worth investigating, and the first place to look is whether a website change removed a tag. That story is in your website relaunch broke your leads.

What breaks the relationship between them entirely?

A conversion action that fires on the wrong thing, most often.

If Google Ads is counting page loads and GA4 is counting form submissions, the two are not measuring the same event and no reconciliation is possible. This is common, it is invisible from a report, and it takes ten minutes to check. The plain version is what a conversion action is, and why yours might be lying.

What is the practical routine?

Once a month, put three numbers next to each other: what the ad platform says it produced, what analytics says happened, and what your own records say you received. Do not try to make them equal. Look at whether they move together.

When all three rise, the marketing is working and you can stop reading reports. When the platform rises and your records do not, you have a measurement problem rather than a performance problem, and fixing the campaigns will not help.

That is the whole discipline, and it takes about fifteen minutes. It also happens to be the fastest way to know whether the person running your account is telling you the truth, which is the subject of how to tell if your ads agency is doing a good job.