GA4 key events are what GA4 used to call conversions, and the rename was more consequential than a relabelling usually is. One word was being used for two different measurements, in two tools, with different counting rules, and a great many arguments came from that alone.
Nothing about the counting changed. What changed is that the two things now have separate names, which makes a distinction visible that was always there.
This is for you if you use GA4 and Google Ads together and have ever been confused about which conversion number somebody meant.
What GA4 key events actually are
They are events you have marked as mattering, inside GA4, for analytics reporting.
GA4 collects a great deal automatically. Page views, scrolls, outbound clicks, file downloads. Marking one as a key event tells GA4 to treat it as an outcome rather than as activity, so it appears in reports built around outcomes.
That is the whole scope. It is an analytics label describing importance within analytics.
What it is not is an instruction to Google Ads. Marking something as a key event does not make it a Google Ads conversion, does not feed bidding, and does not appear in your advertising reports. Those require a separate import step.
Why the rename mattered more than it looked
Before it, both tools called their outcome measurement a conversion, and the two numbers routinely disagreed.
Somebody in a meeting would say conversions and mean the Google Ads figure. Somebody else would hear it and picture the GA4 figure. Both would then discuss a discrepancy that was partly a vocabulary problem.
The counting differences behind that are real and structural. Different attribution, different dating, different treatment of view-through activity. None of those got fixed by the rename, and none of them were ever going to be.
What the rename did was stop one word implying the two measurements were the same thing. That is a modest change with a genuinely useful effect on conversation.
I have watched meetings improve immediately once people had two words available. It is the cheapest measurement improvement in this entire cluster.
The step people skip
Marking a key event in GA4 and expecting Google Ads to act on it is the most common mistake here.
The chain is: the event fires, you mark it as a key event in GA4, you import it into Google Ads as a conversion action, and you set whether it is primary or secondary. Bidding responds only to primary conversion actions.
Miss the import step and you have an event that looks important in one tool and does nothing in the other. Your reporting shows outcomes. Your bidding has no idea they exist.
Miss the primary setting and it counts in reports without influencing optimisation, which is a subtler version of the same gap.
Both are worth checking directly rather than assuming, because nothing in either interface flags the disconnect.
How many GA4 key events you should have
Fewer than you currently do, in most accounts I open.
The purpose of marking something is to distinguish it from everything else. Mark twelve events and you have distinguished nothing, because the label no longer separates the important from the ordinary.
Two or three that map to money is usually right. An enquiry, a qualified enquiry, a sale. Everything else can be measured without being elevated.
The temptation runs the other way, because marking an event feels like taking it seriously and costs nothing. The cost arrives later, when somebody builds a report on outcomes and finds that outcomes include newsletter signups.
Which ones belong in Google Ads
A narrower set again, and this is where real money is affected.
Only import what you want the bidding to chase. If a newsletter signup and a purchase both arrive as primary conversion actions, you have told the algorithm those outcomes are comparable, and it will optimise accordingly.
That is not a hypothetical failure. It is one of the most common causes of an account that produces plenty of cheap conversions and disappointing revenue.
The test I would apply is simple. Would you spend the same amount to acquire one of these as the other? If not, they should not both be primary.
Secondary conversion actions exist precisely for the things worth watching and not worth bidding towards, and they are underused.
What did not change
Worth stating plainly, because the rename created a fear it did not warrant.
No counting method changed. Historical data was not recalculated. Attribution behaviour stayed exactly where it was. GA4 key events count precisely what GA4 conversions counted, under a different name.
So if your numbers shifted around the time of the rename, the rename is not the cause and something else deserves investigation. Attributing a genuine data change to a cosmetic relabelling is how a real fault survives another six months.
I have seen exactly that happen, twice. The rename made a convenient explanation and delayed finding what actually broke.
Why GA4 key events still matter for cross-channel questions
The rename separated the two tools, and it is worth being clear about what GA4 is still uniquely good at.
Google Ads sees the clicks it bought and nothing else. It cannot tell you how paid compares against organic, email or direct, because those never appear in it.
GA4 sees all of them, which makes it the only place a cross-channel question can be answered honestly. When somebody asks whether paid search is worth more than the newsletter, that comparison has to happen there.
The caveat is that GA4’s own view is partial too. Consent refusals remove visitors, and since June 2026 the advertising storage permission is the sole gate on advertising data reaching Google Ads from GA4, so the two tools now diverge on consent as well as on method.
So the honest position is that GA4 key events answer the broadest question available to you, from an incomplete sample. That is still considerably better than answering a cross-channel question with a single-channel tool, which is what happens when people avoid GA4 because the numbers disagree.
External evidence on how far measurement systems can drift is worth holding beside all of this. Seer Interactive found Meta reporting 87% of conversions as incremental where GA4 said 67% across $1.05 million of spend on six accounts.
What to change this week
Three steps.
List your GA4 key events and check which are imported into Google Ads. The gap between those two lists is usually larger than people expect.
Then check which imported actions are set as primary, because that is the set your bidding is chasing, and it should contain only outcomes you would genuinely pay comparable amounts for.
Then adopt the vocabulary properly in your own reports. Say key event when you mean GA4, say conversion when you mean Google Ads, and watch how much confusion disappears from your meetings.
The GA4 vs Google Ads discrepancy explains why the two numbers differ structurally, what conversion numbers actually measure covers the counting underneath, conversion event deduplication is what happens when too many things count, and sending outcomes back is how you make the imported action mean something. Getting the primary actions right is where my Google Ads work starts on most accounts. To find out what your bidding is currently chasing, book a teardown.