GA4 Key Events: The Mistake That Made Our Conversions Look 49 Times Better
Table of contents
Short answer: A GA4 key event is any event you mark as important to your business. Small businesses should mark only actions that prove intent, such as a confirmed enquiry, a phone click, a booking or a purchase. Never mark a page visit or an event that can fire while someone is merely browsing, because your conversion numbers will look strong while telling you nothing.
Open our GA4 report for the last 90 days and you will see 2,845 key events. For a company our size that is a superb number. You could put it in a sales deck.
Only 58 of them are real.
The other 2,787 come from an event called online_business_directory_visit. Someone, at some point, decided that a visit to our directory page should count as a conversion. It sounded reasonable at the time, because that page matters to us. So for months, every person who loaded the page became a "conversion". That is 1,155 users. The event that means something, thank_you, fired 58 times from 22 people.
The first time we noticed this kind of thing, in 2026, the gap was 1,101 reported against 22 genuine completions. It was still there when we pulled the numbers for this article, which should tell you how easily this survives inside a business. Nobody lied. A setting was made once, and then everyone trusted the report.
What a key event is, in plain terms
GA4 records events: page views, scrolls, clicks, form starts. Most of it is background noise, in the good sense. A key event is simply any of those events you have flagged as important. Google renamed "conversions" to "key events" in GA4, and it kept the word "conversion" for Google Ads.
So the flag means whatever you decide it means, and GA4 will not question you. Mark "page_view" as a key event and it will proudly report a 100% conversion rate. Think of a fire alarm wired to the kitchen light. The alarm works perfectly. It just rings all day, and after a week nobody hears it.
Why this mistake costs real money
A wrong key event does more damage than a missing one, because the number looks healthy. Three decisions go wrong downstream.
- Channel decisions. Our GA4 credits Direct with 1,534 conversions and Organic Search with 885. Almost all of those are page visits, so the table cannot tell us which channel brings enquiries.
- Ad spend. Key events can feed Google Ads conversions. If a smart bidding campaign learns from a page visit, it will find you more page visits, and it will be very good at it.
- Engagement figures. A session with a key event counts as engaged, whatever its length. A loose key event inflates your engagement rate too.
Which events deserve the flag
Apply one test: if this event fired 100 times, would you expect more revenue? If the honest answer is "maybe", it is not a key event. It may still be worth watching as an ordinary event.
| Business | Good key events | Keep as ordinary events |
|---|---|---|
| Local service company | Enquiry thank-you page, phone number click, booking confirmed | Contact page view, form started |
| Online store | Purchase | Add to cart, begin checkout, product view |
| Consultant or agency | Proposal request, call booked | Case study read, PDF download |
| Directory or marketplace | Listing submitted, listing claimed | Directory page visit, search used |
Most small businesses need between one and four key events. If you have ten, you have probably not chosen.
The trap inside form_submit
GA4's enhanced measurement can record form_start and form_submit with no code. It looks like a gift. Our site logged 2,290 form_submit events from 1,010 users in the same 90 days. Set that against 58 thank-you events and you can see the issue.
Our site has search boxes, filters and several other forms, and we cannot say that all of those submissions were leads. An automatic event knows that a form was sent. It does not know which form, or whether the message went through. Use a thank-you page or a custom event that fires only after success. If you can only do one thing after reading this, do that.
A 15-minute audit you can run today
- List them. In GA4, go to Admin, then Data display, then Key events (the menu wording can vary). Write down each one.
- Match them to reality. Open your enquiry inbox or orders for last month and count them. Compare that with the GA4 figure for the same dates. Expect a small gap, since GA4 misses people who decline cookies.
- Look at the ratio. If GA4 shows ten times your real leads, a key event is too loose. If it shows a tenth of them, the tag is failing.
- Test it yourself. Submit your own form with DebugView open and watch exactly which events fire.
- Unmark what fails. Removing the flag does not delete the event. You keep the data and lose only the false label.
- Write down the date. Your conversion chart will fall on that day. Tell your team first, or someone will report it as a crisis.
Step 6 is real. When you clean up key events, the graph falls off a cliff, and it looks like the worst month in the company's history. That is the moment your data becomes honest. Note it with a Search Console annotation too, so anyone comparing the two tools understands the change.
Three settings worth checking while you are in there
- Data retention. Standard properties can keep user-level data for 2 or 14 months. It affects explorations and funnels, and it is worth setting to 14 months. Aggregated reports are unaffected.
- Timing. Attribution credit for key events can change for up to 12 days, so yesterday's channel split is not final.
- Consent. If you run a cookie banner, GA4 will count fewer people than your inbox does. Modelling helps only above certain traffic levels. We explain this in why GA4 and Search Console disagree.
Once your events are clean, conversion rate work finally has a reliable number to improve.
The day your conversions fall by 98% is a good day
We are publishing our own mistake because we suspect it sits in a lot of accounts. Fixing ours will take our reported conversions from about 2,845 down to 58. It will look terrible on a slide, and it will be the first number in that report we can act on.
ONE Dashboard lists your key events by name beside your traffic and Search Console data, using read-only access, so a page view dressed up as a conversion is hard to miss. It cannot tell you which events matter to your business. That part remains your call.
Check what your conversions are made of in five minutes: connect GA4 to ONE Dashboard free, and look at the first name on the list.
Frequently asked questions
What is a key event in GA4?
A key event is an event you mark as important to your business, such as a completed enquiry or a purchase. Any collected event can become a key event. GA4 then shows key event counts in reports, and you can use them to create Google Ads conversions.
What is the difference between an event and a key event in GA4?
An event is anything GA4 records, such as a page view or scroll. A key event is one you have chosen to treat as a business result. The event itself does not change. Marking it simply tells GA4 and your team that this action matters.
Which key events should a small business track?
Track actions that show real intent to buy: a form that reached its thank-you page, a click on your phone number, a booking or a purchase. Keep the list short. If an event could fire for someone who is just browsing, it should not be a key event.
Why is the form_submit event not reliable as a key event?
Enhanced measurement can fire form_submit on forms that have nothing to do with a lead, such as search or newsletter forms, and it may fire when a submission fails validation. Test it. A dedicated event that fires only on your thank-you page is usually safer.
Can I change a key event later without losing data?
Yes. You can stop marking an event as key and mark another at any time, and the underlying event data stays. Reports will show the new setting going forward, so note the date of the change. Key event credit can keep shifting for up to 12 days after recording.
