GA4 Conversion Tracking - Setting Up the Events That Actually Map to Revenue
GA4 doesn't track conversions out of the box. It tracks events — clicks, scrolls, page views — and waits for you to tell it which of those actually mean money. Skip that step and you get a dashboard full of activity that looks like data and tells you nothing.
So here's what this post is really about. Not "what is GA4" in the abstract — you can get that from Google. It's about the gap between having GA4 installed and trusting what it says, because that gap is where most teams quietly live for years without realising it.
First, the Thirty-Second What-Is
GA4 — Google Analytics 4 — is the version that replaced Universal Analytics in 2023. The big change: it's built around events, not sessions and pageviews. Everything a user does is an event. A pageview is an event. A click is an event. A purchase is an event. There's no special category called "goal" anymore the way the old version had.
That sounds like a technicality. It isn't. It's the whole reason conversion tracking trips people up now. In the old world you set up goals in the interface and moved on. In GA4 you have to decide what counts as a conversion, make sure that thing fires as a clean event, and then flag it. Three steps where there used to be one. And every one of them is a place to get it wrong.
Why "Installed" and "Working" Are Different Things
I'll be blunt: most GA4 setups I've audited were installed correctly and tracking nothing useful.
The tag was on every page. Real-time showed traffic. Looked healthy. But ask the obvious question — how many people actually signed up last week, and from where — and the numbers either didn't exist or didn't match the database. The events that mattered were never configured, so GA4 was dutifully counting pageviews while the business question went unanswered.
This is the seam where things fall apart, and it's a familiar one. Marketing knows what should count as a conversion — a trial start, a qualified form fill, a checkout. Engineering knows how to make that fire reliably in the code. But the two conversations happen in different rooms, weeks apart, and the event that ends up in GA4 is some watered-down compromise nobody fully owns. A button-click proxy instead of a real confirmed signup. Close enough, except it isn't.
Crucial Insight
A conversion event is only as trustworthy as the moment it fires. If it fires on a button click instead of a confirmed server response, you're counting intentions, not outcomes — and you'll happily optimise your ad spend toward people who clicked and then bounced off a broken form.
The Events Worth Tracking (and the Ones That Just Look Busy)
Not everything deserves to be a conversion. The instinct to track everything is how you end up with forty events and no idea which three matter.
- Track the money moments. Trial start. Demo booked. Qualified lead submitted. Purchase. Subscription. The handful of actions that actually move revenue. These are your conversions. Mark them as such and protect that list.
- Track the steps right before money. Add-to-cart, pricing-page view, checkout-started. Not conversions themselves, but they tell you where people fall out of the funnel. Worth having, not worth flagging as conversions.
- Be suspicious of engagement vanity. Scroll depth, time on page, video plays. Occasionally useful, mostly a comfort blanket. They go up and to the right and tell you nothing about whether the business is growing.
- Kill the duplicates. GA4's enhanced measurement auto-tracks some things, and people often re-track them manually too. Now your numbers are doubled and you don't know it. Pick one source per event.
The test is simple. For each event ask: if this number doubled next month, would I do anything differently? If the honest answer is no, it's not a conversion. It's noise wearing a conversion's clothes.
A 5-Step Setup That Won't Lie to You
Do these in order. The order matters more than it looks.
- Define the conversion in plain words first. Before touching GA4, write the sentence: "A conversion is when a user [does exactly what]." If marketing and engineering can't agree on that sentence, no tool will save you. This is the step everyone skips and everyone regrets.
- Fire the event at the right moment. Make the event trigger on the confirmed outcome — the thank-you page, the server's success response — not the click that hopefully leads there. This is an engineering call, and it's the difference between counting outcomes and counting wishes.
- Send the event cleanly. Use a consistent name, pass useful parameters (value, currency, plan tier), and route it through one path — GA4 directly, or a tag manager, but not both firing the same thing.
- Mark it as a key event. In GA4, flag the event as a key event (Google's current name for what used to be "conversions"). Only then does it show up in conversion reports. Plenty of correctly-fired events never get this flag and stay invisible.
- Verify against reality. Trigger the event yourself, watch it land in DebugView, then cross-check the count against your actual database a day later. If GA4 says 50 signups and the database says 38, stop and find the gap before you trust a single report.
That last step is the one nobody does and everybody needs. Verifying against the source of truth — your own database — is the only way to know your tracking isn't quietly lying to you.
Match the Symptom to the Fix
When the numbers feel off, they usually are. Here's the usual list.
| Symptom | Likely Cause | First Fix |
|---|---|---|
| GA4 count higher than the database | Event fires on click, not confirmation | Move the trigger to the confirmed outcome |
| Conversions show as zero | Event fires but isn't marked a key event | Flag it as a key event in GA4 |
| Numbers roughly doubled | Same event tracked twice | Pick one source; turn off the duplicate |
| Can't tell which channel converts | No source/medium on the event | Check attribution settings; confirm UTMs land |
| Revenue missing from reports | No value parameter on the event | Pass value and currency with the purchase event |
| Everything tracked, nothing clear | Too many events, no priority | Cut to the handful that map to money |
Give any change a couple of days before you judge it. GA4 processing isn't instant, and a fix you made Monday morning often only reads clean by Wednesday.
Rule of Thumb
If you can't reconcile a GA4 conversion number against your own database to within a small margin, treat the GA4 number as a rumour, not a fact. Reconciled tracking is the only tracking worth optimising spend against.
Why This Is Worth the Bother
Because every downstream decision rides on it. Your ad platforms optimise toward the conversions you feed them. Your reports tell the founder where growth is coming from. Your team decides what to build next based on what's "working." If the conversion data is wrong, all of that is wrong — confidently, expensively wrong, in a way that's hard to catch because the dashboard looks so convincing.
We had a client pouring budget into a channel GA4 swore was their best performer. It wasn't. The channel's landing pages fired the signup event on page load — so every visit counted as a conversion, and the worst-performing traffic looked like the best. They'd been scaling the loser for months. Once the event moved to the actual confirmed signup, the real picture showed up, the budget moved, and cost per real signup dropped by about a third in the first month. No new traffic. Just numbers that finally told the truth.
What To Actually Do With This
GA4 conversion tracking isn't a setup task you finish once. It's a measurement layer you have to get honest and keep honest.
- Define each conversion in a plain sentence before you build anything — and get marketing and engineering to agree on it.
- Fire events on confirmed outcomes, not hopeful clicks, or you'll optimise toward the wrong people.
- Track the handful of money moments, flag them as key events, and resist tracking everything else.
- Reconcile against your database, because a number you can't verify is a number you can't trust.
- Then, and only then, point your ad spend at it — because now it's pointing at something real.
Most teams don't need a new analytics tool. They need the one they've already got to stop lying to them. That's usually a day of work with a marketer and an engineer in the same room — and it's the cheapest clarity you'll buy all quarter.
