Conversion tracking before you run ads: are your conversions counted correctly?
Measurement, reporting and GA4
10 min read
Published: · Updated:
If the conversion count is wrong, every decision about the ad budget rests on a wrong number. Conversion tracking is the recording of a valuable action on your site (a form sent, a call, a purchase) once, and to the right source. Setting it up is often short work. The real effort comes after, in proving the number is right. Ten checks for that are below.
How to open the measurement account is walked through in GA4 setup step by step, and how to work out what a new customer costs is in how to calculate customer acquisition cost. This post starts from a finished set-up and asks one thing: do the numbers match what happened?
What does counting conversions correctly mean?
Accuracy depends on two questions. Is the thing being counted a valuable action, and is each action counted once when it happens? Undercounting and overcounting damage decisions in the same way: a number twice as high makes the business look healthier than it is, and a number that never moves makes it look worse. A third case turns up in reports. The same action can appear with different counts in Analytics and in Ads, and that is not always an error. The reasons are in the seventh section.
Google Ads Help’s page about conversion measurement describes a conversion as an action you define as valuable, such as a purchase, a sign-up or a phone call. The definition is yours, so the first decision comes before any counting.
Which actions should count as conversions?
Picture a small catering company (fictional). Its site has a quote request form, a click-to-call button, a sample menu page and pages for each event type.
- A sent quote request: a conversion. It is exactly what the company wants.
- A tap on the call button: often counted, but a tap does not mean a conversation took place. If it is counted, it should be counted knowingly.
- Opening the sample menu: an in-between step. Counting it inflates the total and makes it impossible to compare with real quote requests.
- A page view: not a conversion.
A working rule: few actions, each close to real business. What GA4 calls a key event, Ads calls a conversion; the difference is defined in the key event section of the GA4 setup post. A one-line note per conversion saves time later: its name, why it counts, which event fires it, who manages it.
The counting flow: from click to report
The path of one quote request, from ad to report, has six steps:
- The visitor clicks the ad and lands on the site. If auto-tagging is on in Ads, the click travels to your site as a parameter added to the page address; you need this if you import GA4 data into Ads.
- They fill in the quote form, send it and see the thank-you page.
- The site sends a “form submitted” event. One of the recommended event names is
generate_lead. - GA4 counts that event if it is marked as a key event.
- Ads can take the conversion in two ways: with its own tag, or by importing it from GA4. Imported data can arrive up to 24 hours late.
- Someone checks whether the same action was counted through both routes.
Step six is easy to skip. The rest of this post is about it.
Two ways to count: a tag, or an import from GA4?
Google’s page on creating conversions from Analytics events compares the two methods and says they can be combined in any way you like. Combining them also opens the door to double counting.
| Criterion | Ads’ own tag | Import from GA4 |
|---|---|---|
| What it counts | Only conversions that come from Ads | Tracks all traffic on the site; may include conversions from sources other than Ads |
| Set-up | A tag goes on the site or into the tag manager | Accounts linked, auto-tagging on, the event marked as a key event in GA4 |
| Double-count risk | Present if the same action is also imported from GA4 | Present if the same action is also counted by the Ads tag |
| Data delay | The page names no separate delay | Up to 24 hours |
| Who edits the settings | The Ads account | GA4 is treated as the main place; some settings cannot be edited in Ads |
Conversions created from the GA4 side arrive in Ads as “secondary”. Google says it does this so the same event is not counted twice for bidding. So there is a built-in guard against double counting, but it covers only that much.
Our own view is that a sound start is one route per action. If you want a second route, switch both on and run them through the test in this post.
How can you tell a conversion is counted twice?
Think of two people with hand clickers at the door of an event, both counting the same guest list. The total says more people came in than the guest list holds. Double counting works the same way: the number grows, and real life stays the same.
The signs:
- The conversion count in Ads is clearly higher than the key event count in GA4.
- You sent the form once and the panel shows two rows.
- The count rises when you refresh the thank-you page or come back to it with the back button.
- You opened one page and the report shows two page views. That means two measurement codes sit on the site.
The causes are usually ordinary. A tag fires both when the page loads and when the button is clicked. The thank-you page can be refreshed and the trigger fires on every load. The same code sits in the tag manager and in a plugin. Both routes treat the same action as primary. The cure always passes through one place: the action should have a single trigger.
One distinction: a person truly sending two forms is not a double count. The fault is one submission being written twice.
A ten-item counting test before you run ads
Do the test like a real visitor, on a phone and on a computer. Write down the date, the time and what you did; tomorrow you will match it to a line in the report.
- Fill in the test form and send it once.
- Open DebugView in GA4 and check that the event appears, and appears once.
- Are the event name and parameters what you expect?
- Is the event marked as a key event?
- Is the conversion action defined in Ads and tied to the right event? If you import, wait up to 24 hours and confirm the test conversion shows in the account.
- Did the same test show up once in both tools?
- Refresh the thank-you page, then return to it with the back button. Did the count rise?
- Send the form with a required field empty. The error message appears, but was the event sent anyway?
- Retry with “reject” on the cookie banner. Does the result fit the set-up you built?
- Try the phone, email and messaging buttons from a phone. Is each one an action you decided to count?
Google describes DebugView as a way to troubleshoot while loading tags. Data collected in debug mode can affect report data, so a developer traffic filter is recommended; details are on the DebugView page.
Our own rule: if any of 1, 6, 7 or 8 fails, the ads do not start, because those four distort the count directly. The rest are noted and fixed in the first week. Measurement set-up and the ad account meet in our measurement, reporting and GA4 and performance marketing and advertising services.
What to check when the numbers do not match
A gap between Analytics and Ads sometimes points to two tools working differently, not to a fault. Google’s same page lists the causes:
| Cause | What happens |
|---|---|
| Date | Ads records a conversion on the day of the click, Analytics on the day the action happened. Someone who clicks on Monday and books on Wednesday lands on Monday in Ads and on Wednesday in Analytics. |
| Counting method | In Ads you can count every conversion or only unique ones. If a person sends two forms after the ad, Analytics shows two and an Ads action set to unique shows one. |
| Invalid clicks | Ads does not record suspicious clicks; some conversions visible in Analytics can be filtered out when they are imported. |
| Time window | An Ads cookie expires 90 days after the click and the conversion window runs from 1 to 90 days. Analytics cookies last up to 2 years. Someone who decides late is counted in Analytics and may not be counted in Ads. |
| Delay | Data imported from Analytics arrives up to 24 hours late. |
None of these is a fault. A fault is a gap that no cause in the table explains; then go back to the double-count signs in the fifth section. Another source for this post is Google’s page on making sure conversions are tracked correctly.
How does cookie consent affect the count?
If a visitor refuses analytics cookies, what GA4 measures depends on the consent set-up you built, so part of the actions missing from a report is expected. Google says reports in Ads can include modelled conversions for cases where it cannot observe every conversion. When you refuse cookies during testing, the event may also not show in DebugView: Google writes that when consent mode is applied and the visitor has not given consent, events do not appear in debug mode.
How to build the consent set-up is in the cookie consent section of the GA4 post; for the legal answer, ask your legal adviser.
Frequently asked questions
How do you set up conversion tracking in Google Ads?
You first define a conversion action in Ads, then either place a tag on the site or import a GA4 event. Because the interface changes often, it is safer to follow Google’s page on setting up your conversions. After set-up, the accuracy of the count still needs its own test.
How do you import Google Analytics conversions into Google Ads?
Link the accounts, turn on auto-tagging in Ads and mark the event as a key event in GA4. Then pick the event from conversion management in GA4 or from the create-a-conversion screen in Ads. Google says it can take 24 to 48 hours for the options to appear in Ads.
How can I be sure conversions are tracked correctly?
Perform a real action yourself and confirm it shows up once in both GA4 and Ads. The check should not be a one-off; repeat it when the site changes or a form or tag is added. Google also stresses that correct set-up is a basic part of measuring results.
Why do Google Analytics and Google Ads show different numbers?
The two tools date an action differently, use different counting methods and different cookie lifetimes. Ads filters suspicious clicks, and data from Analytics can arrive late. With a correct set-up, small gaps are expected; a large gap that nothing explains points to double counting or a missing tag.
Related service
07
Measurement, reporting and GA4
We define what you should measure, plan the tracking and turn data into reports you can decide from.
Let us check your measurement before the ads go live
Let us look at where to start together.
Quick message on WhatsApp
Pick a ready-made message; it opens in WhatsApp.
Sending a message is up to you; this panel doesn't ask for your name or phone number.
Services
- Brand strategy and positioning
- Brand identity and design
- Websites and search visibility (SEO)
- Content, social media and written communications
- Performance marketing and advertising
- Email and customer relationships
- Measurement, reporting and GA4
- Campaign planning and print-ready design
- Consulting and workshops