Double counting is the most damaging tracking defect there is, and the reason is counterintuitive: it does not look like a problem. It looks like good news.
Your conversion volume doubles. Your cost per conversion halves. The campaign that was borderline is suddenly your best performer, and you put more budget behind it. Meanwhile your sales team is having the same number of conversations they had last quarter and cannot work out why everyone upstairs seems so pleased.
Worse, Smart Bidding believes it. It sees twice the conversions from a particular audience, keyword, or time of day, and it doubles down on whatever produced the phantom. You are not just reporting wrong, you are buying wrong.
Here are the five ways it happens.
1. Two conversion sources for one event
The most common version by far.
You have a Google Ads conversion tag firing on your thank-you page. You also have a GA4 key event for the same form submission, imported into Google Ads as a conversion. Both are set to "Include in conversions."
One lead. Two conversions. No warning anywhere in the interface.
This usually arrives when someone new takes over the account. The new person sets up tracking the way they prefer and does not remove what was already there, because removing another person's tracking feels risky.
How to prove it: Goals, then Conversions, then Summary. Add the "Conversion source" column. Look for two rows describing the same event, one sourced from "Website" and one from "Google Analytics 4." Then check "Include in conversions" on both.
The fix: pick one source. Keep the other as Secondary so it still reports without inflating totals or steering bids. Native tags are generally the better primary for Google Ads.
2. One tag, two triggers
In Google Tag Manager, a conversion tag can end up attached to more than one trigger that fires on the same interaction.
A form submission trigger and a page view trigger on the thank-you URL will both fire for the same lead. So will a click trigger on the submit button paired with a form submission trigger.
How to prove it: open GTM Preview mode, submit a test lead, and watch the tag list. If your conversion tag appears twice in the summary, you have it.
3. The thank-you page is reachable more than once
If your conversion fires on a thank-you page view, and your conversion action is set to count "every" conversion, then every arrival counts.
People refresh. They hit back and then forward. They bookmark the page. Some email clients prefetch links. Each one is another conversion from the same lead.
The fix: set lead generation conversion actions to count one per click. That alone neutralizes this whole category.
4. Two containers, or a container plus a hardcoded tag
Sites that have been through a redesign or an agency change often carry more tracking than anyone realizes.
A legacy GTM container left in the template plus the new one. A hardcoded gtag snippet in the head plus the same Google tag deployed through GTM. Each copy fires its own conversion.
How to prove it: view the page source and search for googletagmanager.com. Count the distinct container and tag IDs. Anything you cannot account for deserves investigation before it gets removed.
5. Multi-step forms firing at each step
Multi-step and conditional forms often fire their submission event on each step transition rather than only on final submit. A four-step form can produce four conversions from a lead who never finished.
How to prove it: start the form, advance one step, abandon it, and check whether a conversion registered.
The symptoms, if you want to look before you dig
- Conversion rate above what is plausible for your industry, for example 30% on a cold traffic B2B landing page
- Google Ads conversions sitting at almost exactly twice your CRM record count
- Reported cost per lead that looks excellent while pipeline stays flat
- A step change in conversion volume on a date with no campaign change behind it, which points at a site release
- Conversion rate that varies wildly by device, which often means a tag fires differently on mobile
That last pattern is worth taking seriously. It is rarely a real behavioral difference.
Fix it in this order
- Check for duplicate conversion actions with different sources. Demote one.
- Set every lead generation action to count one per click.
- Run GTM Preview and submit a test lead. Count the fires.
- Audit the page source for extra containers and stray hardcoded tags.
- Test multi-step forms by abandoning them partway.
Then let it run for two weeks before drawing conclusions, because the historical data in your reports does not retroactively correct itself. Your before and after will not be comparable, and you should expect your conversion numbers to fall. That drop is the point. It is the number getting honest.
Warn people first
Conversion volume dropping by half is alarming if nobody was told it was coming. Tell your leadership before you make the change, not after, and frame it as what it is: the reported number is being corrected to match reality, and the real performance has not changed at all.
This is worth saying out loud because the fear of that conversation is the actual reason duplicate tracking survives in so many accounts.
Want a second pair of eyes?
The Conversion Tracking Audit checks every conversion action and its source, your tag manager container, how many times your tags fire on a real submission, and whether your numbers reconcile against your CRM. You get a written list of what is duplicated and what to fix first.
$599, fixed fee, no retainer and no call required first.
