Google Ads conversions not tracking: how to check
When conversions are missing, real customers are absent from your advertising reports. Campaigns that are producing business look like campaigns that are producing nothing, and the usual response to a campaign that looks unprofitable is to switch it off.
Six things that cause it, and how much each one shows from outside
These are possibilities, not a diagnosis of your account. The right-hand column is the honest part: it says how much of each cause an external check can actually establish before anybody looks inside the advertising account.
No conversion action is configured for the thing you care about
The account measures something, but not the enquiry, call or purchase you actually judge performance by. Common where the account was set up quickly and never revisited.
Visible without account access Partly. An external check can see whether any conversion request is sent when an enquiry completes.
The tag is missing from the page where the action completes
Measurement is installed sitewide but the confirmation step, thank-you page or booking step was added later, or lives on a subdomain or a third-party booking system.
Visible without account access Often. The check follows the visible enquiry route and records where measurement stops.
The form submits without triggering anything measurable
Some form plugins and page builders submit without a page change or an event a measurement system can attach to, so the submission happens and nothing is recorded.
Visible without account access Often. The check observes what the page emits when a form is submitted.
Phone enquiries are not measured
The website encourages calls, and calls are the main enquiry route, but no call measurement exists. Campaigns producing phone customers then look like campaigns producing nothing.
Visible without account access Often. The check can see whether call measurement is present, though not whether calls are imported from elsewhere.
The ad click is lost before the enquiry happens
The click arrives, but the information identifying it does not survive the journey to the enquiry: a redirect strips it, a subdomain does not carry it, or a consent choice prevents it being retained.
Visible without account access Often. Attribution persistence across navigation is one of the things a public check can test directly.
Something changed on the website
A redesign, a new form plugin, a consent banner change, a domain migration, or a developer removing a script they did not recognise. The advertising account is untouched and the measurement still stops.
Visible without account access Sometimes. The check sees the current state, not what changed or when.
Why is Google Ads showing zero conversions?
Most often because the signal never reaches Google at all: no conversion action configured for the thing you care about, or a tag absent from the page where the enquiry actually completes.
A zero is at least honest about being wrong, which makes it the easier version of this problem to deal with. Somebody notices within days and it gets investigated.
The harder version is a number that looks plausible and is incomplete. A conversion count that has always been roughly thirty a month gives nobody a reason to ask what the thirty things are, or which enquiry routes are missing from it.
Why are some conversions recorded and others missed?
Usually because different enquiry routes are measured to different standards. The website form is measured and the phone call is not. One landing page carries the tag and a page built later does not.
This pattern is more damaging than a total failure, because the number is not obviously broken. It is quietly incomplete, and it systematically favours whichever enquiry route happens to be measured properly.
The consequence compounds. Automated bidding learns from the conversions it can see, so it optimises toward the measured route and away from the unmeasured one, regardless of which produces better customers.
What this cannot tell you Establishing what share of your enquiries arrive through each route needs your own records. An external check can identify which routes appear unmeasured; only you know how much business each one represents.
Can conversion tracking stop working without anyone touching the account?
Yes, and this is the reason a setup verified once is not verified permanently. The advertising account can sit untouched for a year while the website underneath it changes repeatedly.
A website redesign, a new form plugin, a consent banner change, a domain migration or a developer removing an unfamiliar script will each do it. None of those involve logging into Google Ads.
This is also why the date matters when somebody tells you the tracking was checked. Checked before the redesign is not the same as checked.
Can you tell whether conversions are missing without access to my account?
Partly, and it is worth being exact about which part. A public check can establish whether a conversion request is sent when an enquiry completes, whether the ad click survives the journey, and whether calls appear to be measured.
What it cannot see is anything happening away from the browser. Imported conversions, server-side measurement and offline conversion uploads are all invisible from outside, and their absence from a website inspection is not evidence that they are missing.
So the useful output of a public check is not a verdict. It is a specific list of things to ask about, each with the evidence attached, and an explicit statement of what the evidence could not settle.
What this cannot tell you If your agency tells you conversions are imported from a CRM, an external check genuinely cannot confirm or contradict that. What it can check is whether the information that import would need is being captured in the first place.
Find out which of the six it is.
The free check follows the enquiry route on your own site and records where measurement stops. No account access, nothing changed, and the limits stated in writing.
How long should it take for a conversion to appear in Google Ads?
Usually within a few hours, though up to a day or so is normal and some conversion types take longer. Before assuming measurement is broken, this is the first thing to rule out.
There is a second, less obvious delay. Google attributes a conversion back to the date of the original click, so a conversion recorded today can appear against last week once it lands. A report pulled too early will understate recent days and then quietly correct itself, which is not a fault but does look like one.
Could our cookie banner be stopping conversions from being recorded?
It can, and this is a genuinely common cause that gets missed because the banner was added for good reasons by somebody who was not thinking about measurement.
Where a visitor declines, or where the banner blocks measurement scripts until a choice is made, the conversion may never be recorded or may lose the information connecting it to the ad click. A banner that blocks scripts before any choice is made will affect every visitor, including the ones who would have accepted. An external check can observe how the site behaves under each choice, which is one of the more useful things it can establish without account access.
We rebuilt our website. Should the tracking be re-checked?
Yes, and it is worth treating as part of the launch rather than something to look at if the numbers seem odd later. A rebuild replaces the pages where measurement was installed.
The failure is rarely total, which is what makes it expensive. The homepage tag survives because somebody remembered it, and the thank-you page, the booking step or the second landing page does not. Conversions keep arriving at a lower rate, and nobody can say when they changed.
Will Google tell us if conversion tracking stops working?
Sometimes, and not reliably enough to depend on. Google Ads does flag conversion actions that have recorded nothing recently, and that warning is worth acting on when it appears.
What it cannot flag is the more common situation: measurement that is still working, still recording, and recording the wrong thing. From the platform’s point of view a conversion action that fires steadily is healthy, whatever it happens to be counting.