Google Ads Conversion Tracking Problems

The Conversion You Are Counting Probably Did Not Happen

Here’s a test worth running before you read any further.

Open your Google Ads account and note how many conversions it recorded last month. Then open your diary, your inbox or your CRM and count the enquiries you actually received.

If those two numbers are close, you’re in a minority, and the rest of this post is maintenance reading.

If the dashboard says forty and you remember about twelve, something in between is wrong. And every decision made on top of that forty was made on a number that doesn’t exist.

Why this matters more than any other setting

Conversion tracking isn’t one report among many. It’s the input everything else runs on.

Your Smart Bidding strategy optimises towards it. Your cost per conversion is calculated from it. Google’s recommendations are justified by it. Performance Max learns from it.

Get it wrong, and you don’t get slightly worse results. You get an automated system working very hard, very efficiently, in the wrong direction.

A broken conversion signal doesn’t fail quietly. It fails at scale, on your budget.

That’s why every other post we’ve written this month ends up back here. Fix this first, or nothing else you do is measurable.

THE THREE WAYS THE NUMBER GOES WRONG

Most articles treat this as one problem. It’s three, and they need different fixes.

Failure one: counting the wrong thing

This is the most common and the most expensive.

The classic version is a tag placed on the contact page rather than the thank-you page. Everybody who looks at your contact form counts as a conversion, whether or not they fill it in.

The other version is counting page views as conversions. Every page load registers, so your numbers inflate, and Smart Bidding starts optimising towards people who browse rather than people who buy.

You can spot this without any technical work. If your conversion rate looks implausibly good, or your conversions roughly track your traffic rather than your enquiries, this is usually why.

TLC

Failure two: counting the same thing twice

This one has become much more common since GA4 arrived.

The pattern is simple. You have a Google Ads conversion tag firing on your thank you page. You’ve also linked GA4 and imported its equivalent event as a conversion action.

Both fire for the same customer doing the same thing. Every enquiry gets counted twice.

Your conversion numbers double and your cost per conversion halves, and neither reflects anything that happened in the real world.

Failure three: not counting at all

The quiet one, and the one that costs the most in wasted budget.

A conversion action stops recording. Usually a tag disappeared during a website redesign. Or a consent setting stopped it from firing. Or somebody tidying up renamed an imported GA4 event.

Nobody notices, because the number goes down gradually rather than to zero, and there’s always a plausible explanation for a slow month.

Meanwhile, any campaign optimising towards that action is bidding towards nothing at all.

WHY THIS GOES WRONG SO OFTEN

It’s worth understanding why broken tracking is so common, because it explains why checking yours is worth an afternoon even if somebody competent set it up.

This isn’t a story about incompetence. It’s a story about ownership.

Nobody owns it

Tracking sits between two jobs and belongs to neither.

Your web developer builds the site and installs whatever they were asked to install. Your marketing person or agency runs the campaigns and reads the numbers.

The bit in between, checking that what’s installed still measures what the campaigns assume it measures, isn’t in either job description.

So it gets set up once, correctly, and then nobody looks at it again for three years.

Websites change, and tags don’t come with them

The single most common cause of a dead conversion action is a website redesign.

The new site launches and everyone admires it. A tracking snippet that lived on the old thank you page didn’t make the move. Nobody notices because nobody was watching that number closely in the week the site changed.

It happens on a smaller scale too. A form plugin gets swapped, a thank you page URL changes, somebody tidies a tag manager container.

The GA4 migration left debris

A lot of accounts picked up duplicate conversions during the move to GA4.

The pattern was predictable. Somebody linked GA4 and imported its events. Nobody removed or demoted the Google Ads tags already doing that job.

Both kept firing. Conversion counts roughly doubled, and because everything was changing at once, it looked like a reporting quirk rather than an error.

The person who built it has gone

Agencies change. Developers move on. Marketing managers leave.

Tracking is the least documented part of most setups, so the knowledge walks out with them. What’s left is a list of conversion actions with unhelpful names and nobody who can say what any of them measure.

None of that is anybody’s fault exactly. It just means the default state of conversion tracking is drift, and the only fix is to check it deliberately.

PRIMARY AND SECONDARY, AND WHY IT DECIDES EVERYTHING

This is the setting most business owners have never heard of, and it quietly determines what your budget chases.

Every conversion action in your account is set as either primary or secondary.

Primary actions are used by Smart Bidding. They directly influence how Google bids on your behalf.

Secondary actions are recorded for reporting only. They appear in your columns; they tell you things, and they don’t affect bidding at all.

The mistake this causes

The common error is marking low-value actions as primary. Page views. Scroll depth. Newsletter signups. A click on a phone number that nobody answered.

When those are primary, Smart Bidding optimises towards them, because you told it to. It will get very good at finding people who scroll and sign up for newsletters.

It will get correspondingly worse at finding people who buy, because nothing in the signal rewards that.

How to set it properly

The rule is straightforward. Primary is reserved for the thing that makes you money.

For a service business, that’s usually a completed enquiry form or a phone call over a meaningful length. For a shop, it’s a purchase.

Everything else goes to secondary. Newsletter signups, brochure downloads, video views. Useful to know about, dangerous to bid on.

If it wouldn’t make you happy to receive a hundred of them, it should not be a primary conversion.

That single sentence resolves most of these decisions.

THE THIRTY MINUTE AUDIT

Here’s the whole thing, in the order I’d do it, with no tag manager required.

  • Open Goals, then Conversions, then Summary. This screen lists every conversion action in the account and its current status.
  • Read the status column first. Anything showing as inactive, or as having no recent conversions, is a problem. That action is not recording, and any campaign chasing it is bidding towards nothing.
  • Check the primary and secondary columns. Anything set as primary that is not a genuine enquiry or a sale should be moved to secondary.
  • Look for duplicates by name. A Google Ads tag conversion and an imported GA4 event describing the same action means every enquiry counts twice.
  • Install Google Tag Assistant in Chrome, then load your thank you page with it running. Your conversion tag should fire exactly once. Twice means duplication in your setup.
  • Now go back and load the form page itself, without submitting anything. If a conversion fires there, the tag is on the wrong page and your numbers are inflated.
  • Check the conversion window and the counting setting on your main action. One conversion per click is usually right for a service business, and every conversion suits a shop where repeat purchases matter.
  • Finally, compare a full month of Google Ads conversions against the enquiries in your own records for the same dates. That comparison is the real test and everything above is just diagnosis.

Most accounts find something in the first ten minutes. The status column alone catches a surprising number of dead tags.

WHAT TO DO WITH WHAT YOU FIND

Different findings need different responses, and the order matters.

Fix the dead actions first

An inactive conversion action is an emergency, because campaigns may be optimising towards it right now.

Find out when it stopped. If it lines up with a website change, the tag went missing in the rebuild and needs putting back.

While it’s broken, be very careful about trusting recent performance data or making bidding changes based on it.

Then fix the duplicates

Pick one source of truth. Either the Google Ads tag or the imported GA4 event, not both.

For most small businesses, the Google Ads tag is simpler and more direct. Set the other one to secondary so you keep the reporting without the double count.

Expect your conversion numbers to roughly halve when you do this, and your cost per conversion to double. Nothing got worse. The old numbers were wrong.

Then fix the wrong-page tags

Moving a tag from the form page to the thank you page will cut your conversion count sharply.

That’s the correct outcome. You’ve stopped counting people who looked at a form and started counting people who submitted one.

Then reset the primary flags

Once the counting is accurate, decide what the account should be optimising towards, and set only that as primary.

Give Smart Bidding a fortnight to adjust after any change here. The signal it learns from has just changed shape.

TLC

TRACKING PHONE CALLS WITHOUT COUNTING WRONG NUMBERS

For a lot of service businesses, the phone still brings in most of the work, and call tracking is where the numbers go wrong most quietly.

There are two things you can count, and they’re not the same.

Clicks on a phone number are not calls

The easiest thing to track is a click on your phone number on a mobile site. It’s simple to set up, and it’s what a lot of accounts count.

The trouble is that a click means somebody’s phone opened the dialler. It doesn’t mean they pressed call, and it certainly doesn’t mean anyone answered.

Some of those clicks are accidental. Plenty are people who changed their mind between tapping and dialling.

Count them as conversions, and you’re inflating your numbers with intentions rather than conversations.

Calls of a meaningful length are conversions

The better measurement uses a Google forwarding number or a call tracking provider, and records calls that actually connected.

Then you set a minimum duration, and this is the setting that matters.

A call lasting eight seconds is a wrong number or a hang-up. A call lasting ninety seconds is somebody describing a job.

Set the threshold to whatever a real enquiry sounds like in your business. For most trades and professional services, somewhere between sixty and ninety seconds separates the two reliably.

Too short and you count misdials as sales. Too long and you’ll miss genuine quick enquiries from people who already knew what they wanted.

The trap with counting both

If you track number clicks and connected calls, you can end up counting the same person twice. They tap the number, the click registers, then the call connects, and that registers too.

Pick one. For most businesses, connected calls of a decent length are the honest measure. Phone number clicks belong in secondary, if you keep them at all.

And check what happens to calls that come in outside opening hours. If nobody answers, that isn’t a conversion, however keen the caller was.

ENHANCED CONVERSIONS, IN PLAIN ENGLISH

You’ll see enhanced conversions recommended everywhere, usually without an explanation of what it does.

The problem it solves is under-counting. Cookies are less reliable than they used to be. So some genuine conversions never get connected back to the click that caused them.

Enhanced conversions use hashed first-party data, such as an email address the customer has already given you, to help make that connection. The data is hashed rather than sent in the clear.

Enhanced conversions for leads applies the same idea to offline conversion import. A sale closed weeks later gets matched back to the original click.

Is it worth it for a small business?

If you sell online and take email addresses at checkout, it’s usually worth setting up.

The offline version earns its place in a service business, where an enquiry becomes a job three weeks later. It teaches the bidding which enquiries turned into money, rather than treating them all alike.

That said, it’s a refinement. Enhanced conversions on top of a tag that fires on the wrong page will simply give you more accurate wrong numbers.

Get the basics right first. Then improve the accuracy.

CONSENT, AND WHY YOUR NUMBERS MIGHT HAVE DROPPED

If your conversions fell noticeably at some point and nothing obvious changed in the account, consent settings are worth checking.

When a visitor declines cookies, tags may not fire. That’s the system working as intended, and it means some real conversions go unrecorded.

Google applies modelling to estimate some of what’s missing, and you can see whether that’s active on a conversion action’s tracking status.

The practical point for a business owner

You don’t need to become an expert in this. You need to know two things.

First, a drop in recorded conversions isn’t always a drop in real enquiries. Check your own records before panicking or making changes.

Second, check when your consent banner was installed or changed. If that lines up with the drop, there’s your explanation, and whoever built it should check the configuration.

The failure mode to avoid is reacting to a measurement change as though it were a performance change. That’s how good campaigns get switched off.

THE ONLY TEST THAT REALLY PROVES ANYTHING

Everything above checks whether the plumbing works. None of it tells you whether the conversions were worth having.

For that, you need your own records, and this is the comparison I’d build into a monthly routine.

Take a full month. Count what Google Ads reported. Count the enquiries that actually reached you. Then count the ones that were genuinely worth having.

Three numbers, and the gaps between them tell you different things.

A gap between reported and received usually means a tracking fault. Duplicates, wrong-page tags, or form spam being counted.

A gap between received and worth having is a different problem entirely, and it isn’t a tracking problem. That’s a targeting and qualification issue, and it’s the subject of its own post.

Do this once, and you’ll know more about your advertising than the dashboard has told you all year.

The platform can count actions on your website. Only you can count the ones that mattered.

A WORKED EXAMPLE, START TO FINISH

A hypothetical makes the shape of this clearer. Take a commercial locksmith on nine hundred pounds a month, running two Search campaigns on Maximise Conversions.

The account reports 63 conversions last month at just over fourteen pounds each. On paper that’s excellent for the trade.

The owner counts the actual enquiries. Twenty-two, of which about fifteen were worth having.

Sixty-three against twenty-two. That gap is the whole problem, and it takes about half an hour to explain.

They open the conversions summary

Seven conversion actions, five marked primary.

Two of them are clearly duplicates by name. A Google Ads tag called “Contact Form” and an imported GA4 event called “generate_lead”, both firing on the same submission.

One is called “Phone Click”, and it’s set to primary. Another, “Newsletter Signup”, is also primary.

One shows as inactive with no recent conversions, which nobody had noticed.

They run Tag Assistant

Loading the thank you page shows the conversion tag firing twice, confirming the duplicate.

Then they load the contact page without submitting anything. A conversion fires there too, which explains a chunk of the remaining gap. Everyone who looked at the form was being counted.

What they change

The GA4 import goes to secondary, keeping the reporting without the double count.

The tag on the contact page gets moved to the thank you page.

“Phone Click” goes to secondary and gets replaced with connected calls over sixty seconds. “Newsletter Signup” goes to secondary because a hundred newsletter signups wouldn’t make anybody happy.

The inactive action turns out to be a quote request form removed in a site rebuild eight months ago. It gets deleted rather than fixed, because the form no longer exists.

What the numbers do

The following month reports 19 conversions at about forty-seven pounds each.

On the dashboard, that looks like a disaster. Conversions down two-thirds, cost per conversion more than tripled.

In reality, nothing about the business changed. Nineteen against twenty-two actual enquiries is close enough to believe. For the first time, the account is optimising towards something real.

Six weeks later, with Smart Bidding working from an honest signal, the qualified enquiries went up on the same spend. That’s the payoff, and it was invisible for the first fortnight.

WHEN TO REBUILD RATHER THAN PATCH

Sometimes the right answer is to start again, and it’s worth knowing the signs.

Rebuild when nobody can explain what the existing conversion actions measure. If there are eleven of them with names like “Conversion 3” and no documentation, the fastest route is a clean setup.

Rebuild when the person who built it has gone, and the site has been redesigned since. The chance of the tags still being correct is low.

Rebuild when you can’t get the reported numbers within a sensible distance of your own records after fixing the obvious faults.

Patch when the setup is broadly sensible and one or two things are wrong. Most accounts are in this position.

And do it before, not after, any significant change to bidding, budgets or campaign types. Every one of those decisions depends on this data being right.

THE MISTAKES THAT WASTE THE MOST MONEY

  • Placing the conversion tag on the form page rather than the thank you page, so every visitor counts.
  • Running a Google Ads tag and an imported GA4 event for the same action, doubling every enquiry.
  • Setting page views, scrolls or newsletter signups as primary conversions, so Smart Bidding chases them.
  • Never checking conversion action statuses, and leaving a dead tag in place for months.
  • Treating a drop caused by consent settings as a drop in genuine enquiries, and cutting a campaign that was working.
  • Adding enhanced conversions on top of a broken setup, producing more accurate versions of the wrong number.
  • Making bidding or budget decisions during a period when tracking was known to be faulty.
  • Never comparing the platform’s numbers against your own records, which is the only check that proves anything.

TLC

FREQUENTLY ASKED QUESTIONS

How do I check my Google Ads conversion tracking is working?

Start in Goals, then Conversions, then Summary, and read the status of every conversion action. Anything inactive or with no recent conversions needs investigating straight away.

Then install Google Tag Assistant in Chrome and load your thank you page with it running. The conversion tag should fire exactly once.

Then load the form page without submitting. Nothing should fire there at all.

Why are my Google Ads conversions double counting?

Almost always because two things are recording the same action.

The usual pair is a Google Ads tag on the thank you page, plus the same event imported from GA4.

Choose one as your source of truth and set the other to secondary. Your numbers will halve, which is the correct result rather than a loss.

What is the difference between primary and secondary conversion actions?

Primary actions are used by Smart Bidding and directly influence how Google bids. Secondary actions are recorded for reporting only.

The test I’d use: if receiving a hundred of them wouldn’t make you happy, it shouldn’t be primary.

Enquiries and sales are primary. Newsletter signups and brochure downloads are secondary.

What are enhanced conversions?

A way of improving measurement accuracy using hashed first-party data, such as an email address the customer already gave you.

They help connect conversions back to clicks where cookies alone fall short.

The version for leads extends this to offline conversion import. A job won three weeks later gets matched to the click that started it.

Why do my Google Ads conversions not match my CRM?

Some difference is normal. The platform counts actions on your website; your CRM counts enquiries a person received.

A large gap points to duplicates, a tag on the wrong page, or form spam being recorded as conversions.

Compare the same dates in both and look at which direction the gap runs. Over-reporting and under-reporting have completely different causes.

What does it mean when a conversion action shows as inactive?

Google isn’t recording that action any more.

Common causes: a tag removed during a website redesign, a consent setting blocking it, or an imported GA4 event renamed or deleted.

Treat it as urgent. Any campaign optimising towards an inactive action is bidding towards nothing.

Should I track phone calls as conversions?

Yes, for most service businesses, but with a length threshold.

A call that lasts eight seconds is a wrong number. A call that lasts ninety seconds is a conversation.

Set the minimum duration to something that reflects a real enquiry in your business; otherwise, you’re counting misdials as sales.

WHERE TO START THIS WEEK

Do the two-number test first. Google Ads conversions for last month, against enquiries you actually received.

If they’re close, spend twenty minutes on the audit above as a health check and get on with your month.

If they’re not close, stop making changes to bidding and budgets until you know why. Every decision you make in the meantime rests on a number that isn’t real.

Then work down in order. Dead actions first, duplicates second, wrong-page tags third, primary flags last.

None of this is glamorous, and none of it will make your dashboard look better. Most of it will make the numbers look worse, because the old ones were flattering you.

But every other improvement depends on this one being right. That’s why it’s worth an afternoon before anything else gets touched.

Did You Enjoy This Blog Post?

I hope you enjoyed this blog post, and thank you so much for being here. We also upload videos to our YouTube channel every weekday. Please subscribe so you are one of the first to be notified.

If you enjoyed this blog, you may also like:

Tools We Use & Love!
Scroll to Top