Your Google Ads conversions are lying to you

Your account reported 43 conversions last month. Your calendar says you closed six jobs. Nobody is lying — those are just two different questions, and only one of them pays your rent.

The gap between them is the single most expensive misunderstanding in small-budget advertising. Not your bids. Not your keywords. What your account counts as a win.

The short answer

A “conversion” is not a customer. It’s whatever action you once told Google to record. Most accounts we open are counting button presses, page loads, and the same person twice — and then handing that number to an algorithm and asking it to go find more.

Fix what counts before you touch your budget, your bids, or your keywords. Everything downstream of a bad conversion number is guesswork with a decimal point.

The four things quietly inflating your number

1. You’re counting every conversion instead of one. Google lets you pick, per conversion action, whether to count every conversion or only one per ad click. “Every” makes sense for an e-commerce site where each order is real money. For a lead-gen business it means one anxious homeowner tapping your call button three times in ninety seconds shows up as three leads. Leads should almost always be set to One.

2. Page views are wearing a conversion costume. A tag fires on /contact instead of on the thank-you state, so every person who looked at your contact page counts as someone who contacted you. Same thing when a thank-you page is reachable by direct link, bookmarked, or crawled.

3. A tap on a phone number is not a phone call. Click-to-call and actual call tracking are different mechanisms — Google supports several distinct ways to track conversions, and the useful one for phone leads counts calls that lasted a minimum duration. Set that minimum to 60 seconds and watch a third of your “calls” evaporate. They were wrong numbers, hang-ups, and people ringing to ask if you’re hiring.

4. Micro-actions got promoted. Newsletter signups, PDF downloads, a click on your map pin — all worth observing, none of them a customer. If they’re in your primary conversions column, they’re in your reported total and in your bid strategy.

Why this costs real money

If you’re on any Smart Bidding strategy, Google is not optimizing toward revenue. It’s optimizing toward the thing you labelled a conversion. Feed it tap-to-call events and it will get very good at finding people who tap and don’t talk. That’s not the algorithm failing — that’s it succeeding at the job you gave it.

It also poisons every decision above it. Your cost per conversion looks like €22 when your real cost per job is €157. You conclude the channel works, raise the budget, and buy more of the same. Or the inverse: a genuinely healthy account looks broken because half its wins aren’t being recorded at all, and you switch it off.

This is the same failure mode as chasing your optimization score — a number that is easy to move and doesn’t measure what you think. And it’s why the budget math we’ve published only means anything once the conversion column is honest.

The 20-minute audit

Open your account and work through this in order. You don’t need an agency for any of it.

• List what you’re counting. Goals → Summary. Read every conversion action by name. If you can’t say in one sentence what a human did to trigger it, that’s your first problem.
• Check the Count setting on each one. Lead actions → One. Purchases → Every.
• Check what’s primary vs secondary. Only things that represent a real human contact belong in primary. Everything else moves to secondary so you can still see it without bidding on it.
• Check for doubles. Importing GA4 goals and running the Google tag on the same event is the most common duplicate we find. One event, one source.
• Set a call-length minimum. 60 seconds is a sane floor for service businesses.
• Segment by conversion action. In your campaign report, segment → conversions → conversion action. This shows you which campaigns are winning on real leads and which are winning on noise. It’s usually not the ones you’d guess.

Closing the loop is the part only you can do

Even a perfectly configured account stops at the lead. It has no idea which of those calls turned into a €4,000 job and which was a tyre-kicker who ghosted you.

The proper fix is importing offline conversions — feeding closed jobs back into Google so bidding learns what a good lead looks like, not just a lead. Google now steers most advertisers toward enhanced conversions for leads for this.

Here’s the honest version, though: if you’re doing eleven leads a month, that integration is not your bottleneck. A spreadsheet is. Lead date, source, what they wanted, whether it closed, what it was worth. Ninety days of that beats any API, because at your volume you can read every row yourself and see the pattern with your own eyes. Build the plumbing when the volume justifies it.

FAQ

My conversions dropped 40% after I fixed this. Did I break something? No, you deleted a fiction. The jobs you close haven’t changed. Re-baseline your cost-per-lead target against the new number and keep going.

Should I just track everything and sort it out in reporting? Track everything, yes. Bid on everything, no. That’s exactly what the primary/secondary split is for.

Does this matter if I’m on manual CPC? Less for bidding, just as much for judgement. You’re still using that number to decide whether ads are worth it.

Before you spend another euro

If your conversion column and your invoices disagree, nothing else in the account is worth discussing yet. Fix the counting, run thirty days, then decide.

If you want a second pair of eyes on it, tell us what you’re seeing. We’ll look at the account and tell you straight — and if the answer is that your ads are fine and your landing page is the actual leak, we’ll say that instead of selling you more ad management.