Your Google Ads
look better than they are.

The 'Conversions' column looks like a count. It's partly an estimate. When privacy settings or browsers stop Google seeing a conversion, it models one, and Google says plainly that the column reports both modelled and observed conversions. Modelled conversions can take up to five days to settle, so last week's numbers can still move.

Modelling isn't a scam. It exists so automated bidding isn't learning from half the picture. But it does mean the number you're judging the account on is Google's estimate of Google's own performance.

The biggest flattery is usually brand search. Someone types your business name, clicks your ad instead of the organic link right under it, and buys. The ad gets the credit. Would they have bought without it? Often, yes.

This has been tested properly. eBay ran large field experiments, switching ads off for some areas and comparing, and found brand-keyword ads had no measurable short-term benefit. Across its paid search, true returns were a fraction of what the usual reports suggested.

To be fair, you are not eBay. If competitors bid on your name, a brand campaign can be protecting real sales. The point isn't to switch brand off. It's to stop letting brand results stand in for how the rest of the account is doing.

Watch Performance Max here. It can pick up brand searches unless you tell it not to, which is why Google offers brand exclusions for Performance Max and Search.

Some conversions are credited to people who never clicked. Google's default windows are 30 days after a click, 3 days after an engaged view of a video ad, and 1 day after an ad is simply seen. A person who scrolled past a YouTube ad and bought from you the next day can count.

Sometimes that's real influence. Sometimes it's coincidence with good timing. You can't tell from the report, which is the problem.

Google's own advice is to count leads as one conversion per click, because a second form from the same person isn't a second customer. But 'every' is the default for website conversions. Someone who sends the form twice is two leads in the report and one in your inbox.

The other usual culprit: actions set as main conversions that aren't sales or enquiries at all. Page views, button clicks and time on site make the number bigger and the bidding worse.

Pull these from your account for the same date range, then put them in below. It's simple subtraction, and it stays in your browser. If you don't know a number, leave it blank and the calculator skips that step.

Subtraction gets you closer. A test gets you the answer. The idea is simple: show ads to one group, hold them back from a similar group, and compare. Google offers this as Conversion Lift, though it isn't available for every account.

For smaller accounts, a region holdout does the job: pause a campaign in a few comparable areas for a few weeks and watch total enquiries, not just the ones the platform reports. That's the approach behind Cause Testing.

And sometimes the test says the ads are doing more than Google claims, especially when people see an ad, then ring you instead of filling in a form. Testing isn't about proving ads don't work. It's about knowing which ones do.