Your ad platform and your store report different sales. Here is which to trust

Your ad platform and your store report different sales numbers. Neither is lying. Here is what creates the gap, what it costs you in misdirected budget, and how to close it without guessing.

By Yannis Spanenburg · 2026-08-04 · 7 min read

Meta says the campaign produced 180 purchases last month. Shopify says 121. Nothing has been stolen and neither system is broken. The two are counting different things, and the space between them is where a lot of ecommerce budget quietly disappears.

Most owners settle this by picking whichever number they prefer and scaling spend on it. That is the expensive move. If you cannot say how big your own gap is, every decision downstream of it is a guess with a decimal point on it.

What is the tracking gap?

The tracking gap is the difference between the conversions your ad platform reports and the orders your business actually recorded. Four things create it: browsers shortened cookie lifetimes, iOS let users decline tracking, consent banners block pixels before they fire, and every ad platform counts the same sale as its own.

Why the gap got so wide

Browsers shortened the memory

Safari caps cookies written by JavaScript at seven days, and in some cross-site cases at twenty four hours. Firefox blocks known trackers by default. A customer who clicks your ad on Monday and buys three weeks later arrives, as far as the pixel is concerned, out of nowhere.

iOS let users opt out

Since iOS 14.5 in 2021, apps have had to ask permission before tracking users across other companies' apps and websites. Most people declined. For any brand with meaningful iPhone traffic, a large share of app-sourced conversions stopped being visible to the pixel more or less overnight.

Your consent banner sits in front of your pixel

If you sell into the EU or the UK, your cookie banner is part of your measurement stack whether you designed it that way or not. Google has required Consent Mode v2 signals for EEA traffic since March 2024. Configured badly, the banner deletes your data. Configured well, it models around the gap.

Every platform claims the same sale

Meta credits itself for a purchase within seven days of a click, and in most default setups within one day of a view the customer never consciously noticed. Google credits the same order. Your email tool credits it too. Add the dashboards together and you will have sold more than you actually sold.

So which number is correct?

Neither, and that is the useful answer. The ad platform over-reports: it is built to claim credit, and it now models the conversions it can no longer observe. Your backend under-reports: last-click attribution hands the entire sale to whichever channel happened to be standing there at the end.

The only number that cannot argue with you is the one in your bank account. Revenue in, spend out. Everything above that line is an estimate, and estimates stay useful as long as you know which way each one leans.

What the gap costs in real money

One client came to us spending roughly €12,000 a month on ads with a purchase event that had been firing incorrectly for over a year. The platform was optimising towards a broken signal, which means the algorithm was confidently buying the wrong people. That is €144,000 of annual ad spend being steered by bad data. The fix was quoted at €3,000.

To be precise about it, fixing the tracking did not hand back €144,000 in profit. It moved €144,000 a year from guesswork into intent. That is the part most owners never put a number on, because the invoice for bad measurement never actually arrives.

How to close it

1. Move the important events server side. A conversion sent from your own server to Meta's Conversions API, or through Google's server-side tagging, does not depend on a browser that has been told to forget things. For most stores this is the single biggest recovery available.
2. Treat consent as configuration, not as a compliance formality. Consent Mode v2 lets Google model the behaviour of visitors who declined, so a banner costs you a slice of precision rather than a blackout.
3. Pick one source of truth and write it down. For ecommerce that is almost always your store platform or your accounting. Ad dashboards then become directional instruments rather than scoreboards, which is what they were always better at anyway.
4. Judge total spend on blended numbers. Total revenue divided by total advertising spend cannot be double counted, cannot be modelled and cannot be negotiated with. Platform ROAS tells you how to steer inside a channel. Blended tells you whether the business is actually winning.

If you cannot say how wrong your numbers are, you are not measuring. You are hoping, with extra steps.

Three checks worth an afternoon

Three checks, before you change anything.

1. Compare last month's platform-reported conversions against real orders in your store and write the percentage difference down.
2. Open your tag manager and confirm the purchase event fires exactly once, carries a value, and sits on the real confirmation page.
3. Check whether your consent banner is blocking the pixel outright or passing it a proper consent signal.

That is an afternoon of work and it answers a question worth far more than an afternoon. Most businesses assume they have a marketing problem. A good number of them have a measurement problem instead.