The campaign is running, the ads are fine, traffic is coming in. Meta’s interface says there were three purchases. In your store there were eleven. Nobody is lying — Meta simply no longer sees everything that happens on your site, and has started estimating the difference.
That is the normal state of things today, not a fault. But it is worth understanding why, because it decides which numbers you will believe and how much money you will waste if you believe the wrong ones.
What the pixel does, without the jargon
The pixel is a small piece of code on your website. When somebody arrives, that code tells Meta what happened: viewed a product, added it to the basket, submitted a form, bought. Nothing more mysterious than that.
Two consequences matter. First: without it, Meta doesn’t know what happened after the click, so it also can’t go looking for more people who behave like your buyers. Second: that very ability is what makes advertising effective in the first place. The system learns from results. Show it no results and it stays guessing.
What changed, and why the numbers don’t match
From roughly 2021 onwards, measurement has been substantially eroded, from three directions at once:
- Apple gave users a choice. With App Tracking Transparency in iOS 14.5, apps have to ask explicitly whether they may track across apps and sites. The large majority of people decline. All of those visits become partly invisible to Meta.
- Browsers closed off third-party cookies. Safari and Firefox some time ago, the others heading the same way. A pixel that once recognised a visitor a week later often no longer does.
- Ad blockers. A share of visitors never load that code at all.
The result: Meta’s reports are partly modelled — some measured, some estimated statistically. They will therefore never match your store, your accounts or your CRM exactly.
A practical rule worth holding to: the source of truth about revenue is your own system, not the advertising interface. Use Meta’s numbers to compare ads against each other (which is better than which), and your own to answer whether the whole thing pays.
Server-side tracking, which makes the difference
This is where the Conversions API comes in — tracking where the event is sent by your server rather than by the visitor’s browser. An ad blocker doesn’t interrupt that, and the dependence on cookies is lower.
For stores that advertise seriously it is effectively mandatory today: without it the system gets substantially less feedback, campaigns learn more slowly, and that shows up in the cost per order. Setting it up isn’t something you click into place in the interface; it is work on the site, which is precisely why plenty of people skip it.
Consent isn’t bureaucracy, it is a precondition
Now the uncomfortable part, which advertising advice often leaves out. You may not load tracking before the visitor consents. Advertising and retargeting cookies are not necessary for the site to function, so they require clear consent — and it has to be given before loading, not after.
In practice that means:
- a cookie banner where declining is as easy to reach as accepting — not hidden in small print;
- the pixel and similar code load only once consent is given;
- the visitor can change their consent later;
- the privacy policy states who processes the data and why.
Yes, this reduces the volume of data. The alternative isn’t “more data”, it is the risk of a fine and the loss of trust when it emerges that you tracked people without asking. How we handle it is described in our privacy and cookie policy.
Retargeting is still the best buy
Despite all the eroded measurement, retargeting remains the most profitable thing in advertising. Someone who has been on your site already knows you. The ad isn’t a complete interruption for them, it is a reminder.
Three cases that almost always pay: whoever left a full basket; whoever viewed a particular product and didn’t buy; whoever was on the pricing page and didn’t enquire. The audiences are smaller than they were a few years ago, but the cost per result is still usually the best in the whole account.
And the page on the other side of the click
The best tracking in the world doesn’t help if the ad brings someone to a page that takes five seconds to load on a phone. You paid for that visit and you will never see it — they left before anything rendered.
So the same applies as to all paid traffic: a fast page on mobile, a message that matches the ad, and one obvious next step. For stores that is often the difference between a campaign that works out and one that almost works out — more on that on the online stores page.
A short checklist
- Is the pixel installed, and does it only fire after consent?
- Is server-side tracking (Conversions API) set up?
- Is an order or a submitted form being measured — not a page visit?
- Is the ad account owned by the company rather than a personal profile?
- Is at least one retargeting campaign running?
- Do you check the return in your own store or CRM, not only in Meta’s report?
If you answered “I don’t know” to more than two of those, that is expected — this is the part of advertising that isn’t visible, and so nobody mentions it until it starts costing money.
Setting up tracking and running campaigns starts at 249 €/month for Facebook and Instagram, with the ad budget separate; what is included is on the advertising pricing page. If you would first just like to know whether your tracking is even set up correctly, we look at that as part of a website audit. Write through the contact form or call +386 40 529 425.