Meta Ads and Google Ads show fewer sales than your store
When the ads dashboard shows fewer sales (or more) than the store, there are specific causes: blockers, cookies that are rejected or cut short, customers who never come back from the payment gateway, attribution windows and duplicates. We explain how to measure how much is being lost, what server-side tracking fixes and how it affects ROAS and bidding.
If Meta Ads or Google Ads record fewer purchases than your store actually has, it is almost always because the purchase notification never reached the platform: the browser blocked it, the customer rejected cookies or closed the tab before returning from the payment gateway. If they record more, there are usually duplicate events. In both cases the ROAS in the dashboard is not the real one, and automated bidding learns from the wrong data.
It is normal, up to a point, for conversions not to match orders: an ads platform and a store count different things. The problem starts when the gap is large, changes from one month to the next or nobody knows where it comes from. This article goes through the causes in plain language, suggests a simple way to measure how much is being lost and separates what can be fixed from what cannot. We cover the gap between GA4 and orders in another article on the blog; here we are talking about ads platforms and ROAS.
When the dashboard shows fewer sales than the store
Blockers
Extensions that block ads and trackers, and some browsers by default, stop the pixel from loading. The customer buys as normal, but the platform receives nothing.
Rejected cookies
If the customer says no in the cookie banner, the advertising pixel must not fire. That purchase is not attributed, and rightly so: it is their decision.
Browsers that cut cookies short
To attribute a purchase to an ad, the platform needs to recognise that the person buying today is the one who clicked a few days ago. It does this with cookies, and some browsers shorten their lifetime. If the customer takes a while to decide, the link to the ad is lost even though the pixel works.
Customers who never come back from the payment gateway
In many stores the purchase event fires on the order confirmation page. Anyone who pays on the bank's page, or in a payment app on their phone, and closes it without returning to the store never loads that page. The order is paid and sitting in your back office; for the platform it does not exist.
Another device or another browser
The click happens on a phone inside a social network and the purchase that evening on a computer. Platforms try to join the two when the customer is identified, but they cannot always do it.
The pixel is broken or has stopped working
A theme change, a store update or a new module can stop the purchase event from being sent, or make it go out without an amount. Nobody warns you: conversions simply drop and the campaign gets the blame.
When the dashboard shows more sales than the store
- Duplicates. The purchase event fires twice: when the confirmation page is reloaded, because two pixels are installed at once or because browser and server send the same purchase without a shared identifier.
- Orders that turn out not to be orders. Failed payments, cancelled orders and returns still count as conversions if nobody corrects them.
- Different amounts. The dashboard may add up the total including VAT and shipping while you look at the net amount, or the other way round.
- Generous attribution. Each platform claims the sale if it took part, even if the customer also came through another one. Add up every platform's dashboard and you get more sales than there were.
Differences that are not errors
Even when everything works properly, the figures will not be identical, for two reasons:
- Attribution windows. Each platform decides for how many days after the click (or after the ad was seen) it claims a purchase, and each one uses its own. The same sale may count in one and not in another.
- Conversion date. Some platforms record the sale on the day of the click and others on the day of the purchase, so totals per day or per week do not match the store's even if the total for the period is similar.
That is why the goal is not to match to the penny, but to know how big the gap is and to keep it stable.
How to check how much is being lost
- Pick a closed period, for example the last two full weeks, and wait a few days so the platforms can finish attributing.
- Pull the store's valid orders for that period: paid, with no cancelled or test orders. Note the number and the amount.
- Pull the purchase conversions from each platform for the same period, using the same time zone.
- Compare the total number of events received first, not only those attributed to ads. The platform's events manager shows how many purchases arrived, whether or not they came from an ad. If the store has considerably more orders than purchases received, there is a technical loss.
- Do an order-by-order check on a sample: look up ten specific orders and see whether their event arrived, with the correct amount.
- Look for duplicates: order identifiers that appear more than once.
- Repeat it every month. A gap that suddenly grows is usually a broken pixel, not a change in the market.
How it affects ROAS and automated bidding
ROAS is revenue divided by ad spend. If the platform only sees part of the sales, the ROAS in the dashboard comes out lower than the real one, and decisions go astray: campaigns that were profitable get switched off and the budget stays on the ones that are simply measured better.
The second effect is less visible and more expensive. Automated bidding strategies learn from the conversions they receive: who to show the ad to, how much to bid, which creative works. With fewer conversions, or with conversions skewed towards people who do not use blockers, they learn more slowly and less well. With duplicates the opposite happens: the platform believes that something works when it does not and spends more there.
As well as the ROAS of each dashboard, it is worth calculating a real store ROAS: total revenue from orders divided by total advertising spend. It does not tell you which campaign works, but it does not depend on any pixel and it gives you a reference for whether things are going better or worse overall.
What server-side tracking fixes and what it does not
| Cause | Does sending from the server fix it? |
|---|---|
| Blockers that stop the pixel from loading | Yes: the event does not go through the browser. |
| Customer who never comes back from the payment gateway | Yes: the purchase is sent when the order is recorded, not when a page loads. |
| Cookies cut short by the browser | Partly: storing the click identifier in the store helps to keep the link to the ad. |
| Pixel broken after a change in the store | Yes, and it is also detected: there is a delivery log that can be compared with the orders. |
| Duplicates | Yes, if browser and server share an event identifier. |
| Cookies rejected by the customer | No. Consent is respected on the server too. |
| Different attribution windows and models | No. These are each platform's own rules. |
| Purchase on another device without identification | Only when there is customer data the platform can use to match it. |
How to solve it in PrestaShop
RKR Relay receives the events from your PrestaShop store and delivers them to Meta, GA4, Google Ads, TikTok and other platforms from our server, with automatic retries if a platform fails and the status of every delivery. It works with the RKR API Conversions Pro module, which stores click identifiers (90 days by default), sends the same event identifier from browser and server so that each conversion counts once, and includes a daily reconciliation that compares orders with the deliveries to each platform: the check described in this article, done automatically every day. RKR Relay plans cost €19, €39 or €79 a month + VAT depending on event volume, with no minimum term.
If you would rather have everything set up in one go, the Pack Tracking that doesn't get lost brings together two modules, three hours of setup and one year of RKR Relay.
Frequently asked questions
Why doesn't Meta Ads record all my purchases?
Because the pixel depends on the customer's browser: blockers, cookies that are rejected or cut short and customers who never come back from the payment gateway mean that some purchase notifications never arrive.
Is it normal for Google Ads and the store not to match?
Yes, a small, stable difference is normal because of attribution windows and the date on which each sale is recorded. A large difference, or one that changes suddenly, points to a measurement problem.
How do I calculate my store's real ROAS?
Divide the revenue from all valid orders in the period by the total ad spend in the same period. It does not depend on the pixel and serves as a reference against the ROAS in each dashboard.
Does server-side tracking make the figures match completely?
No. It reduces the loss from technical causes and prevents duplicates, but each platform's attribution rules and consent decisions still make a difference.