4 min readBy Marcel Sattler
Track Native Ads the Right Way
Check the pixel and the server event before you spend. A dashboard that disagrees with the shop is not a result. It is a broken test.
From the post
The platform pixel, the server event, and the shop's own order list are enough to start, including the Shopify and Taboola pieces if that is your stack.
, Marcel Sattler
↓ read on
If you launch before tracking works, the first 14 days teach you nothing. You will see clicks. You may see sales in the shop. You will not know which clicks became which sales, and you will "optimize" a story the numbers made up.
Do this after the advertorial has a URL and the store path can take a test order. Do it before the first campaign.
Two records, not one
The ad platform has a pixel in the browser. It tries to see the click and the purchase. Browsers block a lot of that. A pixel alone will undercount, and sometimes it will count the wrong thing.
The second record is a server event: your shop tells the platform, server to server, that an order happened. That message does not depend on the reader's browser finishing a script.
You want both. The pixel helps the platform learn. The server event is what you reconcile. Neither one is "the truth" until you have watched them fire on a purchase you made yourself.
A named tracker product is optional. The build does not depend on one. The platform pixel, the server event, and the shop's own order list are enough to start, including the Shopify and Taboola pieces if that is your stack. The longer setup notes are in native ads tracking setup and server-side tracking.
Watch it in the browser before any spend
Open the advertorial in a browser where you can see network requests, or use the platform's own helper if it has one. Then walk the path a buyer walks.
- Load the article. Confirm the base tag loads, and that it loads once.
- Click through to the product. Confirm you did not lose the click identifier on the way.
- Add to cart if you track that event. Confirm the event name is the one the campaign will optimize toward, not a name you made up in a different account.
- Place a real test order you can refund. Confirm the purchase event fires, with a value and an order id.
- Open the shop admin. Confirm the same order is there, with the same value.
Write down the time, the order id, and what each screen showed. That note is your baseline. When a week looks "dead," you compare against this, not against a memory.
If the purchase event does not fire, stop. Do not launch and "fix tracking on Monday." Monday's data will be a hole, and you will make decisions inside it.
Make the shop the referee
Once a day is enough at the start, and once is enough before launch. Export or simply read the orders in the shop. Compare them to what the ad platform claims.
You are looking for three kinds of disagreement:
- The shop has orders the platform never saw. Your server event or your pixel is missing.
- The platform reports purchases the shop does not have. You are counting something that is not a sale, or you are double-counting.
- The values do not match. A tax, a shipping line, or a currency is being sent differently on each side.
Pick one definition of "sale" and keep it. If you decide the shop's paid orders are the definition, say so in the note and stick to it. Switching definitions mid-test is how a loser looks like a winner on Thursday.
Fixing a Taboola and Shopify gap is the practical cousin of this page if that is the pair you run. Hyros, Tracify, and the rest is useful when you are choosing whether a third-party view is worth adding. It is not a requirement for the first campaign.
What you are allowed to optimize toward
On day one you may not have purchases. That is normal. The events you still need, even then, are the ones that show the path: article view, click from the article to the product, add to cart if you use it, and purchase when it happens.
Name them in a short list and put the list next to the campaign. In the first two weeks you will read the funnel, not a profit number. That reading is Your First 14 Days on Native Ads. Tracking is what makes the reading possible.
Do not add five extra events "just in case." Extra events you never check become noise. The list above is the list.
Done means you saw it once
Tracking is done when you have personally seen a test purchase in the browser event and in the shop, and you know which of those two you will believe if they ever disagree. It is not done because a pixel status dot is green.
Leave the campaign off until that is true.
The tracking build in Native Advertising Mastery for E-Commerce is this sequence with the checks written down, including the Shopify and Taboola side. What the course costs and who it is for is on the course FAQ.
The build, in order
FAQ
Is the platform's conversion column enough?
No. It is one report. The shop's paid orders are the other. If you only read the platform, you will scale a number you have not checked.
Do I need a third-party tracker to start?
No. Pixel, server event, and the shop admin are the start. Add another tool when you can explain what question it answers that those three do not.
What if my test purchase shows in the shop and nowhere else?
You are not ready to launch. The click is not tied to the order. Find the break, place another test order, and watch it land in both places before you spend.
▸ Keep reading
Three more on the same topic.

Native ads
▸ From the video
7 min read
Native Ads Tracking Setup: The Reliable Way to Track in 2026
Read article
Native ads
▸ From the video
6 min read
The Right Mindset for Native Ads: 3 Things You Need First (2026)
Read article
Taboola
▸ From the video
7 min read
How to Target the Right Audience on Outbrain & Taboola (2026)
Read article
