How Meta counts results
A Meta recap comes from Meta Ads Manager, whoever puts it together. The results in it are the ones Meta credits to its own ads, using Meta’s own rule for attribution. Meta can see what happened on Facebook and Instagram and, through its tracking on your site or app, some of what happened next. It can’t see your TikTok ads, your emails, or the creator video someone watched on YouTube.
Two settings shape every number in the recap. The first is the conversion window: how many days after someone clicked or saw an ad a result still counts. The second is whether views count at all. A view-through conversion is a result credited to an ad someone saw and never clicked. Meta reports often combine clicks and views into one number. Find the window in the report before you read anything else.
Meta also estimates some results it can’t observe directly, because phones and browsers share less than they used to. That’s a reasonable fix for missing data. It also means part of the number is a model, and the recap rarely says which part.
Where results get counted twice
Every ad platform credits itself, so the same buyer can land in more than one report. Meta is often the biggest single number in a recap, so it’s where double counting shows up first. Here’s how it happens on Meta:
- View-throughs. Facebook and Instagram reach a lot of people. Many of them buy for other reasons in the days after seeing an ad, and a view rule credits Meta with those results.
- Partnership ads. When paid spend goes behind a creator’s post, its results can show up in your paid-media report and your creator report. Two reports can count the same post. See creator whitelisting.
- Retargeting. Ads shown to people who already visited your site reach buyers close to a decision. Those results often belong to whatever brought them the first time.
- Your own channels. A buyer who opens your email and buys can still be credited to Meta if they saw an ad that week.
None of this means anyone did bad work. It means Meta’s number is Meta’s view, and it needs to sit next to every other channel’s.
A checklist for your next Meta recap
| Look at | What to look for | Why it matters |
|---|---|---|
| Window | The click and view windows behind the results | Longer windows reach further back for credit |
| Clicks vs. views | Results split by click and by view | View-throughs are the easiest results to overcount |
| Modeled results | Whether some results are estimated | Estimates should be labeled as estimates |
| Partnership ads | Creator posts with paid spend behind them | Two reports can count the same post |
| Total vs. your numbers | Meta’s results plus every other channel’s, against your own total | A total above what you sold means some results were counted twice |
| Reach | How many different people saw the ads | See reach vs. impressions |
Questions worth asking
Ask them of whoever runs the account, in-house or a partner.
- What are the click and view windows on this report, and can we see results at click only?
- How much of this number did Meta estimate instead of observe?
- Which results came from partnership ads, and do they also show up in our creator reports?
- How much of the spend went to people who had already visited our site?
- If we add Meta’s results to every other channel’s, does the total fit inside our own numbers?
Most of these can be answered from Ads Manager in a day. Write the answers next to the recap, so every month gets read by the same rules.
What a regions-on, regions-off test looks like on Meta
Lining up Meta’s results with your own numbers removes double counting. It doesn’t prove Meta caused anything. For that, pick regions that look alike, run Meta ads in some of them, pause them in the others for a few weeks, and compare sales in your own numbers. That’s a regions-on, regions-off test, also called a holdout test.
Meta’s targeting by location makes this easier than on most channels. The hard part is keeping everything else steady: no extra promotions in one group, no different creative, no surprise press. Agree on the length, and on what you’ll do with each outcome, before you start.
Meta also offers its own lift studies. They’re useful, and worth running. They’re also run by the company selling the ads, so read them next to a test measured on your own numbers. Our guide to incrementality walks through the steps.
Want to see what worked for your brand? Try it free: pick a question, and we answer it with your own numbers and what people say online. No card.