Meta's Ads Manager is not a neutral measurement tool. It is a reporting interface built by a company that earns revenue when you spend more on Meta. That is not an accusation. It is a structural fact that should inform how you read every number in that dashboard.
Meta's attribution system is designed to give Meta's ads the most favorable possible reading of their contribution. That does not mean Facebook ads do not work. It means you cannot use Facebook's reporting to determine whether they work.
Three signals determine whether Facebook ads are generating real business impact
What Meta's attribution actually measures
Meta's default attribution window is 7 days after a click and 1 day after an impression. Any conversion that happens within those windows, for any reason, gets credited to your campaign.
Consider a retargeting campaign. Your audience is everyone who visited your site in the last 30 days. These people already know your brand. They were already considering buying. On a typical e-commerce site, 2-5% of site visitors convert within any given week even without seeing ads. Meta will count every one of those conversions as ad-driven, as long as they happened within the attribution window.
This is not fraud. It is attribution logic applied to an audience that already had high purchase intent. The result is that retargeting campaigns almost always look excellent in Meta's dashboard, and that reported performance is almost always inflated relative to the actual incremental impact.
The three honest tests
There are three ways to evaluate Facebook ads without relying on Meta's own reporting. Each has trade-offs.
The MER test
Marketing Efficiency Ratio is total revenue divided by total ad spend, calculated from your backend data. No attribution, no platform windows.
Track your MER over four-week rolling windows. Then change your Meta spend meaningfully, either scale it up by 30-40% or cut it by a similar amount. Watch what happens to MER over the following three to four weeks. If MER improves when you scale and deteriorates when you cut, Meta is generating real incremental revenue. If MER barely moves, you have a problem the dashboard is not showing you.
This test is imprecise. Other factors affect MER: seasonality, email performance, organic traffic. But it is directionally reliable over multi-week windows, and it does not require running a formal experiment.
The holdout test
A holdout test is the most rigorous way to measure Facebook incrementality without needing a data scientist.
Divide your target audience into two groups before the campaign starts. Expose 80-85% of the audience to your ads as normal. Suppress ads entirely to the remaining 15-20%, your holdout group. Run for four weeks. Then compare backend conversion rates between the two groups.
If your exposed group converts at 4.2% and your holdout group converts at 3.5%, the incremental lift is 0.7 percentage points. About 17% of your exposed group's conversions are incremental. The rest would have happened without the ads.
Meta's Conversion Lift product runs a version of this automatically, suppressing ads to a randomized holdout. The limitation is that Meta controls the test design and reports the results, so you are still relying on Meta's methodology. Running your own holdout using a suppression list gives you more control.
The pause test and why it is tricky
Pausing Facebook and watching whether sales drop seems logical. It rarely works as expected.
When you pause campaigns, there is typically a carryover effect of one to three weeks where the pipeline built by the ads continues to convert. People who saw your ad last week but have not bought yet may still convert this week. A one-week pause will almost always understate Facebook's impact because you are measuring during the carryover window.
A longer pause, four to six weeks, gives cleaner data but costs you real sales while you wait. For most brands, a formal holdout test is a better design than an ad pause.
What good incrementality looks like for Facebook
Prospecting campaigns targeting cold audiences tend to show higher incrementality than retargeting. Cold audiences have low organic purchase intent, so a larger fraction of their conversions are actually caused by the ads. In holdout tests, prospecting campaigns typically show 60-80% incremental conversion rates.
Retargeting shows much lower incrementality, typically 20-50%, because retargeted audiences already have high organic intent. A Meta-reported ROAS of 8x on a retargeting campaign with 25% incrementality means the real incremental ROAS is closer to 2x. That may still be above your profitability threshold, but it looks very different from what the dashboard suggests.
The implication: evaluate prospecting and retargeting separately. They have fundamentally different incrementality profiles and should not be compared using the same ROAS benchmarks.
What to do with the results
If your MER moves predictably with Meta spend and your holdout tests show 50%+ incrementality on prospecting: Meta is working. Scale it.
If your MER is flat regardless of Meta spend and your holdout shows low incrementality on retargeting: you are probably spending more than you should on retargeting. Tighten the audience (cart abandoners and high-intent visitors only), shorten the window (7-14 days instead of 90), and reallocate the budget toward prospecting.
If you have never run either test: start with the MER analysis. It requires no campaign changes, just four weeks of consistent tracking and one deliberate spend change. The result will tell you more than six months of reading the Ads Manager dashboard.