Meta ads reporting shows the revenue Meta believes it caused, inside the attribution window you chose. That makes it useful for comparing ads against each other and unreliable as a verdict on the channel. The fix is not ignoring the report. It is knowing which decisions belong on Meta’s numbers and which belong on yours.

What is Meta’s report actually telling you?

Meta credits a purchase to an ad when someone clicked or viewed it within your window and then bought. It reports that revenue whether or not the person would have bought anyway, and whether or not another channel also played a part. Google, TikTok, your email platform and your affiliates are doing the same thing with many of the same orders. So the Meta figure is a real measurement of something narrower than it appears: the purchases Meta was in a position to observe. Inside Meta, that is perfectly workable. Comparing ad A with ad B, both measured the same way, tells you something true about which is performing better. The trouble begins when that figure is used to decide how much to spend on Meta in total, because the report has no way to see what would have happened if the ads had never run. That question needs data from outside the platform.

Which attribution setting should you use?

The right window depends on how your customers buy. I made this point in Triple Whale’s Moby Deep Dives session: an expensive product that people research for days needs a longer window and a multi-touch view, because the buyer sees several ads and visits the site several times before committing. An impulse product reads more truthfully on last click with a short window, because the decision happens in minutes and a longer window mostly gathers credit for purchases that were already coming. Choose the setting that matches your product, then test the choice rather than trusting it. My check is simple. Over the same weeks, does Meta’s reported ROAS rise and fall in step with the number of first time buyers your store records? If Meta’s number climbs while new customer counts stay flat, the window is claiming credit it has not earned, and the setting needs to change.

How do you check whether channels are double counting?

Overlap checks are where most of the useful insight sits. In a multi-touch tool such as Triple Whale, look at the journeys behind each channel’s conversions and see how often a purchase credited to one channel was touched by another first. In that same Moby Deep Dives session, I walked through a check where more than half of the customers won through Shop campaigns, 52% of them, had first come into contact with a Meta ad. That single number reshaped what each channel was worth and how much each one received. Without it, one channel was being paid for the other’s work. Anyone evaluating a Facebook ads agency for ecommerce should ask to see this kind of check run on their own data, because it shows whether a partner is managing the account or managing the report. Run it again whenever a new channel launches or budgets shift materially.

What should the weekly Meta readout contain?

A useful weekly readout answers four questions in order. Is Meta working, judged by blended new customer acquisition cost and marketing efficiency ratio week over week rather than by Meta’s ROAS? Which ads are winning, judged by Meta’s own ROAS and cost per purchase alongside how many viewers the opening holds, how many keep watching, and what share of buyers are new? How much should we spend, answered by the highest budget at which each order still clears its margin target while the business grows at the pace it wants? And is Meta collecting credit for sales another channel started, or losing credit it earned, answered by an overlap check before any budget moves? Keep the readout to one page. An owner should be able to read it in five minutes and know exactly what changes next week, and why. Anything that does not lead to a decision belongs in an appendix, or nowhere.

Choosing the right measure changes what you conclude. Take a womens fashion brand I work with: Facebook spend grew 56% to $829K across January to August 2026, while net sales for the whole business rose 105%, from $1.27M to $2.60M, against the same months of 2025. Meta’s report on its own could never have shown that relationship, because it only sees Meta.

The free 30 minute growth audit starts with your Meta reporting setup and attribution window, then tells you whether the numbers you act on are the right ones.