Client identifiers removed. Metrics directionally preserved.
This is a real BSD audit with the brand and exact figures anonymised. The pattern, the diagnostic logic, and the directional numbers are reported as they were found.
The account looked healthy. Meta was reporting a 2.1 ROAS in the home market, the team was reading that as a green light, and budget was being pushed accordingly. The P&L told a different story. True acquisition return was 1.07. The 2.1 was attribution, not performance.
That gap is not a rounding error. It is the difference between an account that is buying new customers profitably and one that is buying them at break-even. Every budget decision sitting on top of the 2.1 was being made on a number that did not exist. The first job of the audit was to take that number apart.
Men's grooming and skincare, multi-region, wide product range across deodorant, shampoo and hair loss.
Split roughly 60% Meta, 40% Google, running across its home market and two offshore markets.
Reported Meta ROAS fell to 1.29 on 7-day click and 1.2 incremental. The P&L showed 1.07.
Every campaign was running 7-day click plus 1-day view, crediting impressions nobody interacted with.
What 1-day view actually does.
The setting sounds technical and harmless. It is neither. With 1-day view switched on, Meta claims a conversion any time a user is served an ad and then purchases within 24 hours. Served, not clicked. A single pixel appearing on screen for under a second counts. The user never had to engage with the ad, never had to register it, never had to do anything other than scroll past it and buy something the next day for a reason that had nothing to do with paid media.
This brand had it running on every campaign. That is the default, which is exactly why it is dangerous. Most operators never change it, so they spend their entire history reading a credited number and calling it a performance number. The two are not the same, and the distance between them is widest precisely where it matters most: on existing customers, who Meta serves ads to constantly and then claims credit for whenever they repurchase.
The double count.
Here is the cascade in full. Reported Meta ROAS in the home market: 2.1. Strip 1-day view and read on 7-day click only: 1.29. Apply incremental attribution on top of that: 1.2. The P&L acquisition figure, the actual number, was 1.07. So the corrected platform read and the commercial reality finally agreed, and the original 2.1 was revealed as nearly double the truth.
Google was doing the same thing through a different mechanism. It was reporting 4.3, and even after stripping brand search, still 3.1, against that same 1.07. Meta was generating the demand through upper-funnel activity, the user later searched the brand by name, and Google booked the sale at the point of transaction. Both platforms were claiming the same customer. Add the two reported numbers together and the account looked like it was printing money. It was not.
If your reported ROAS is sitting at two times your P&L acquisition number, the gap is almost never performance. It is attribution claiming conversions it did not cause.
What it was costing.
An inflated number is not a reporting nuisance. It is a budgeting instrument, and it was pointing the money in the wrong direction. The three regions had genuinely different economics, but the blended, inflated read flattened all of that into one optimistic headline.
The offshore market with the strongest profile was sitting on healthy LTGP:CAC and above-benchmark one-year retention. That region should have been scaled aggressively from day one. The home market had been unprofitable on acquisition for almost its entire history. Reading at the account level on a 2.1, leadership had no reason to treat those two regions differently. They were funding the weak region and under-funding the strong one, because the inflated number gave them no signal to do otherwise.
The ROAS inversion: the campaign that looked better was 20% worse.
Live in this audit, two campaigns in one offshore market reported 1.44 and 1.23. The obvious move is to back the 1.44. On incremental attribution with a new-customer filter, they were actually 0.8 and 1.1. The second campaign was performing 20% better. Acting on reported ROAS would have actively worsened the account.
The fix.
- Remove 1-day view across every campaign, immediately. Move the whole account to 7-day click only. Review existing campaigns against the 7-day click column, not the default reported ROAS. Every creative and budget decision gets made on that basis from here.
- Judge regions on their own economics, never the blend. Scale the profitable offshore market now. Hold the weak home market to its real acquisition return. Stop averaging across regions for spend decisions, because the average is hiding both the opportunity and the risk.
- Validate Google with a geo holdout before reallocating a dollar. A 3.1 ROAS sitting on top of Meta-generated demand is not incremental until a holdout proves it. Start with a regional geo test in the break-even market, where the read is most actionable.
What changed.
Once the corrected numbers were on the table, the account stopped looking like a strong performer that needed more spend and started looking like what it was: a near break-even acquisition engine carried by genuinely strong retention, with one region worth scaling hard and another that needed watching, not feeding. None of that was visible at a reported 2.1. All of it was obvious at a true 1.07.
That is the whole point of pulling 1-day view first. It is not a measurement tidy-up. It reframes every other number in the account, and it changes where the next dollar goes. Leave it running and you optimise an entire account against a number that was never real.