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 single biggest waste finding in this account was not a broken pixel or a bad bid setting. It was that more than half of the Meta budget was being spent advertising to customers the business already had. The account looked like it was working. Most of it was paying to reach people who were always going to come back, or not, regardless of the ad.
Commercial equipment retailer selling high-value kit on long, procurement-driven buying cycles.
Blended across search and social. Meta was the smaller slice, and the one carrying the worst attribution debt.
Over half of Meta spend served past buyers at roughly one ad a day. Frequency without purpose.
Account-wide ROAS dropped from 10.8 reported to 6.6 incremental once view-through credit was stripped out.
The pattern.
Meta was reporting an account-wide ROAS of around 10.8 on a 7-day click plus 1-day view setting. On the surface that is the kind of number that ends the conversation. Nobody questions a 10.8. The problem is that almost none of the diagnostic value lives at the account average. It lives in how that number breaks down by audience.
Strip the 1-day view credit and model it incrementally and the real account-wide figure was closer to 6.6. Still a healthy business. But the gap between 10.8 and 6.6 is the platform crediting itself for purchases that were going to happen anyway. In a B2B account that gap is wider than usual, because buyers here have already researched on Google and visited the site before a Meta ad ever reaches them. The ad gets the credit. It rarely did the work.
Why it was happening.
Two structural decisions were quietly funding the problem. The first was that existing customers had never been excluded from the prospecting layer, so over half the budget was being spent advertising to people who had already bought, at roughly one impression a day. In a B2B business with long reorder cycles, that frequency does nothing. These buyers come back when they need to replace or expand, not because they saw a carousel that morning. Email and direct sales are the right channels for that relationship, not paid frequency.
The second was that cold prospecting was being run through dynamic product ads. On an incremental basis that campaign returned 0.9, which means it was losing money. Every dollar in produced ninety cents of revenue that would have existed without it. Product carousels with prices work on warm audiences who already trust the brand. Pointed at a cold buyer planning a serious commercial purchase, they offer no context, no credibility, and no reason to act.
A 10.8 reported ROAS with cold prospecting at 0.9 incremental is not a strong account. It is a strong retargeting line item carrying a dead acquisition engine on its back.
What it was costing.
The commercial cost was straightforward. The half of the Meta budget sitting on existing customers, plus the cold DPA losing money on an incremental basis, was budget producing almost no new business. It was inflating the reported number while the part of the account that was supposed to find new buyers was either switched off or pointed at the wrong format. For a brand with a genuinely small addressable market, that is the most expensive mistake available, because new customer acquisition is the only lever that actually grows the business.
There was a second cost underneath it. Search was underused. For a high-intent B2B buyer typing procurement queries into Google, search is where the real demand declares itself, and the account was leaning on shopping and social while the highest-intent prospects went uncaptured. The Meta over-attribution was hiding how much of the genuine acquisition work search should have been doing.
When reported ROAS is high but cold incremental is near 1.0, the account is not acquiring. It is harvesting.
A flattering blended number almost always means the spend is concentrated on people who were already going to convert. The diagnostic is never the average. It is the spread between warm and cold, reported and incremental.
The fix.
- Move to 7-day click only. Drop the 1-day view credit so the account optimises against purchases the ads plausibly caused, not ones they happened to be near. Easier to do now than after budgets have scaled against the inflated read.
- Exclude existing customers from prospecting and redirect the budget. Cap any genuine retention activity at a low frequency and push the reclaimed spend, a meaningful share of the Meta budget, entirely into reaching new buyers.
- Turn off cold DPA and rebuild search. Stop running a retargeting format against cold audiences, and build out the search coverage that captures high-intent B2B procurement queries the account was missing.
None of this is exotic. It is account hygiene that had never been done, sitting underneath a reported number high enough that nobody had reason to look. That is exactly why it had survived.
What changed.
The reframe was the whole point. The brand was not failing. The read was. Once the budget stopped funding people it already had and the acquisition layer was pointed at genuinely new buyers through the right formats and through search, the spend started doing the job it was hired for. The headline number came down, because it was always partly fiction. The business underneath it got healthier, because the dollars were finally buying growth instead of crediting it.