In short
Ad waste, here, is the share of platform-credited sales your ads didn't cause, and the ad dollars that bought that credit. Platforms hand out credit inside an attribution window. They count shoppers who would have ordered anyway and miss paths they can't see, especially Meta ads that later become Amazon brand search. Subtract measured incremental sales from credited revenue, or apply a careful would-have-bought guess when you have no holdout yet. Divide that wasted sales figure by credited revenue, then multiply by spend. The result isn't a license to slash budgets from a guess: it's a planning signal that Meta ROAS and Amazon Attribution aren't the same as lift. Size a geo holdout before you cut or scale on the waste number alone.
What is ad waste for Meta → Amazon?
When Meta reports purchases, it's counting credit under its rules, not the sales you would lose if the ads stopped. For an Amazon brand that gap is often large: Meta can't see Amazon checkouts, and Amazon Attribution only sees tagged clicks inside a 14-day window. Both leave you with a ROAS that can look healthy while a share of the credited sales would have happened anyway.
This ad waste calculator turns that gap into dollars of spend. It asks for what you spent, what the platform credited, and either a holdout measurement of caused sales or your best guess of the would-have-bought share. The output is how much spend sat on non-incremental sales in that period.
How to calculate ad waste
caused sales = holdout incremental sales
(or credited × (1 − would-have-bought share))
wasted sales = max(0, credited − caused)
waste share = wasted sales ÷ credited
wasted spend = ad spend × waste share- Ad spend: Meta amount spent on the campaigns for the same dates.
- Credited revenue: Meta purchase value or Amazon Attribution product sales for those campaigns.
- Caused sales: incremental Amazon sales from a geo holdout over the same window. Prefer this over a guess.
If caused sales exceed credited sales, waste is zero on this definition: the platform under-credited you. That still matters for decisions, but it's the opposite of waste. Use the halo effect calculator for that case.
Worked example
A made-up brand spends $5,000 on Meta in a month. Meta credits $14,000 of purchases. A holdout later measures $8,000 of Amazon sales the ads caused. These are the calculator's example values.
- Wasted sales = $14,000 − $8,000 = $6,000.
- Waste share = $6,000 ÷ $14,000 = 42.9%.
- Wasted spend = $5,000 × 42.9% = $2,142.86.
- Credited ROAS is 2.80; incremental ROAS is 1.60.
On Meta's number the ads look strong. On the holdout, about $2,142.86 of the month's spend bought sales that weren't incremental. That's the planning signal, not an automatic cut.
What's a useful waste reading?
There's no universal "good" waste share. Categories with strong brand search and Subscribe & Save often show more credited sales that aren't incremental. New brands with little organic demand often show the opposite pattern: under-credit and a halo.
Treat a high waste share from a guess as a prompt to measure, not a prompt to pause. Treat a high waste share from a holdout as a reason to redesign creative, audience, or landing, and to re-check break-even with the break-even ROAS calculator against incremental ROAS, not Meta ROAS.
Common mistakes
- Cutting spend from a guess. A would-have-bought percentage from a podcast isn't your brand. Measure.
- Mixing date ranges. Spend, credited revenue and holdout results must cover the same period.
- Adding Meta and Amazon Attribution. The sum still counts credit. It can double-count and still miss lift.
- Ignoring under-credit. If caused sales beat credited sales, waste is zero here: you have a halo problem, not a waste problem.
How this connects to true incrementality
Waste is the flip side of incremental ROAS. Plan a test with the holdout test calculator or the brand size holdout calculator, then turn the result into iROAS with the incremental ROAS calculator. Compare every report at once with the attribution gap calculator.
Spillmark runs geo holdouts for Meta ads and measures Amazon orders by ship-to state, with a range on every lift number. See how lift is calculated.
Frequently asked questions
What is ad waste in this calculator?
The share of platform-credited sales a holdout (or your careful guess) says the ads didn't cause, and the ad spend that sits on that share.
Can I use Amazon Attribution instead of Meta purchase value?
Yes. Use whichever credit number you are making decisions from. The waste math is the same: credit minus cause, then apply the share to spend.
Is a high waste share always bad?
It means you're paying for credit that isn't lift. Sometimes that spend still supports brand search you care about; a holdout is how you know whether cutting it hurts Amazon sales.
What if I have no holdout number yet?
Enter a would-have-bought guess only for planning. Then size a geo holdout before you change budget from the result.
Does this work for Shopify as well as Amazon?
The formulas work for any credited vs caused pair. For Meta ads that sell on Amazon, a geo holdout on ship-to state is the usual way to get caused sales.