Meta & Facebook Ads
4 min read
Your top ROAS product is probably not your most profitable product. A 3.5x on a thin margin SKU can lose money on every order while a 2.1x on a fat margin one quietly prints, and revenue ROAS reports both as if the first is winning.
The free Claude skill product-winner-finder ranks products by what they actually contribute per ad dollar, using your real margins instead of the account average.
This post covers the 4 classifications it uses, how it maps spend to products, and why the fake winner is the most expensive item in most accounts.
Key takeaways
Products are ranked by contribution ROAS, which breaks even at 1.0, rather than by revenue ROAS.
Fake winners are flagged: high revenue ROAS, thin margin, quietly below breakeven once real costs land.
Sleepers get surfaced too, the modest ROAS and fat margin products that are underfunded because they never look impressive.
Every ranking states which attribution rung produced it, so you know how much to trust it.
The math that changes the ranking
contribution ROAS = (net sales x CM%) ÷ ad spend breakeven at 1.0
margin per ad dollar = contribution ROAS − 1
That's the whole idea. Revenue ROAS asks how much money came in per ad dollar. Contribution ROAS asks how much money you kept, which is the only version that pays salaries.
Because it needs your margin percentage per product, this skill depends on the margin truth calculator having run first. Without real margins, ranking by contribution is just ranking by revenue with extra steps.
4 classifications
Class | What it means |
|---|---|
True winner | Contribution ROAS at 1.2 or above with real volume behind it |
Fake winner | Revenue ROAS clears your store blended breakeven, but contribution ROAS is under 1.0 |
Sleeper | Modest revenue ROAS, contribution ROAS at 1.2 or above, underfunded |
Loser | Below 1.0 on both measures |
The fake winner is the one worth hunting. It's usually the flagship: high price, heavy discounting or thin margin, and a ROAS that looks respectable against the number you carry in your head. It survives every review because nobody checks it against its own margin, only against the store average.
The sleeper is the opposite failure. Nothing about it demands attention, so it never gets more budget, and it keeps returning more per dollar than the products that do.
The attribution ladder
Mapping ad spend to individual products is genuinely hard, so the skill uses the highest rung available and tells you which one it used.
The best rung is a naming convention, where campaigns are named by product so spend maps directly. Second is catalog ads with Meta's product level breakdowns. The last resort is a channel blend, where total spend is allocated by each product's share of attributed sales. Anything from that bottom rung gets labelled as blended, because it hides cross product differences and the rankings are directional only.
That labelling matters. A fake winner verdict built on blended attribution is a strong hint, not a final answer, and the skill will say so rather than let you defund a product on weak evidence.
Move the budget, then feed the sleepers
Reallocating toward true winners and sleepers only works if those products have creative to spend it on. Vibelets generates on-brand product ads from your store in minutes, no designer or brief required.
How to set it up in Claude
Download the skill file from the skill's Google Drive folder, free, no email gate, then add it to Claude by uploading it as a skill under Settings or attaching it to a project or chat. Anthropic's Help Center guide to skills covers the upload path.
Run the margin truth calculator first so per product margins exist. Then paste an Ads Manager export plus a sales by product report covering the same 14 to 30 day window. Matching date ranges matters more than the length of the window.
Conclusion
Revenue ROAS tells you what sold. Only margin tells you what earned. Rank on contribution, move budget from the fake winners toward the sleepers in steps of 20% or less, and let the account average stop making decisions for you.
Rank your products by profit, not vanity ROAS
Free Claude skill, yours in 1 click.
Frequently asked questions
Why can a high ROAS product still lose money?
Because revenue ROAS ignores cost of goods, shipping, fees, and returns. A 3.5x on a 12% margin SKU returns less per ad dollar than a 2.1x on a 55% margin one. Only contribution ROAS reflects what you keep.
What is contribution ROAS?
Net sales multiplied by contribution margin percentage, divided by ad spend. It breaks even at 1.0, so anything above that is profit per ad dollar and anything below is loss.
What is a fake winner?
A product whose revenue ROAS clears the store blended breakeven people carry in their heads, but whose contribution ROAS is below 1.0. It looks fine on the account average and quietly loses money on every order.
How does the skill know which spend belongs to which product?
It uses the highest available attribution rung: a product naming convention on campaigns, then catalog ad breakdowns, then a channel blend as last resort. Blended results are labelled as directional rather than final.
Do I need to run another skill before this one?
Yes, the margin truth calculator, because this ranking needs real per product contribution margins. Without them you are ranking by revenue again with extra steps.
Sources
Make ads that sell
your Products.
Generate on-brand product ads in minutes, no designer or brief required.
Trusted by 400+ growing brands