Trading a target ROAS for contribution margin to quadruple a household brand's sales

ROAS looked strong almost everywhere: campaigns sat budget-capped while running 3.6–7.8× ROAS. These high ROAS figures look incredibly profitable, but it's actually a massive signal that profits are being left on the table. Since ROAS is just a ratio of ad spend to ad sales, it says nothing about what a sale is actually worth after COGS, referral fees, FBA, and storage. Nobody could tell whether pushing spend further would still be profitable once real costs were subtracted, so the only safe number was the one already being hit.

The gap was a measurement problem: a target ROAS was standing in for a profit number nobody had actually calculated.

We stopped measuring performance against a target ROAS. Instead, we built the brand's true contribution margin on Amazon like we do for all our clients (revenue minus COGS, referral fees, FBA, and storage, per ASIN) and turned it into a TACOS guardrail: the spend ceiling at the point where the next dollar of ad spend would push overall margin negative. That guardrail, not an arbitrary ROAS number, is the constraint our models at Laurence scale against.

From there it's just cold hard math and statistics: Laurence models the conversion rate of every keyword and product target as a probability distribution, prices each bid to buy incremental sales up to the margin guardrail, and re-runs every hour as the account grows. This means each new layer of budget is underwritten against fresh data, never set once and forgotten. That's why the curve above keeps climbing instead of plateauing once the easy wins are gone.

Daily sales doubled inside the first week: from roughly $2,800 to $5,600. TACOS never breached 20%, well within the blended contribution margin..

The real wins came after our data and strategy compounded over the next five months: total store revenue grew from about $2,800/day before launch to roughly $13,000/day, a 4× lift. Ad sales scaled 3.6×, and even as spend tripled to keep pace, TACoS never drifted: it sat near 14% the entire time, which was below the ceiling our contribution margin math had set at go-live. Total revenue grew faster than ad spend because the budget bought rank and velocity rather than just clicks, so organic compounded alongside paid. Holding a margin-derived guardrail for five months straight, through a 4× scale-up, is the proof that we measure the right things.