ROAS is a lagging question. Contribution is the leading one.
Why the most quoted number in paid media should rarely be the number that makes your decisions.
The number everyone watches
Return on ad spend is easy to read and easy to report. That is exactly why it gets over-trusted. A platform-reported ROAS tells you how much revenue the platform believes it influenced, inside its own attribution window, under its own rules.
None of that is wrong. It is simply incomplete. Revenue is not margin, platform credit is not incrementality, and a single blended figure hides which customers you are actually paying to acquire.
Start from contribution margin
Before setting any ROAS target, work out what a sale is worth after product cost, fulfilment, payment fees, returns and discounts. That gives you the break-even ROAS: the point below which every additional order loses money on first purchase.
From there, decide deliberately how far above or below break-even you are willing to operate, based on repeat behaviour you can actually observe in your own data rather than optimistic lifetime value assumptions.
Separate the jobs
Branded search, retargeting and prospecting do different jobs and should not share a target. Blending them flatters the total and starves the channels that create new demand.
A disciplined account structure gives each job its own target and its own review cadence, so that budget moves towards new customers rather than towards the cheapest credited conversions.
What to do this quarter
Calculate break-even ROAS per product line. Split reporting between new and returning customers. Run at least one holdout or geo test on your largest channel. Then let those three inputs, not the dashboard default, set your targets.
Want to discuss how this applies to your company?
Next article
Fix acquisition before you scale it.