Why does ROAS look good but profit is flat?

A homeware brand we audited last month was running a reported 4.2× blended ROAS and reporting it as a win to the board. Twelve months of revenue growth, flat contribution profit. The pattern repeats across catalogues because ROAS hides five failure modes.

First: returns. A 4× ROAS on a category with 30% return rate is effectively 2.8× on goods delivered. If the platform reports gross orders rather than net, the number is a fiction at the margin level.

Second: discount load. Each discount code, sitewide sale and bundle offer compresses the margin that ROAS does not see. A brand running 25% blended discount is silently giving away a quarter of the headline ROAS.

Third: mix shift. PMax and Shopping bid on whichever SKUs convert cheapest. Those are often low-AOV, low-margin SKUs. Revenue goes up, mix shifts down, contribution profit holds or drops while ROAS looks fine.

Fourth: brand cannibalisation. PMax often eats brand search traffic that would have converted organically. That spend shows up as 'incremental' in PMax reporting but is mostly substitution. Revenue is the same, costs are higher.

Fifth: attribution. Last-click and data-driven attribution both tend to over-credit paid touchpoints. The lift Google reports is often higher than the incrementality test would show.