Case study — Shopify
The Shopify account that was scaling itself out of profit
Blended ROAS said keep scaling. The curve said the account had crossed its ceiling months earlier. Here is how the analysis found the number, and what changed.
$[X]/mo
ad spend at the time of analysis
[X.X]
blended ROAS the dashboards showed
[X.X]
marginal ROAS at the ceiling — the number nobody was looking at
$[X]
monthly spend recovered past the ceiling
The situation
A Shopify brand doing $[X]/year, spending $[X]/month on Meta and Google. Every scaling attempt followed the same script: raise budgets, watch blended ROAS hold for a week, then watch contribution margin quietly fall. The founder’s read was creative fatigue. The dashboards agreed, because dashboards report averages.
Finding the curve
We rebuilt 90 days of spend and revenue into a marginal ROAS curve — the return on each next dollar, not the average of all dollars. The curve showed returns decaying from $[X]/month of spend and crossing 1.0 at roughly $[X]/month. Every dollar past that point was buying revenue that cost more than it returned. The account had been operating $[X]/month past its ceiling — profitably on the dashboard, unprofitably on the P&L.
The fixes
Three moves, in priority order from the waste map:
- Cut past-ceiling spend — [X] campaigns spending beyond their marginal break-even, cut or capped.
- Restructure toward marginal winners — budget shifted to the [X] ad sets whose marginal return still cleared 1.0 with headroom.
- Move the ceiling itself — [offer/funnel/creative changes from the real analysis].
The numbers
[X] weeks later: spend at $[X]/month — below the old level — with contribution margin up $[X]/month. The ceiling itself moved from $[X] to $[X] as the fixes landed. Scaling resumed, this time with the curve watched monthly.
The point is not that spending less made more money — sometimes it does, sometimes it doesn’t. The point is that nobody in the account knew where the ceiling was until it was plotted.