Target ROAS vs Profit Bidding: Why Your Best Campaign Might Be Losing Money
ROAS targets treat every euro of revenue as equal. Your accountant does not. Here is how to move an account from revenue bidding to profit bidding without blowing up volume.
The flaw hiding inside ROAS
Return on ad spend divides revenue by cost. It is simple, universal, and blind to the one number that decides whether your business survives: margin.
Two campaigns both at 4.0 ROAS can have completely different outcomes. One sells accessories at 65% margin and prints money. The other sells discounted hardware at 8% margin and loses cash on every order once shipping and returns are counted.
Profit bidding fixes this by sending gross profit, not revenue, as the conversion value to the bidding algorithm.
What “profit” should include
The value you send back should be revenue minus:
- Cost of goods sold, at the SKU level.
- Shipping cost, including any subsidy on free-shipping thresholds.
- Payment processing fees.
- Expected return rate, applied per product category.
- Any per-order packaging or fulfilment cost.
What it should not include: fixed overhead, salaries, or the ad spend itself. You are giving the algorithm a contribution margin, not a net profit line.
The tooling
For most ecommerce stacks, ProfitMetrics or an equivalent handles the calculation and forwards the adjusted value through server-side tracking to Google Ads, Meta, and GA4. For custom stacks, the same result comes from computing margin in the order confirmation payload and sending it as the conversion value via the Conversions API and offline conversion imports.
Either way, the requirement is the same: accurate cost data per SKU, kept current.
What happens when you switch
Expect three things, in order.
Reported ROAS collapses. If your average margin is 35%, a 4.0 revenue ROAS becomes roughly a 1.4 profit ROAS. Nothing has changed in the business. The number is simply honest now.
Budget reallocates. Within two to three weeks, spend drifts toward high-margin products and away from the loss leaders that previously looked like winners.
Volume dips, then profit rises. Order count usually falls slightly. Contribution profit typically rises, and in the accounts we have migrated, the gain has been in the double digits.
Brief your stakeholders before the switch, not after. A collapsing ROAS chart with no context ends good strategies.
Setting the new target
Do not translate your old target with a formula. Run in maximize conversion value with no target for two to three weeks, gather profit-based data, then set target ROAS at roughly the achieved level and adjust by 5% to 10% increments.
A profit ROAS of 1.0 means you break even on contribution margin before overhead. Most businesses need somewhere between 1.3 and 2.5 depending on fixed cost base and growth appetite.
When ROAS is still the right metric
Profit bidding is not universally correct. Stay on revenue-based values if:
- Your margins are near-identical across the catalogue, in which case profit bidding just rescales the same decisions.
- You are deliberately buying first orders below margin to win subscription or repeat revenue. In that case, send predicted lifetime value instead.
- Your cost data is unreliable. Bad margin data is worse than no margin data, because the algorithm will act on it confidently.
The lead generation version
For lead gen, the equivalent is sending different values for different lead qualities through offline conversion imports: a demo request booked is worth more than a newsletter signup, and a closed deal is worth more than both. The principle is identical. Tell the algorithm what an outcome is worth to you, and it will go buy more of the valuable ones.
Measure what you bank, not what you invoice.