What is Target ROAS?

A store at a 38% contribution margin breaks even at 1 ÷ 0.38 = 2.63. To leave a third of the contribution for fixed costs and profit, divide rather than add: 2.63 ÷ (1 − 0.33) = about 3.9. A target of 8.0, chosen because one campaign reached it during a promotion, will not produce a return of 8.0 — it will produce very little spend, because the bidder declines almost every auction it does not expect to clear at that return.

Target return on ad spend tells a value-based bid strategy what ratio of conversion value to spend to aim for. The bidder estimates the value each impression is likely to produce and bids in proportion, aiming for the target as an average across the campaign.

Two things follow from that description and both are routinely missed.

The first is that it operates on the conversion value the account sends back, not on any value the platform can see independently. If the value is wrong, the strategy will pursue the wrong thing with great efficiency.

The second is that it is an average and not a floor. Individual conversions will return far above and far below the target. The strategy is working when the aggregate lands near it.

The derivation runs from the business outward, and it has two steps.

Break-even is the reciprocal of contribution margin. A store keeping 38 cents of contribution on a dollar of revenue breaks even at 1 ÷ 0.38, or about 2.63: below that, the campaign loses money on every order.

The working target then divides that threshold by one minus the share of contribution the business keeps for everything else — rent, salaries, tooling and profit. Keeping a third of the contribution takes 2.63 to 2.63 ÷ 0.67, or about 3.9. A business with high fixed costs and a modest media budget keeps a larger share and lands on a higher target; a lean business where media is most of the cost base keeps less.

Two sources that look authoritative and are not: a published benchmark, which describes companies whose margin structure is invisible, and the account's own best week, which was achieved under conditions that are not the normal ones. A target from either will spend most of the year unmeetable.

Return on ad spend and total contribution do not move together, and this is the most consequential thing to understand about the target.

Raising the target makes the bidder more selective. It enters fewer auctions, buys only the highest-expected-value impressions, and reports a higher return on less spend. Lowering it does the reverse. Across most of the useful range, a higher target means a higher ratio and less money.

The question a business actually cares about is where total contribution peaks, and that point is almost never at the highest achievable return. It is somewhere above break-even and well below the best ratio the account can produce. Finding it requires deliberately testing targets and measuring the contribution at each, rather than defending the ratio.

Accounts managed to a high ratio are usually under-spending and conceding volume to competitors who understood this.

A value-based strategy is only as good as the values it receives, and there are three quality levels.

A static value on every conversion is not a value strategy. It is target cost per acquisition wearing different units, because every conversion is worth the same to the bidder.

Revenue as the value is the common implementation and it is adequate where margins are even. Where they are not, it will find the highest-revenue orders regardless of what they contribute, which in a mixed catalogue means high-volume, low-margin products.

Profit as the value is the version that matches what the business wants. It requires product-level cost to reach the conversion event, usually via a server-side event or an enriched feed, and it changes what the bidder buys. Where the catalogue has real margin variation, this is the single highest-value change available to the account.

Every material change — the target, the budget, the creative set, the campaign structure — puts the strategy back into a state where its estimates are being rebuilt. Delivery is unstable during that period and the reported return is not representative of anything.

An account that changes the target in response to what it sees during learning will never leave learning. The discipline is to change one thing, wait at least the conversion window and preferably longer, then read the result.

Step sizes matter too. A change of ten or fifteen percent is absorbed; a change of fifty percent is a different campaign.

A single target applied to products with different margins does two wrong things simultaneously.

For the thin-margin range, the target sits below the true break-even, so the campaign buys orders that lose money. For the high-margin range, it sits well above break-even, so the campaign declines orders that would have been profitable.

Two remedies exist. The structural one is to split campaigns or product groups by margin band and give each its own target, which is straightforward in search and shopping and awkward in feed-driven campaigns. The better one is to send profit as the conversion value, which makes the split unnecessary because each order is already weighted by what it contributes.

The cheapest conversions are bought first, so achieved return falls as spend rises. This is a property of the market rather than a failure of the campaign.

Holding a high target while increasing budget therefore produces a campaign that cannot spend what it has been given. Reports describe it as limited; the limit was set deliberately by somebody who did not intend to set it.

Where growth is the objective, the target has to be walked down toward break-even as budget grows, with the floor set by what the business can afford rather than by what the account managed once. Knowing the break-even figure is what makes that walk safe.

• Derive the target from contribution margin, then divide by one minus the share of contribution the business keeps for fixed costs and profit. A target imported from an article describes a cost structure you cannot see.
• Send the best conversion value you have, and send profit rather than revenue where the catalogue has uneven margins. The bidder pursues whatever value it is given.
• Set different targets for different margin bands. One target across a mixed catalogue over-invests in the thin end and under-invests in the profitable end.
• Move the target in steps and wait out the learning period. Large changes reset the model and buy a week of unrepresentative delivery.
• Expect achieved return to fall as spend rises. Holding a high target while scaling is how a campaign becomes limited without anyone choosing to limit it.