What is target ROAS bidding?

Target return on ad spend bidding works from a prediction. For each auction the system estimates how likely a conversion is and how much that conversion is likely to be worth, then bids an amount that should produce your stated ratio on average.

That means the bid is not fixed. Two clicks on the same product can cost very different amounts, because one query, device or moment looked more promising than the other. A search that names your exact product may attract a high bid. A vague browsing query may attract almost nothing.

For this to function, conversions have to be sent with values. A conversion recorded without revenue tells the system nothing about worth, and a value based strategy has nothing to work with.

It also means the strategy is only as good as the value you send. If every order reports its full revenue regardless of margin, the system optimises for revenue, and revenue is not profit. Stores with widely varying margins often send a margin adjusted value instead, so that the optimisation chases the money they actually keep.

The important mental shift is that you have stopped managing prices and started managing an objective.

People expect the target to act like a floor. It does not. It is the average the system aims for across the whole campaign over time, which means some conversions will come in far above it and some far below.

This matters for two practical reasons.

First, a campaign hitting its target exactly can still contain products that lose money and products that make a great deal. If you look only at the campaign average you will never see that. Check the product level report and decide whether the losers belong in the campaign at all.

Second, short windows are meaningless. A day, or even a week at modest volume, will sit well above or below the target through normal variation. Judging the strategy on a few days leads to constant target changes, and constant target changes keep the campaign in a permanent learning state where it performs at its worst.

So read it monthly, or at least over enough conversions that a handful of large orders cannot swing the figure. And when you do change the target, move it in small steps rather than large ones, because a big change is treated as a new objective.

Two numbers decide the right target and you need both.

The first is break-even, which comes from your gross margin. Take price less cost of goods, payment fees, packaging and any shipping you absorb, express that as a share of revenue, and divide one by it. That is the ratio at which advertising pays for itself and nothing more. Anything below it loses money on every order.

The second is what the campaign currently achieves. That is your starting point, because setting a target far above current performance does not produce better results, it produces less delivery. The system reaches an ambitious target by bidding on fewer, safer auctions, which usually means retargeting your existing demand and abandoning growth.

So start at or near the achieved figure, confirm the campaign is stable, then tighten gradually if you want more efficiency, or loosen if you want more volume. Each move should be modest, with a few weeks between changes.

And revisit break-even whenever costs change. A supplier increase or a rise in shipping rates moves the floor, and a target set against last year's margin can quietly become unprofitable.

The two settings interact in a way that catches people out. The target controls efficiency. The budget controls scale. Set them in conflict and the campaign behaves strangely.

A strict target with a large budget produces underspend. The system cannot find enough auctions it believes will hit the ratio, so it leaves money unspent. That looks like a delivery problem and it is actually a target problem.

A loose target with a small budget produces a campaign that hits its budget early in the day and stops. That is not necessarily bad, but it means you are choosing volume over reach and probably missing the afternoon.

The way to think about it is that lowering the target buys volume at a lower efficiency, and raising it buys efficiency at a lower volume. Neither is automatically right. The question is which combination produces the most gross profit after ad spend, and that is worth testing deliberately by moving the target in one direction for a month and comparing total profit, not the ratio.

If you want more revenue and the campaign is already spending its budget, raise the budget. If it is not spending its budget, the target is the constraint.

Setting it on day one. A new campaign has no conversion history, so the system has nothing to predict from. Start with a cost based approach, gather several weeks of conversions, then switch.

Changing the target every few days. Each change restarts learning, and a campaign that never leaves learning performs worse than one left alone at an imperfect target.

Mixing margins in one campaign. If a campaign contains items earning very different margins, one target cannot serve them. Split them, or adjust the conversion values you send.

Broken or duplicated conversion tracking. Duplicated purchases inflate reported value, the system thinks it is beating the target easily, and it bids up on traffic that is not really performing.

Ignoring the conversion window. If your buyers take a week to decide, a campaign judged on same day results will look worse than it is, and tightening the target in response makes it genuinely worse.

And forgetting the feed. The strategy decides the bid. What gets shown, and for which searches, still comes from your product data and your product selection. A poor feed under a good bid strategy is still a poor campaign.