Target ROAS vs conversion target: which fits your business?

Google Ads offers a menu of automated bidding targets, and two of the most common get chosen more by habit than by fit: target ROAS and target CPA, or its simpler cousin, Maximize Conversions. Both hand the moment-to-moment bid decision to the algorithm. They optimize for genuinely different things, and picking the wrong one for your business means the algorithm works hard, correctly, toward the wrong number.

What is the actual difference between the two?

Return on ad spend, or ROAS, is revenue divided by ad spend, usually shown as a percentage: five dollars of revenue for every one spent is a 500% ROAS. A target ROAS strategy bids to hit a revenue-per-dollar goal, which means it needs to know the value of each conversion, not just that a conversion happened.

A target CPA strategy bids to hit a cost-per-conversion goal instead, treating every conversion as equally valuable. Maximize Conversions is the simpler relative: rather than a set cost target, it spends the available budget to return as many conversions as it can find. Both of those optimize for count. The dividing line between the two families is whether your conversions actually differ in value: ROAS answers to revenue, CPA and Maximize Conversions answer to volume.

How do the two compare, side by side?

Which business fits a target ROAS strategy?

A business where order value genuinely varies, such as a retailer selling both a low-cost accessory and a high-cost centerpiece from the same catalog. Revenue, not raw conversion count, is the real goal in that setup, since ten cheap sales and two expensive ones can produce the same order count with very different returns. ROAS is the target built to tell those apart, but only if the order values reaching Google are accurate; a target chasing a number it cannot trust will chase it confidently anyway.

Which business fits a conversion or CPA target?

A business where each conversion is roughly interchangeable at the ad level, even if it is not interchangeable later. Lead generation is the clearest case: a form fill is a form fill to the campaign, even though one lead will close and another will not, because that difference gets decided in the sales process, not at the moment of the click. Subscription signups and service bookings usually fit the same pattern. A CPA or Maximize Conversions target fits because there is no per-conversion value worth feeding the algorithm yet.

What does it cost to switch between them?

Automated bidding builds its judgment from the recent pattern of conversions under the current target. Change the target and that pattern's meaning changes with it: a click that looked like a good bid under a conversion-count goal may not look like one under a revenue goal, so the system has to relearn which signals predict success under the new instructions. Expect a stretch of less predictable performance right after any switch, not because something broke, but because the model is rebuilding its judgment from scratch.

That stretch interacts badly with a switch made right before a spend-critical period, since unstable bidding is the last thing you want during a week where hitting the monthly number matters most; see why the calendar matters as much as the target when you are timing a change.

How do you decide, step by step?

1. Ask whether your conversions carry meaningfully different value. If a forty-dollar sale and a nine-hundred-dollar sale both come through the same campaign, that difference matters and points toward ROAS.
2. Check whether that value reaches Google accurately. A ROAS strategy fed wrong or missing order values will optimize confidently toward the wrong number.
3. Count recent conversion volume under the current strategy. A switch needs enough history behind it to relearn from something, rather than starting cold.
4. Weigh the calendar. Avoid switching right before a period where stable delivery matters most, such as the final week of a spend-critical month.
5. Set a review window before judging the switch. Hold off on a second change until that window has actually passed, or you will be judging noise instead of the target.

Picture a mid-size home goods retailer selling a forty-dollar candle and a nine-hundred-dollar sofa from the same campaign. A conversion-count target treats both sales as identical wins, so it might chase a flood of cheap candle sales while a shopper behaving like a sofa buyer loses the auction to a cheaper bid elsewhere. A target ROAS strategy, fed the real order value, bids more for the visit that looks like a sofa sale and less for the one that looks like a candle browse. The retailer's total conversion count might even fall after the switch while total revenue rises, because the strategy is now chasing the number that was the actual goal all along.

Neither target is the safe default. The safe default is whichever one matches how your conversions actually differ, or do not. A target that is also held down by a budget cap will not reach its own goal no matter how well it is set, so check whether a budget cap is holding a good campaign back before blaming the target itself. And when Google's own panel recommends a target switch, it deserves the same scrutiny as any other suggestion; see when to decline a Google Ads recommendation. Programmatic CMO's Google Ads agent tracks a campaign through the settling period after a target change and reports honestly on whether the new target is actually working, not just whether it is still running.