Target ROAS in Google Ads: picking a number that doesn't stall the account

Target ROAS is an instruction: "spend whatever it takes as long as the return stays at this level." It sounds like a profitability guarantee and behaves more like a volume regulator, and confusing the two is the most expensive mistake in the whole setup.

What it actually does

With Target ROAS, Google estimates for each auction how much revenue it expects from that click. If the ratio between that estimated revenue and the cost meets your target, it bids. If not, it doesn't.

The direct consequence is that the target doesn't force more profitability: it filters opportunities. A high target doesn't make conversions worth more — it makes the campaign enter fewer auctions.

That's why raising the target almost always reduces spend. It isn't a misconfiguration, it's exactly what was asked for.

Choosing the starting number

The rule that prevents most problems: start from the return the campaign is already delivering, not the one you'd like.

If the last ninety days show a three-to-one return, start there. The strategy begins with an achievable target, accumulates data and stabilizes. From that point you can raise it gradually and watch what happens to volume.

Starting at six when the history says three produces the predictable result: spend collapses, the campaign barely delivers, and someone concludes automation doesn't work.

The trade-off nobody states

There's a tension worth putting on the table before choosing: return and volume move in opposite directions.

A high target gives better return on less revenue. A low target gives more revenue at worse return. There's no point where both rise at once, unless something structural improves — page conversion, margin, the product itself.

So the business question isn't "what's the best ROAS" but "do we want more profit per sale or more sales." The answer depends on the company's stage and almost never belongs to the media team.

The volume requirement

Like every automated strategy, it needs data. With few monthly conversions, the model can't estimate expected revenue per click with reasonable precision.

The symptom of missing volume is inconsistency: very good weeks followed by very bad ones, with no identifiable cause. That isn't the market moving, it's the model guessing.

In that scenario, Maximize Conversions or even well-managed manual CPC usually performs better, until volume justifies the switch.

Conversion value has to be real

One technical detail that invalidates everything above if it's missing: Target ROAS needs every conversion to carry a value.

In ecommerce that usually comes resolved by the integration. In lead generation it doesn't — and that's where the problem shows up: if every lead carries the same assigned value, the strategy optimizes quantity, not quality, and the calculated ROAS describes nothing real.

The honest fix there is to assign different values by lead type or, better, to send the conversion from the CRM with the real sale value. It's integration work, and without it the strategy optimizes a fiction.

Different targets per campaign

An adjustment that pays and rarely gets made: not every campaign deserves the same target.

Brand campaigns — people searching your name — have naturally very high return. Giving them the same target as prospecting wastes margin: they could sustain a much stricter target without losing anything.

Conversely, discovery campaigns aimed at people who don't know the category yet will never hit brand-level return. Demanding it just switches them off.

One target per campaign, calibrated against each one's history, produces a considerably more profitable account than a single number applied to everything. It's more maintenance work, and it pays for itself.

The attribution window trap

A detail that distorts the read and creates pointless arguments: the return you see today for last week is still going to change.

Conversions that happen days after the click get credited retroactively to the click date. For slow-decision products, a week's ROAS can improve substantially up to two or three weeks later.

That has two practical consequences. First, don't judge an adjustment on data that's too fresh. Second, don't compare the current week against a closed one — the running week will always look worse, and it isn't.

How often to adjust

The strategy needs stability. Changing the target daily prevents it from finding its footing, because each meaningful adjustment resets part of the model's learning.

A reasonable rhythm: moderate changes, with two weeks of observation between them. And always evaluate over the full period, not the first days — which are usually the worst.

If you need to correct quickly because spend ran away, it's better to use the daily budget as a brake and leave the target alone.

When not to use it

Three situations where another strategy pays more.

New campaigns with no history. There's nothing to estimate from. Start with Maximize Conversions and migrate once there's data.

Strong seasonality. In atypical weeks, a model trained on normal behavior gets it wrong in both directions.

Goals that aren't revenue. If what matters is qualified lead volume rather than direct revenue, ROAS is the wrong metric to govern bidding.

Raising the target without losing volume

There is one way to get both sides of the trade-off, and it doesn't happen inside Google Ads.

If page conversion rate improves, or average order value rises, or margin improves, the same click is worth more — and the same target now qualifies more auctions. The campaign can spend more and return more at the same time.

That's the reason the most profitable accounts aren't usually the ones with the cleverest bidding setup. They're the ones where someone worked on what happens after the click, and let the bidding strategy inherit the improvement.

Worth remembering whenever a target adjustment is being debated for the third week in a row: the lever may not be in that field.

What to watch to know it's working

Three numbers together, never one alone: effective ROAS, total revenue and spend.

ROAS alone misleads, because you can always improve it by spending less. A campaign that went from three to five while revenue halved didn't improve — it shrank.

The combination that confirms the adjustment worked is stable or better return with equal or higher revenue. Any other combination deserves a conversation before it gets celebrated.

If the cost of the click is the symptom, see high CPC on Google Ads. For the most common source of wasted spend, see Google Ads search terms. And for the weekly read without exporting, see the paid traffic report.