Budget-limited campaigns with target ROAS: what changed in August 2026

The target ROAS you set is now the price at which Google buys revenue

In a budget-limited campaign, the target could be wrong for a long time without anyone noticing, because actual ROAS was higher and paid the bill. According to Google, the change rolled out from 17 August 2026 and was complete globally on 27 August. Since then, campaigns with the status "Limited by budget" and a target ROAS or target CPA have optimised more consistently towards the value you entered. In Demand Gen this also applies to target CPC.

Search, Shopping, Performance Max, Demand Gen and Travel are affected. According to Google, campaigns with sufficient budget do not change their behaviour. Google does not touch targets or budgets, so the decision stays with you. Google's own example works with CPA: a target of USD 10, an actual figure of USD 5, and the campaign now moves towards USD 10. With ROAS the direction is the same, only downwards.

Spend stays the same, profit does not

Google offers reassurance with the statement that the change will not directly increase spend. That refers to the money going out, not to what comes back for it. The same source concedes that campaigns which previously beat their targets may be affected if you do nothing.

Worked example with invented figures: a shop spends €10,000 a month in a budget-limited campaign, with a target ROAS of 3, an actual ROAS of 5 and a contribution margin before advertising of 40% of revenue. The break-even ROAS is therefore 2.5.

The right-hand column is the full swing, not a forecast for every account. But it shows why the problem is easy to miss: ROAS 3 is above break-even, the campaign is meeting its target, and nothing in the account is red. Yet €8,000 a month is missing.

There are no reliable effect sizes. The only early signal I would cite comes from Amsive: in the first week, ROAS for the retail target ROAS campaigns it looked at fell by around 29 to 35%. It is a small group, and the agency itself calls it an early signal, not a benchmark.

Why campaigns used to beat their targets

Google does not name a cause, only its reasoning: the old behaviour was confusing and produced unpredictable results when budgets changed. Mike Ryan of Smarter Ecommerce offers an explanation: Google lowered bids in budget-limited campaigns so as not to exceed the budget. That is his reading, not a confirmation from Google.

My assessment: for many retailers, beating the target was an unnoticed part of their margin. The budget cap doubled as an efficiency safeguard, and nobody ever had to check the target against the margin. Ginny Marvin, Ads Liaison at Google, describes the new division of roles clearly: the target is the lever for efficiency and spend control, while the Maximise strategies are meant for fixed budgets.

The adjustment tool locks in the old side effect

Google's bid target adjustment tool shows historical performance and sets the target to that value with "Apply". It preserves the status quo and does not answer whether that is commercially right. In addition, a recommendation is missing below 7 conversions and only appears for campaigns that recent data shows are likely to be affected. No recommendation therefore does not mean not affected.

I would therefore decide by situation, not by tool. The yardstick is contribution margin in euros, and break-even ROAS is the floor, not a target, because that is exactly where nothing is left after advertising. In the table, actual ROAS means the figure before 17 August, as today's may already have fallen.

The example from above shows why the third row needs a calculation. If the shop doubles the budget to €20,000 and then runs at ROAS 3, it makes €60,000 of revenue and €24,000 of contribution margin before advertising. After advertising, €4,000 is left instead of the previous €10,000, with more revenue and a target above break-even. Whether a lower target pays off depends on the incremental ROAS of the next step, and the best way to estimate it is with the bid simulator and a limited test.

The table does not apply if you deliberately buy below short-term contribution margin, for example for new customers with proven repeat revenue. In that case you calculate with different values.

In Black Week, both routes lead to the target

The usual checklist looks at today's status. That falls short. When demand rises, campaigns that had no budget limit before can run into one, and in the peak exactly the value you entered then applies to them. In the past, the cap would have protected efficiency as a side effect. Limited by budget is therefore not a fixed state of a campaign, it depends on the week.

If instead you follow Google's recommendation and set the budget at three times the expected spend, the cap usually no longer binds, and the campaign scales along the target. Google also earns from this recommendation, which is worth bearing in mind. Both routes end in the same place: in that week, the target is the only efficiency control left.

On top of that, the discount shifts the yardstick. In the worked example, the goods sell for €100 net with €40 of contribution margin, break-even 2.5. With a 20% discount the value is €80, the contribution margin €20, break-even 4. A target of 3 then brings €300 of revenue and €75 of contribution margin per €100 of ad spend, which is a loss of €25. Before August, a budget-limited campaign might have masked this. I would no longer rely on that. The calculation for the discount itself is in the article on preparing for Black Week.

For target changes ahead of promotions, Google advises a gradual approach, for example in 30% steps with one to two conversion cycles in between. Going from 3 to 4 is already a full step, so start early enough. According to Google, seasonality adjustments are meant for events of 1 to 7 days and large expected jumps in conversion rate, and whether they are worthwhile for you is covered in the article on seasonality bid adjustments.

The limit of this reasoning: it is my conclusion from Google's rule, not a measured observation. For retailers whose campaigns have ample budget headroom even in the peak, little changes.

Campaign list before recommendations

1. List all campaigns with a target ROAS or target CPA whose budget is limited today or would be tight at expected peak demand. Work from your own list, not from the recommendations in the account.
2. Enter three figures per campaign: the target you set, actual ROAS before 17 August, and actual ROAS since 27 August. Seasonal effects distort the comparison, so only read the direction.
3. Calculate break-even per campaign, using the margin mix it actually sells, and once more with the promotional margin.
4. Decide using the table above. For portfolio strategies and shared budgets, this is only possible at that level, and with a limited shared budget the effect is spread across all campaigns in the group.
5. Evaluate after one to two conversion cycles at the earliest, which is also what Google advises. Adjusting earlier means reacting to incomplete data.

This assessment is as of November 2026. According to Google, campaign total budgets behave as before, so they do not need this check on this occasion.