What changed on 17 August
From 17 August 2026, Google changed how target-based bid strategies behave when a campaign is limited by budget. Target CPA, Target ROAS, and Target CPC for Demand Gen now optimise toward the target you configured, rather than toward the performance level the campaign had been delivering.
That sentence is easy to read past. It is the most consequential thing Google has shipped to bidding this year, and it takes effect without you touching anything.
Why your numbers move even though you changed nothing
A budget-limited campaign is one that would spend more if you let it. Google throttles delivery to fit the budget, and in doing so it has historically kept the campaign near the efficiency it was already achieving.
Here is the part that matters. Plenty of accounts are running a target that no longer reflects reality. Someone set a £40 target CPA eighteen months ago, the account improved, and it has been delivering at £28 ever since. The budget cap was doing the work of holding it there.
With this change, the £40 you typed in becomes the number the system aims at. The campaign now has explicit permission to spend up to that target, because that is what you told it you would accept. Nothing is broken. It is doing exactly what you configured — for the first time.
The same logic runs the other way. A target tighter than your actual performance will now pull volume down rather than let the campaign coast.
Who is affected
Any campaign that is both using a target-based strategy and limited by budget. Campaigns that are not budget-constrained were already being governed by the target, so there is far less to see there.
In practice this hits smaller accounts and agency-managed accounts hardest, because those are the ones running deliberately capped budgets across many campaigns. If you manage twenty client accounts with monthly budget ceilings, this touches most of your book at once.
The check to run today
For every budget-limited campaign, compare the last 30 days of actual CPA or ROAS against the target set in its bid strategy. The gap between those two numbers is the distance your results are now free to travel.
- Actual much better than target — the highest-risk case. Costs can rise toward the target. This is the one nobody notices until the monthly report.
- Actual roughly equal to target — little should change. The target was already binding.
- Actual worse than target — expect the system to buy less and chase efficiency. Volume may fall.
Sort by spend and start at the top. A 40% CPA gap on your largest campaign matters more than a perfect audit of the small ones.
What to do about it
The instinct is to leave it alone and watch. That is reasonable for a week. It is not reasonable for a quarter, because the drift is gradual and looks like ordinary noise until it has compounded.
If you want to keep the performance you have been getting, reset the target to the performance you have been getting. Set the target CPA to something close to your genuine recent average rather than the aspirational number from a year ago. The target is now a statement of what you will actually accept, so it should say that.
Move in steps rather than all at once. Large target changes reset the learning behaviour and you will spend a fortnight finding out whether the change or the reset caused what you see next.
How to see it in your own data
Look at daily CPA and daily spend either side of 17 August for your budget-limited campaigns, not weekly totals — a weekly view will bury a gradual drift inside normal variance. What you are looking for is a slow upward slope in CPA that starts around that date and does not correspond to any change you made.
The trap is attribution. Late August brings its own seasonality, and if you only look at one campaign you will not be able to tell a platform change from a market change. The signal is whether it happened across many budget-limited campaigns at once while your unconstrained campaigns held steady. A pattern across accounts is a platform change. One campaign moving is a campaign.
That is a tedious comparison to run by hand across a client book. It is the kind of question 1ClickReport answers in one go — ask it to compare CPA before and after a date across every campaign and account you have connected, and it reads Google Ads, Meta and GA4 together rather than one dashboard at a time. You can try it free.
What we do not know yet
Google described the behaviour change; it has not published the magnitude. How far any individual campaign drifts depends on your target, your actual performance, your budget headroom and your auction. Anyone telling you a specific percentage impact a week in is guessing.
What is knowable today is the gap between your target and your actual performance, because that is sitting in your own account right now. That number is the exposure. Go and read it.
Frequently asked questions
What changed in Google Ads bidding on 17 August 2026?
Target-based bid strategies — Target CPA, Target ROAS, and Target CPC for Demand Gen — now behave differently when a campaign is limited by budget. Google states these campaigns optimise toward the configured target rather than toward the performance level they had been delivering.
Why would my CPA rise if I did not change anything?
Because a budget-limited campaign that was outperforming its target was previously held near that better level. Now the configured target is what the system aims at, so a loose target gives the campaign permission to spend up to it.
Which campaigns are affected?
Campaigns using Target CPA or Target ROAS, and Demand Gen campaigns using Target CPC, that are limited by budget. Campaigns that were not budget-constrained are far less exposed.
What should I check first?
For every budget-limited campaign, compare the last 30 days of actual CPA or ROAS against the target in its bid strategy. Where actual performance is meaningfully better than the target, that gap is what can now drift.
Sources: Google Ads Help — new features and announcements; PPC Land, August 2026. Behaviour described by Google; magnitude in any individual account will vary.
Related reading: Google Ads reporting tools compared, audit a Google Ads account in five prompts, and how campaign total budgets actually work.