Menu
From 17 August 2026, Google Ads will start bidding to your target CPA/ROAS’s rather than trying to beat them. This is huge.
Google is changing how target CPA and target ROAS bidding works when a campaign is limited by budget. Right now, some campaigns are quietly beating their target and getting away with it. For example, if you’ve set a target CPA of $10 but the campaign has actually been getting conversions for $5, Google will start pushing that campaign back up towards the $10 you asked for. Google isn’t touching your budgets or targets itself; it’s just going to start taking the numbers you typed in more literally. If you haven’t looked at those numbers in a while, now’s the time.
Imagine this – it’s late September, and you have an important client review coming up, but their campaigns that have been set at tCPA $10 but actually achieving under $5 for months have started tanking. Budgets have not changed. You are frantically checking the change history, reviewing device performance, and checking landing pages, but there is no evidence anywhere of changes that could be causing this. You’ve just been stung by the 17th August bid changes, and they’ve silently been eroding your performance like a slow leak you didn’t notice until the tire had already gone flat.
This isn’t a Google glitch or a punishment; it’s a deliberate change in how target-based Smart Bidding behaves when a campaign is budget-constrained. Additionally, “review your campaigns” is the advice everyone is giving right now, and it’s correct as far as it goes. It just doesn’t tell you which campaigns to use, which number to use, or how you’re supposed to do that across a large client book rather than one account at a time. That’s exactly what this article actually gets into.
We’ll cover why an overperforming target CPA is often a mirage rather than a win, how to easily find every campaign across every account this touches (including the ones that don’t look like Target CPA campaigns at all), what Google is actually offering you as a fix, how to work out a target that reflects reality instead of a guess, and how to protect yourself with a simple, scaled and controlled update that runs across all your accounts in one sweep.
In short:
Google won’t alter your targets or budgets, but bidding behavior will change automatically as the rollout begins.
Google has not published a detailed account of which auctions the old bidding behavior selected or rejected, and it says the auction itself is not changing. What it has confirmed is the result: some budget-constrained campaigns currently overperform their stated targets and can exhibit unpredictable changes in efficiency when a large budget shift occurs.
That overperformance is real – you genuinely paid the lower CPA. What it may not represent is the efficiency the campaign would maintain at greater scale. Under the old behavior, increasing the budget could quickly expose that gap, making the result look like a sudden deterioration even as the campaign was also buying more volume.
Starting 17 August, Google says budget-limited campaigns using affected target-based strategies will deliver more consistently towards the stated target, including when budgets change. Google’s own example is blunt: a campaign with a $10 target CPA that has recently achieved a $5 actual CPA may move closer to $10 unless the target is changed.
That doesn’t automatically mean the campaign is performing worse, and Google says the update will not directly increase spend. If CPA rises while profitable conversion volume grows, the trade-off may be exactly what the business needs. The problem is that it’s allowed to happen because nobody remembers why $10 was entered in the first place.
The number in the target field is about to carry much more weight. It needs to represent a deliberate business decision, not a relic from a historical campaign setup.
Google’s eligibility guidance splits campaigns into three groups:
| Campaign type | What happens starting 17 August 2026 |
|---|---|
| Search, Shopping, Performance Max, Demand Gen and Travel | Move to the updated behavior when limited by budget and using an affected target-based strategy |
| Display and Hotel | Already use the newer behavior, so there is no fresh change to prepare for |
| App, Video Reach and Video View campaigns | Continue using the previous bidding behaviour |
For most advertisers, the affected strategies are target CPA and target ROAS. Demand Gen campaigns using target CPC are also covered by the change.
The labels need some care. In June 2026, Google began renaming “Maximize Conversions with a target CPA” to “Target CPA” and the equivalent-value strategy to “Target ROAS.” Different Google Ads surfaces may show either name while that transition continues, but the underlying strategies are the same. A Maximize strategy with no target is different and is not in scope as target CPA or target ROAS.
So don’t audit by the displayed label alone. Pull the effective target as well. Google explains the naming transition here.
Google says its account notifications are triggered for advertisers whose affected campaigns were budget-limited at some point in the previous 12 months. That is a notification rule, not a statement that every campaign ever constrained during that period will remain permanently affected, even if it has plenty of budget.
Google states that unconstrained target CPA and target ROAS campaigns will not change behavior, but there is still a good reason to use the 12-month window for your audit: a campaign that became constrained earlier can become constrained again. That said, it makes sense to prioritize campaigns that are currently limited or repeatedly run into budget constraints. Treat the historical lookback as a safety net, not as the technical definition of eligibility.
A portfolio bid strategy has a single bidding target shared across its campaigns. If five campaigns use the same base target and you review them as five independent line items, you can end up recommending five different changes to one shared setting.
For a portfolio strategy-level review, filter for those campaigns to get a one-line aggregated view of the data, or check the portfolio strategy report directly. Do not take a simple average of each campaign’s CPA or ROAS. That gives a tiny campaign the same influence as the account’s largest spender and can produce a number the shared strategy never actually achieved. Look for the filtered view totals.
Google adds one more distinction:
Portfolio targets can be managed through the Google Ads interface and through the Google Ads API. The important point to note is that they must be updated once at strategy level, with the consequences assessed across all member campaigns. Any bulk process that treats them like ordinary campaign-level targets is modeling incorrectly.
The arithmetic is simple:
As always, choosing the right bid targets is where your judgment and knowledge of the accounts become essential.
As a guide, look back in 30-day blocks to find the point at which each campaign has seen 30 conversions, so lower-volume campaigns get a reliable CPA base on which to make your decision, and higher-volume campaigns use recent data.
However, if the campaign has had a major promotion, pricing change, conversion-tracking fix, or seasonal shift that could make the data misleading, adjust the time period accordingly.
If you need to prioritize a large, complex account or are reviewing many clients at once, calculate the percentage gap between target and actual performance rather than sorting by the raw dollar difference. A $5 gap against a $10 target matters much more than a $5 gap against a $200 target.
The maths can identify the gap, but it cannot decide whether the business values efficiency or growth more highly.
| What you find | Likely action | What that decision means |
|---|---|---|
| Actual CPA is materially lower than target, or actual ROAS materially higher | Tighten towards recent actual performance | Protect more of today’s efficiency, with volume and seasonality checked first |
| Actual is close to target | Keep the target | Avoid changing a setting that already reflects performance |
| Recent actual and the old target are both wrong for the business | Set a custom target | Choose the efficiency-to-growth trade-off deliberately |
| Actual CPA is above target, or actual ROAS below it | Investigate before loosening | A looser target makes weaker efficiency acceptable; it does not repair the cause |
| Volume matters more than a stable CPA or ROAS | Remove the target | Let Maximize Conversions or Maximize Conversion Value pursue volume within budget |
| The target is profitable and more volume is available | Increase budget | Scale towards the stated target under the more predictable new behaviour |
| Data is thin or there are no conversions | Manual decision | Do not turn an unstable calculation into an automatic live change |
If you are managing just one account, updating your bids is quite simple. It becomes much more difficult when you are managing fifty or a hundred accounts, each with multiple campaigns; it quickly becomes an operational problem. Google’s adjustment tool works within an individual account; an agency still needs a single view of every exposed target, currency, portfolio, and low-volume exception so the largest gaps can be prioritized consistently.
A safe, scaled process should be based on compiling one approval list containing:
This is the problem the Adpulse Bid Target Recalibration Skill is built to solve.
Just connect Adpulse via MCP and Google Sheets to your AI of choice, load the skill, and run it. The skill will not make any changes without your prior approval. It is free to access and uses your own token limits within your AI environment, so no additional costs.
Here’s how it works:
Where a campaign is already beating its target, the sheet will suggest a tighter target as a starting point. Where the suggested change would loosen efficiency, it leaves that as an explicit human decision. Shared strategies are surfaced as shared decisions, and anything that cannot safely follow the standard application path is flagged for manual action.
Nothing is pushed to Google Ads until the review has been edited and approved. Immediately before applying an approved change, Adpulse checks the live strategy against the reviewed version. If the setup or target has changed in the meantime, it skips the update rather than silently overwriting it and records the result in the audit log.
If you want to run this but don’t yet have an Adpulse account, just start a free (no-credit-card) trial, and it will work just fine: dashboard.adpulse.app/signup.
If you manage only a few campaigns, use Google’s tool and work through them manually. If you manage a client book, use a process that lets you review everything together. Either way, the sequence is the same:
None of that is particularly difficult – the danger is assuming that because performance has been acceptable recently, the target must already be sensible.
Get this wrong, and the failure mode will be silent. CPAs will drift towards numbers nobody consciously chose, ROAS will fall towards an old target that no longer reflects margin, and you’ll be left trying to explain declining performance with no change-history entry to point at.
Get it right, and 17 August becomes much less dramatic. Your bid targets are based on real, recent performance, and performance should remain stable.
Target CPA is the average amount you have told Google Ads to aim to pay for a conversion. Actual CPA is the campaign’s cost per acquisition over a chosen reporting window, calculated as total cost divided by total conversions. Under Google’s old bidding behavior, some budget-limited campaigns deliver an actual CPA well below their stated target. Under the new behavior, an overperforming, budget-constrained campaign may see CPA rise or ROAS fall towards its stated target. Whether that represents worse business performance depends on the additional volume and profitability.
Google will not pause the campaign or automatically change its budget or target. If an affected, budget-limited campaign has been performing more efficiently than its target, Google says its performance may move closer to the stated target after the rollout. In Google’s example, a campaign with a $10 target CPA and $5 actual CPA may begin delivering closer to $10.
Not necessarily. Google uses a 12-month lookback period to decide which advertisers receive its notifications. The published description of the bidding change refers to budget-limited campaigns. Reviewing the previous 12 months remains a sensible way to identify campaigns that are repeatedly constrained, but it should not be treated as permanent technical eligibility.
It affects Maximize Conversions when an optional target CPA is attached, and Maximize Conversion Value when an optional target ROAS is attached. With no target set, those strategies optimize for the highest conversion volume or value within the budget and are not target-CPA or target-ROAS campaigns for this change.
Check your actual CPA or ROAS over a window that reflects trading conditions and campaign history by looking back in 30-day blocks until you reach 30 conversions per campaign. Adjust it for promotions, seasonality, tracking changes, or conversion lag that would have distorted the result. Use the answer as evidence for a business decision, not as a number to apply blindly.
We love talking PPC and spend most of our time chatting with agencies, so if you have a question, please do drop us a line!
Also, Adpulse comes with a 14-day free trial, and right now the first month is only $19.99/mth, so it’s a low-risk way to try it for yourself. Sign up now!
