Google Changed How Target CPA and ROAS Bidding Works in 2026

If a Google Ads account runs Target CPA or Target ROAS campaigns that are capped by budget, the real cost per conversion has probably been quietly better than the stated target for months. Since August 17, 2026, Google stopped letting that gap sit there. Budget-limited campaigns that have been overperforming now get pulled back toward the number actually typed into the target field, and several advertisers found out the hard way that the shift cuts both ways.

Quick answer: what changed in Google's Target CPA and ROAS bidding?

• From August 17, 2026, Google enforces the literal number set in Target CPA, Target ROAS, and Target CPC, even on campaigns marked Limited by budget, according to Search Engine Journal.
• The change only touches campaigns that meet three conditions at once: budget-limited status, a target-based bid strategy, and a history of beating that target, per Location3.
• Search, Shopping, Performance Max, Demand Gen, and Travel campaigns are all in scope, per RelevantAudience.
• Google's own guidance, from Ads Liaison Ginny Marvin, is that spend only rises when a team raises the budget on purpose.
• Google shipped a Bid Target Adjustment Tool inside the platform on July 6, 2026, giving advertisers a six week window to reset targets before enforcement began.
• At least one advertiser running Shopping at a 300 percent target ROAS while actually banking 3,000 to 4,000 percent watched sales drop once the rebalancing kicked in, per PPC Land.

What the change actually does

Smart Bidding used to treat a budget cap as the ceiling and let efficiency drift wherever the auction landed. A campaign set to a $10 Target CPA that only ever spent $5 per conversion kept delivering at $5, because the budget was doing the limiting. Google now treats that gap as something to close.

Google frames this as an accuracy fix. Delivery now matches the number an advertiser actually typed in, budget changes included, so a target set once during setup and never revisited stops acting as a hidden safety margin.

The timing tracks with how much more small teams now lean on Smart Bidding to do the work manual bidding used to do. Third-party signal loss pushed more accounts onto target-based automated strategies over the past two years, and Google is tightening how literally it reads the instructions those strategies run on.

For a team that picked an aggressive Target CPA years ago assuming it would never really bind, that assumption just expired. The target now does the work the budget cap used to do quietly in the background.

Which campaigns the change actually touches

Three conditions have to be true at the same time for a campaign to be affected: it has to be marked Limited by budget inside Google Ads, it has to run on a Target CPA or Target ROAS strategy, and it has to have a track record of beating that target. Miss any one of the three and nothing changes.

The affected campaign types cover most of what a small team actually spends on: Search, Shopping, Performance Max, Demand Gen, and Travel. That breadth is the reason a quick account check matters more than another recap of the announcement.

The risk concentrates in accounts that have run the same targets for a long time without anyone revisiting them. A newer campaign that already hits its target consistently sails through this change without a single setting to touch.

A campaign running five figures a month at a Target CPA set two budget cycles ago is a far more likely candidate than one launched last quarter with fresh numbers. Age and inattention put an account in the affected group more than spend level alone does.

What happens when nobody checks the target

One Shopping advertiser had set a 300 percent Target ROAS and simply never revisited it, while the campaign was actually delivering 3,000 to 4,000 percent ROAS in practice. After August 17, Google pulled delivery toward the stated 300 percent, sales fell, and competitors moved up past their listings in the results.

Marvin's own clarification holds at the account-settings level: the update by itself does not raise spend, since only a deliberate budget increase does that. A target nobody has looked at in two years is exactly the gap the Shopping advertiser fell into, and the update has no way to know that number was ever a guess.

This follows the same pattern as when Meta removed ad set placement exclusions a few weeks earlier. Platform logic keeps getting stricter about matching stated intent to actual delivery, and the fix is checking ahead of each enforcement date as it lands.

How to check your account before the next enforcement date

Checking exposure takes under fifteen minutes per account. Filter campaigns by budget-limited status, then compare each one's stated Target CPA or ROAS against its trailing 30 and 90 day actuals. Any target sitting far from reality is a candidate for a rebalancing whether or not Google has already applied one.

A short sequence covers most accounts:

Budget filter: pull every campaign flagged Limited by budget inside Google Ads and set that list aside for review.

Target versus actual: compare the stated Target CPA or ROAS against trailing 30 and 90 day performance for each campaign on the list.

Set it on purpose: where the gap is wide, move the target closer to reality or raise the budget deliberately, instead of letting the next platform update decide for you.

Google's Bid Target Adjustment Tool, built into the platform since July 2026, will flag campaigns likely to see the biggest swing and suggest a new target based on recent delivery. Running it once a quarter, rather than only during an announced enforcement window, turns a scramble into routine maintenance.

Someone on the team should own this check the way they own any recurring platform maintenance: a paid media lead for a mid-size account, or whoever manages the ad accounts directly for a smaller one. Put a name and a calendar date on it, or it quietly becomes nobody's job.

The gap between a stated target and real performance is as much an attribution question as a budget one. The same model-driven blind spots that show up in the difference between attribution and measurement can make a campaign look more efficient than it really is, long before Google's bidding logic catches up to it.

Google will not be the last platform to tighten enforcement like this. Folding a bidding-target check into the same cadence as a broader martech audit, or the review checklist a team already runs before adopting new automation, keeps it from becoming a surprise every time a platform ships one of these updates.

Checking this takes fifteen minutes and a habit of looking before Google looks for you. If that account review is one more thing competing with everything else on this week's list, it is the kind of work we fold into a broader services engagement, or reach out through contact and we will go through the account together.