Google Ads Is Turning Your Target Into a Spend Setpoint

Key takeaway

Google Ads is changing how budget-limited campaigns use Target CPA, Target ROAS, and Demand Gen Target CPC. Beginning August 17, campaigns that have been beating a configured target may move closer to it unless the advertiser changes the setting. The budget is still a ceiling. The target is becoming a more literal instruction about how much efficiency the system is allowed to trade for volume.

That distinction matters because many teams have treated a stale target as a guardrail. A campaign with a $10 target cost per acquisition and a recent $5 actual CPA could keep the $10 setting while celebrating the $5 result. Under the new behavior, Google says the campaign will optimize more consistently toward the $10 target when it is "Limited by budget." In other words, unused efficiency can become permission to bid more aggressively.

What's changing

Google's official FAQ says the change begins rolling out on August 17, 2026. It applies to budget-constrained campaigns using Target CPA or Target ROAS across Search, Shopping, Performance Max, Demand Gen, and Travel, plus Target CPC for Demand Gen. It also reaches eligible campaigns managed through Search Ads 360 and Display & Video 360.

Campaigns that are not limited by budget are not affected. Manual CPC, Target Impression Share, and campaign total-budget behavior are outside the change. Google also says it will not alter an advertiser's budget or target automatically.

The auction is not changing. The bidding system's interpretation of a target is.

Google describes the current behavior as confusing because a constrained campaign can outperform its target and then react unpredictably when the budget changes. The revised system is intended to keep results closer to the configured target as budgets move. Google says daily and monthly budget limits will continue to be respected, so this is not an automatic budget increase.

It can still change the economics inside the same budget. A campaign that pays more per conversion can buy fewer conversions, or it can reach additional auctions and buy more conversions at a higher average cost. In multi-channel products such as Performance Max and Demand Gen, Google warns that spend allocation across channels may also shift.

Search Engine Land offered the clearest numerical illustration in its review of the update: a campaign with a $10 target CPA and a $5 actual CPA could move closer to $10 if the advertiser leaves the target unchanged. The example is not a forecast for every account. It shows why the difference between target and recent performance can no longer be treated as harmless slack.

Why it matters

Automated bidding turns configuration into capital allocation. When a target is loose, the system does not know that finance has quietly adopted the better recent result as the new operating standard. It only knows the number the advertiser supplied.

That creates a governance problem between marketing and the rest of the business. A target CPA may have been set months ago using a contribution-margin assumption that no longer holds. A target ROAS may ignore returns, fulfillment, payment fees, discounts, or the fact that a high-revenue product produces less gross profit. Conversion delays can make a recent improvement look more durable than it is. Offline conversion imports can also change the value signal after the click.

The update makes those weaknesses more expensive because all eligible conversion sources are in scope, including click-through, view-through, and engaged-view conversions where applicable. A team that optimizes against a shallow event can get more of that event while moving away from cash contribution.

The important counterpoint is that tighter alignment can be useful. A campaign genuinely constrained by budget may have profitable demand it cannot reach. If the configured target reflects the real marginal value of another customer, moving toward it can increase total contribution even when the average CPA rises. Protecting the lowest possible CPA is not always the right objective.

The decision is therefore not "lower every target." It is whether the target represents the marginal economics the business is willing to accept today.

What operators should do

First, isolate the affected population. Filter for campaigns that are both "Limited by budget" and using the eligible target-based strategies. Separate portfolio strategies and shared budgets because Google says changes there need to be made at the portfolio or shared-budget level.

Second, compare the configured target with recent realized performance after allowing for conversion lag. Use a window long enough to cover at least one or two full conversion cycles, which is also Google's guidance before evaluating a change. Flag the campaigns where actual CPA is materially below target or actual ROAS is materially above it.

Third, rebuild the target from contribution rather than platform revenue. For an acquisition target, start with expected gross profit over the measurement horizon, then deduct returns, fulfillment, payment expense, variable service cost, promotion, and a risk allowance for uncertain retention. For ROAS, use value adjustments or offline signals where platform revenue overstates economic value.

Fourth, preserve a baseline. Record the target, budget, channel mix, impression share, conversion volume, conversion value, actual CPA or ROAS, and contribution estimate before the rollout reaches the account. Google's planning tools may be less accurate during the August 17-31 transition, according to the FAQ, so realized cohort economics should carry more weight than a short-term forecast.

Finally, define who can change the target. A media buyer can operate the campaign, but finance and product should approve the economic threshold. The setting is no longer just a marketing preference; it tells an automated market how much surplus it may spend to find the next customer.

Bottom line

Google is making automated bidding more faithful to the number an advertiser enters. That sounds like a control improvement, and often it will be. It also removes the accidental protection created when a budget-limited campaign quietly beat a stale target.

The immediate task is not to chase yesterday's best CPA. It is to replace inherited settings with current marginal economics. When the machine starts taking the target literally, the target has to deserve that authority.