Google Ads' August 17 bidding change removes the efficiency surplus

Published: 16 July 2026 | Updated: 17 July 2026 | Effective date: 17 August 2026

Google is converting the advertiser's target from a loose boundary into an operating setpoint. If a limited-by-budget campaign currently performs better than its target, doing nothing can mean paying more for the same result or receiving less output from the same budget.

What Google is changing

From 17 August 2026, Google will make affected budget-limited campaigns perform more consistently towards their stated target. This applies to Target CPA, Target ROAS and Target CPC for Demand Gen across Search, Shopping, Performance Max, Demand Gen and Travel.

Google will not automatically change the campaign's budget or target. It will change the bidding behaviour underneath those settings.

Google's own example is explicit: a campaign with a $10 Target CPA that has recently achieved a $5 actual CPA will start delivering closer to $10 unless the advertiser lowers the target.

The actual economic mechanism

Before this change, a tight budget could force Smart Bidding to concentrate spend in the most efficient auctions. The advertiser might declare a maximum acceptable CPA of $10 while the budget constraint kept actual CPA at $5.

After the change, Google will use the $10 target as a stronger instruction. The budget still caps total spend, but it no longer protects the overperformance created by scarcity.

• Before: the target stated what the advertiser would tolerate; the budget could force Google to do better.
• After: the target becomes the operating setpoint even when the campaign is budget constrained.
• Transfer: the efficiency gap moves from the advertiser to Google's bidding system.

The target is a pain threshold

A Target CPA is not a statement of what a conversion should cost. It is the highest average acquisition cost the advertiser has authorised. A Target ROAS is the lowest return the advertiser has authorised.

Google already receives the budget, conversion values, conversion definitions and enough auction-time signals to decide how aggressively to bid. The August change makes the declared tolerance more usable. If the campaign achieves a £60 CPA against a £100 target, the £40 gap is no longer harmless headroom. It is permission for the system to move towards £100.

This does not prove private intent. It does establish the incentive and the effect: Google gains more auction latitude, while the advertiser must intervene merely to preserve the previous economics.

What happens at a fixed budget

Apply Google's CPA example to a £1,000 budget:

• Current performance: £1,000 at £5 CPA produces 200 conversions.
• Performance at the target: £1,000 at £10 CPA produces 100 conversions.
• Budget needed to restore volume: £2,000 at £10 CPA produces 200 conversions.

At fixed spend, doubling CPA halves conversion volume. To recover the original output at the declared pain threshold, the advertiser must double the budget.

This is arithmetic, not a forecast that every campaign will move all the way to its target. Google has not disclosed the expected movement by account, campaign type or vertical.

What if the advertiser lowers the target to £5?

The calculation above assumes that the full £1,000 is spent. A budget is a ceiling, not guaranteed spend. When the advertiser lowers Target CPA from £10 to £5, the efficiency target can become the binding constraint. Google states that a more efficient target will likely affect daily spend and that setting a target too low can forgo clicks and reduce total conversions.

Target CPA uses predicted conversion likelihood to set a different bid for each auction. A useful economic illustration is justifiable CPC ≈ Target CPA × predicted conversion rate. This is not Google's disclosed bidding formula; it shows why halving the target makes the system less competitive in marginal auctions.

At the lower target, Google does not necessarily stop evaluating those auctions. It submits lower bids and loses more of the marginal ones. The operational result is narrower auction participation, less available spend and fewer leads when the supply of £5 conversions is finite.

• Enough efficient traffic exists: the campaign spends £1,000 at an average £5 CPA and produces approximately 200 conversions.
• Only £600 of traffic meets the target: the campaign spends approximately £600, produces approximately 120 conversions at £5 CPA and leaves £400 unspent.
• The target remains at £10: the new system can spend the full £1,000 closer to a £10 CPA and produce approximately 100 conversions.

Lowering the target does not manufacture 200 cheap leads. It tells the bidder to protect the £5 economics by surrendering auction reach. The advertiser can preserve CPA, but cannot preserve CPA, spend and conversion volume when enough £5 inventory does not exist.

The ROAS version is the same transfer

• Current performance: £10,000 at 600% ROAS produces £60,000 in conversion value.
• Performance at the target: £10,000 at 400% ROAS produces £40,000.
• Budget needed to restore value: £15,000 at 400% ROAS produces £60,000.

The advertiser spends another £5,000 to recover the original conversion value. Scaling is rational only when the marginal return remains profitable after contribution margin, fulfilment costs, refunds and customer quality are included.

Who wins

• Google: it gains more latitude to participate in marginal auctions and a stronger basis for recommending higher budgets.
• Advertisers with accurate profit-derived targets: they gain more predictable scaling at a genuinely acceptable marginal return.
• Businesses seeking additional profitable volume: they can raise budgets with less performance movement caused solely by removing the budget constraint.
• Competent agencies and in-house teams: target governance and measurement quality become more valuable.

Who loses

• Fixed-budget advertisers: the same budget can produce fewer conversions or less conversion value.
• Accounts with stale or deliberately loose targets: unused tolerance becomes bidding permission.
• Small and lightly managed advertisers: inaction preserves the old target but not the old efficiency.
• Poorly measured lead-generation accounts: Google will pursue weak proxy conversions more consistently.
• Performance Max and Demand Gen advertisers: Google says channel allocation can shift, but it does not disclose the direction or magnitude.
• Agencies that fail to act: avoidable performance deterioration becomes an account-management failure.

Google's commercial benefit

The change does not automatically increase spend. Daily and monthly budget limits remain in place. Google therefore does not receive an immediate revenue increase merely because the behaviour changes.

The structural benefit is stronger. Google can consume overperformance by bidding for and allocating traffic closer to the advertiser's revealed tolerance. It also explicitly recommends budget buffers, applying budget recommendations and increasing budgets after the change.

The transferred surplus does not necessarily become Google revenue in full. It can also flow to publishers, more expensive inventory, competitors or incremental conversions. Google has not disclosed that distribution.

What advertisers should do before 17 August

1. Inventory affected campaigns. Export every campaign using Target CPA, Target ROAS or affected Demand Gen Target CPC that has been limited by budget during the previous 12 months.
2. Measure the gap. Compare the current target with actual 28-, 56- and 84-day CPA or ROAS after allowing for conversion lag.
3. Validate the conversion signal. Check primary goals, values, offline imports, duplicate counting, refunds and lead quality.
4. Calculate the economic limit. Set maximum CPA or minimum ROAS from contribution margin and customer value, not from Google's recommendation interface.
5. Fix portfolio and shared-budget targets at their controlling level. Split campaigns when their margins or lead values are materially different.
6. Save a baseline. Record spend, conversions, conversion value, CPA, ROAS, budget status and channel allocation before rollout.
7. Make changes early enough to observe mature conversions. Google recommends waiting one to two conversion cycles before judging performance.

What not to do

Do not blindly apply the Bid Target Adjustment Tool. Recent actual performance is evidence, not automatically the correct commercial target. Do not increase budgets because Google says scaling is now safer. Increase them only when the marginal CPA or ROAS remains profitable.

Do not layer unrelated restructures onto the rollout period. Google warns that forecasts may be inaccurate from 17 to 31 August. Preserve a clean comparison window.

Two opposing readings

These are the strongest versions of the traditional platform-defence and platform-critique arguments. They are not quotations, impersonations or statements attributed to any current spokesperson.

The platform-defence case: targets should mean what they say

The old behaviour made two advertiser controls contradict each other. A campaign could be given a £100 Target CPA and a fixed budget, yet return a £70 CPA because the budget prevented it from entering more expensive auctions. The reported result looked efficient, but it did not describe what the campaign could deliver at scale.

The new behaviour makes the controls explicit. The target defines the acceptable efficiency boundary. The budget defines the maximum amount available. If an advertiser sets both accurately, Google can use more of the budget while keeping performance near the stated target. That produces greater forecasting certainty and makes scaling less dependent on an accidental surplus created by budget constraint.

Advertisers are not forced to accept worse economics. Google is providing a tool to move targets towards recent actual performance, budgets do not automatically increase, and the underlying auction does not change. If a £70 CPA is the real requirement, the advertiser should set £70 rather than declare £100 and expect the system to ignore the extra tolerance.

The defence in one sentence: a target should be an instruction, not a decorative ceiling, and clearer control semantics make automated bidding more predictable at scale.

The monopoly-rent critique: declared tolerance becomes extraction capacity

Google operates the auction, supplies the bidder, controls most of the performance data and recommends the targets and budgets. The advertiser supplies two commercially sensitive inputs: how much can be spent and how much inefficiency can be tolerated. This change gives the system more freedom to consume the distance between current performance and that declared pain threshold.

The advertiser receives no automatic preservation of the existing surplus. Keeping the previous CPA or ROAS requires intervention. Doing nothing allows performance to move towards a less favourable target, and Google's own guidance presents higher budgets as the route to additional volume. Fixed-budget advertisers can therefore pay the same amount for fewer conversions or less conversion value, while Performance Max advertisers may also receive a different and less transparent channel mix.

This is not neutral simplification. A platform with market power has converted unused advertiser tolerance into additional auction demand. Some of that demand may produce genuinely incremental value, but Google controls the mechanism and captures the revenue opportunity. The change transfers the default advantage from the buyer's efficiency surplus to the platform's monetisation system.

The critique in one sentence: when the auction owner also runs the bidder, an advertiser's maximum acceptable cost becomes a price-discovery signal for extracting more of the available margin.

The uncensored take

Summary

This is an algorithmic margin squeeze presented as predictable scaling. Target CPA and Target ROAS are moving from loose boundaries towards operating setpoints. If a budget-limited campaign has been producing £5 leads against a £10 target, Google has explicitly said that performance will move closer to £10 unless the advertiser intervenes. The platform is closing an efficiency gap and gaining room to consume the surplus inside it.

A £100 CPA target was never a wish. It was the advertiser's exposed nerve. Previously, budget scarcity could protect it by forcing the bidder to concentrate spend in the most efficient auctions. The August change loosens that protection while leaving the pain threshold in place.

Factions

• The platform: Google describes the change as clearer control and more predictable scaling. Operationally, it gains more latitude to enter marginal auctions and to recommend higher budgets at the target the advertiser supplied.
• The asleep-at-the-wheel advertiser: businesses set inflated targets as safety margins, then treat a commercially lethal instruction like a decorative ceiling. They disclose their maximum tolerance and fail to maintain it.
• The mercenary adviser: this article turns the change into a case for paid account review. That commercial interest is real. It does not alter the arithmetic, and every material mechanism can be checked against Google's documentation and the advertiser's own campaign data.

The bullshit

Google's framing is incomplete. “Targets should mean what they say” is defensible control logic, but it omits who receives the efficiency surplus when a previously superior result is pushed towards the target. Calling the change predictable does not make the transfer neutral.

Advertiser outrage also contains an element of pretence. A business cannot voluntarily hand the auction operator its maximum acceptable cost, leave that number stale for months and then act surprised when the automated bidder treats it as an instruction.

This is not theft. It is permissioned extraction. The advertiser supplies the tolerance. Google supplies the bidder and operates the auction environment. The target states how much inefficiency can be accepted, the budget states the maximum exposure, and the conversion signal tells the system what to pursue.

In a normal negotiation, maximum willingness to pay is private strategic information. In Google Ads, the buyer gives a version of it to the platform that also sets auction-time bids, measures the result and recommends the next budget. That conflict is the real story.

Winner

Google. The house has the structural advantage because it operates the auction, writes the bidding system, controls the recommendation interface and bills the spend. Not every pound of lost advertiser surplus becomes Google revenue; some can flow to publishers, competitors, more expensive inventory or additional conversions. Google still gains the revenue opportunity and increased auction demand.

Verdict

This is the digital rentier economy in miniature. Advertisers mistook budget-enforced overperformance for a durable entitlement. Google is now operationalising the tolerance they entered and penalising anyone who failed to connect that setting to real unit economics.

Google will not magically discover an advertiser's true profit margin. It does not need to. It can optimise against the proxy the advertiser supplied and consume every unit of inefficiency that proxy permits.

Material extraction. Preferences become parameters. Limits become targets. Tolerance becomes extraction capacity. Survival means auditing the target, repairing conversion quality and refusing every budget recommendation that fails outside Google's interface.

Sources

• Google Ads: Changes to target-based bid strategies
• Google Ads: Frequently asked questions about the changes
• Google Ads: About Target CPA bidding
• Google Ads: About Smart Bidding
• Google Ads: How the auction works

Need the affected campaigns checked?

I can identify the campaigns exposed to this change, calculate the target-versus-actual gap and separate genuine scaling opportunities from surplus Google is about to consume.

Request a Google Ads review