Two days ago, Google flipped a switch that most PPC managers are still complaining about. The August 17 bidding change forces budget-limited campaigns using Target CPA or Target ROAS to actually hit the targets advertisers set, rather than quietly outperforming them. The initial reaction across LinkedIn and Slack channels has been predictable: "Google is taking away our cheap conversions."
That framing misses the point entirely. What Google actually did was decouple the budget lever from the efficiency lever, and for anyone who has ever tried to scale a winning campaign without watching performance crater, this is the structural fix we should have demanded years ago.
The Scaling Problem Nobody Modeled Correctly
Here is a scenario every B2B marketing leader has lived through. A campaign runs at $100 daily budget with a $50 Target CPA. Smart Bidding delivers at $35 actual CPA. Leadership sees the numbers, approves a 5x budget increase, and within 72 hours the actual CPA has blown past $60. The campaign that looked like a growth engine becomes a budget drain requiring emergency intervention.
The standard explanation was always "the algorithm needs time to learn at higher spend levels." That explanation was incomplete. As Noa Amit's analysis at PPC Hero makes clear, the real issue was that budget-constrained campaigns weren't optimizing toward the stated target at all. They were cherry-picking only the cheapest, highest-intent conversions available within the capped spend. The $35 CPA wasn't overperformance; it was a misleading baseline built on artificially restricted inventory.
When you raised the budget, Smart Bidding suddenly had to evaluate auction pools it had never touched. The algorithm wasn't "re-learning" so much as learning for the first time what the broader market actually looked like.
What Changed on August 17
Google's official documentation states the new behavior plainly: campaigns limited by budget will now "more consistently perform toward your bid target, including when you make budget adjustments." The example Google provides is instructive. If your Target CPA is $10 but recent actual performance is $5, the post-August 17 algorithm will deliver closer to $10.
That sounds like an efficiency penalty if you read it in isolation. Read it in the context of scaling, and the math changes completely.
Under the old system, a campaign delivering $5 CPA against a $10 target was not actually capable of scaling at $5. It was buying a narrow slice of inventory that happened to convert cheaply. The moment you increased budget, you discovered the true cost of the broader market, often painfully.
Under the new system, a campaign delivering $10 CPA against a $10 target is showing you what the market actually costs at scale. If you want $5 CPA, you set $5 as your target and let the algorithm tell you whether that volume exists. The target becomes a real constraint rather than a fiction the algorithm ignores when convenient.
The CFO Conversation Gets Simpler
For marketing leaders who spend half their pipeline reviews explaining why "efficient" campaigns can't absorb more budget without performance degradation, this change is a gift. The old dynamic created a credibility problem: you would show a $35 CPA campaign to finance, they would approve incremental budget, and then you would have to explain why the same campaign now runs at $55.
The new dynamic is more honest. If your target is $50 and you're delivering at $50, you can model what happens at 2x or 3x budget with reasonable confidence. The algorithm is no longer hiding a bait-and-switch inside the "Limited by budget" status.
This matters for annual planning, for board presentations, and for any conversation where marketing needs to commit to CAC payback timelines. Predictable scaling beats artificially low baselines that collapse under pressure.
Which Campaigns Are Affected
The June 2026 announcement specified that Search, Shopping, Performance Max, and Demand Gen campaigns all fall under the new behavior. App Campaigns, Video reach campaigns, and Video view campaigns continue using the previous bidding logic. Hotel and Display campaigns already operated this way.
For B2B marketers, the Search and Performance Max implications are the ones to model first. If you have budget-limited campaigns that have been "outperforming" their targets, you need to decide whether that outperformance was intentional strategy or simply a target you never updated as the market shifted.
The Audit Before the Adjustment
Google shipped a Bid Target Adjustment Tool in early July that surfaces historical performance and offers three options: keep the current target, match the target to recent actual performance, or set a custom target. If you haven't run this audit yet, you're already two days late, but the logic still applies.
Pull every campaign currently marked "Limited by budget" that uses Target CPA or Target ROAS. For each one, compare the stated target to the trailing 30-day actual performance. If the gap is significant, you have a decision to make.
Option one: lower the target to match recent performance. This preserves your current efficiency but may reduce volume, since the algorithm will now bid more conservatively across the board.
Option two: keep the target and accept that actual performance will drift toward it. This is the right choice if you believe the stated target reflects your true willingness to pay and you want the algorithm to find all available volume at that price.
Option three: raise the budget and lower the target simultaneously, testing whether the market can deliver more volume at your preferred efficiency. This is the scaling play, but it requires a two-week test window and a clear MDE threshold before you commit incremental dollars.
The Uncomfortable Truth About "Overperformance"
The campaigns that looked best in your last QBR may be the ones most affected by this change. That $35 CPA against a $50 target was never a sign of algorithmic genius; it was a sign that your budget constraint was doing the work your target should have been doing.
Going forward, the target is the constraint. Set it where you actually want performance to land, not where you hope it might land if conditions are perfect. The algorithm will take you at your word.
For marketing leaders who have spent years managing the gap between "what we show the board" and "what we know will happen when we scale," this is a chance to close that gap permanently. The math is more honest now. Use it.
