Google Ads Will Spend Your CPA Surplus Starting August 17. Audit Before July 6.

A Google Ads campaign hitting a 5 dollar cost per acquisition against a 10 dollar target sounds like a win. After August 17 it turns into a leak. Google is changing how budget-limited campaigns on Target CPA and Target ROAS behave, and campaigns that have been beating their targets will be pulled back toward the number you actually set. That 5 dollar CPA is now free to drift up toward 10. Nobody turned this on. It arrives automatically.

Google announced the change on June 15, alongside two other bidding updates, and frames it as more consistent delivery toward your stated targets. Read it from the other side. If your target is stale, set loosely months ago and quietly outperformed ever since, Google just got permission to spend the gap. The fix is not complicated, but it has a deadline. Reset your targets to the performance you actually want before August 17, and use the six-week window Google is handing you to do it.

What changes on August 17

The feature is called Bidding Target Optimization. It is a backend change, not a toggle, and it hits one specific group, budget-limited campaigns using Target CPA or Target ROAS that have historically over-delivered. Google's own illustration is a campaign set to a 10 dollar Target CPA that has been delivering around 5. After August 17 that campaign aims closer to 10 than to 5. The 10 versus 5 numbers are Google's example, not a measured result, but the direction is the whole point. Over-performance gets spent, not banked.

Budget-limited is the trigger word. These are campaigns that consistently spend their full budget and could use more. Google's reasoning is that if the budget caps you anyway, the algorithm should chase volume up to your stated target rather than leave it on the table. For an advertiser who set a deliberately tight target, fine. For the many who picked a round number once and never revisited it, this is an efficiency tax.

Google is shipping a Bid Target Adjustment Tool on July 6 that surfaces your historical performance and offers three ways to respond. That gives you a genuine window. Between July 6 and August 17 you can see which campaigns have been over-delivering and decide, per campaign, whether that surplus was strategy or just a target you forgot to update.

Two more levers, one real deadline

The same June 15 announcement expanded Smart Bidding Exploration, live since 2024 on Search, to all Performance Max campaigns without a product feed, globally, plus a new beta for Shopping and for Performance Max campaigns that carry a feed. You set a ROAS tolerance, and the algorithm uses that headroom to bid on plausible but unproven queries it would otherwise skip. Google's internal data, from a 2025 window and flagged as varying by advertiser, puts it at an 18 percent lift in unique query categories with conversions and 19 percent more conversions overall. Useful, but it loosens efficiency on purpose, so pair it with the target discipline below, not against it.

The third update, Promotion mode, is in beta for Search and Performance Max. It lets you schedule a temporary ROAS tolerance change and extra daily budget for a defined window, a flash sale or a launch, without rebuilding the campaign. It works with daily budgets and with the campaign total budgets Google added back in January. Handy, but optional. The target change is the only one of the three carrying a date you cannot ignore.

The audit to run before July 6

You do not need to wait for the tool to know your exposure. Pull a campaign export and let Claude do the triage. The fields that matter are, per campaign, the bid strategy (Target CPA or Target ROAS), the target value you set, actual CPA or ROAS over the last 30 to 90 days, whether the campaign is budget-limited, plus spend and conversions.

The rule is simple. Flag any budget-limited campaign on Target CPA or Target ROAS where actual performance runs meaningfully better than the target. Those are the campaigns August 17 will loosen. For each one, the corrected target is the number that locks in the efficiency you are already getting, not the round figure from launch day. A campaign holding a 5 dollar CPA against a 10 dollar target should carry a target near 5, with a little headroom, before Google decides for you.

We built a Claude skill that runs exactly this audit. Drop in your Google Ads export and it returns the at-risk campaigns, how far each is over-delivering, and a suggested reset target for each, ready to apply before August 17. Get the free skill.

The window is the gift

Google rarely announces a behavior change two months early and ships a tool to soften it. Treat the gap between July 6 and August 17 as the real product here. The advertisers who reset their targets inside that window keep their efficiency. The ones who skim the announcement and move on will watch their CPA creep up in late August and blame the algorithm.

Sources: PPC Land, Google Ads bidding and budgeting updates (June 15, 2026); Accelerate with Google blog, Ginny Marvin announcement; Digital Applied, June 2026 bidding overhaul playbook; Google internal Smart Bidding Exploration data (March to April 2025, varies by advertiser).