Google's target-based bidding change lands on 17 August: what it does, and the account moves to make now
Google's target-based bid strategies change on 17 August 2026. Here's what actually changes as you scale — and the account moves to make before it lands.
On 17 August 2026, Google is changing how target-based bid strategies behave — and if you run Target ROAS or Target CPA on budget-constrained campaigns, it will change the numbers you report. This isn't an optional new tool you can switch on when you're ready. It's a platform-level shift to how Smart Bidding treats your target as you scale spend.
The framing from Google is "predictable and scalable ROI". That's genuinely useful — but it cuts both ways, and the accounts that do nothing before the deadline are the ones most likely to be surprised. Here's what actually changes, why Google is doing it, and the moves to make in your account this week.
What actually changes on 17 August
Today, a budget-constrained campaign frequently over-delivers its stated target. Set a 5x Target ROAS on a campaign that keeps hitting a £200/day cap and it might quietly run at 10x, because the system optimises hard within a tight budget. The catch is what happens next: the moment you lift that budget towards, say, £1,000/day, efficiency drifts back down towards the 5x you actually asked for — and it does so unpredictably, with a bumpy, hard-to-forecast decline along the way.
From 17 August, Google is tightening that relationship. Target-based strategies will aim to deliver your stated target more consistently, regardless of budget. In Google's own words: reliable delivery of the target, consistent scaling without the unexpected drop, and less need for constant budget fiddling. Set 10x, uncap the budget, and the campaign is designed to hold roughly 10x while spending more — with conversion volume rising as profitable opportunity allows. One mechanic worth remembering: campaigns can still spend up to 2x their daily budget on any given day.
Why the old behaviour existed
Google's explanation is refreshingly candid. Bidding systems that tried to squeeze maximum performance out of a tight budget sometimes created volatility. That made budget changes unpredictable, and on multi-channel campaigns like Performance Max and Demand Gen it could trigger unexpected shifts in where spend went across Search, Display and Video. If you've ever raised a PMax budget and watched the channel mix lurch for no obvious reason, that's the behaviour being addressed.
The part most advertisers will get wrong
This is where the money is. "Predictable performance" sounds like a pure win, but read it carefully: the update delivers the target you set — it does not preserve your accidental over- delivery for free. If you've been setting a 5x Target ROAS and quietly banking 10x because the campaign is budget-constrained, doing nothing means you risk handing that efficiency back as the system relaxes towards 5x while you scale.
The move is a sequence, and the order matters. Before you touch budgets, find your budget constrained, over-delivering campaigns and record the ROAS or CPA they are actually achieving — not the target printed on the tin. Then reset the target to reflect that real efficiency (Google's own worked example is changing the target to 10x tROAS), and only then uncap or raise the budget. Get it backwards — uncap first, reset later — and you scale spend against a slack target and buy a wave of lower-value conversions before you've corrected course.
What it means for Performance Max and Demand Gen
The multi-channel campaigns get a second benefit. After the update, channel-level allocation when you raise budgets should be steadier — fewer surprise swings between Search, Video and Display. Google flags this as illustrative rather than guaranteed, so treat it as a reduction in volatility, not a fixed split you can bank on. For anyone who has struggled to explain a sudden PMax channel shift to a founder, it's a welcome change. Keep watching the asset-group and channel-level reporting you can actually get at, though — "more predictable" is not the same as "transparent".
The context Google wants you to read this in
Google is packaging this inside a bigger "AI moment" story, and it's worth understanding the framing even if you discount the marketing. The pitch: over 5 trillion searches a year, Gemini 3 topping the model leaderboards, AI Mode users asking questions 2–3x longer than a traditional search, and AI Overviews users reportedly searching more often. On the ads side, Google cites AI Max for Search delivering 27% more conversions at a similar CPA or ROAS versus manual campaigns, and Demand Gen driving a 30% lift in conversions or conversion value.
Whether those figures hold in your account is a question for your own testing. But the strategic point stands: as more of the bidding is handed to Google's models, your target is the main lever you still control. Google calls the target "the compass for Google AI", and for once the metaphor is accurate — if the compass points at the wrong metric, predictable delivery just gets you to the wrong place faster.
Set the right target, not just a target
Before the deadline, pressure-test the target itself against the three questions Google's own
deck raises:
Does the metric match what leadership values? Platform ROAS is not the same as profit, market share, customer growth or year-on-year revenue. If the boardroom cares about new- customer acquisition and you're optimising to blended ROAS, predictable delivery won't save you.
Are you feeding the system complete goals? Web, app and store actions — and net new customer acquisition where it matters — not a thin slice of them. A bidder steered by half your goals will scale the half it can see.
Is your source of truth solid? Are you optimising to the same data leadership reports on, and how quickly does the bidder get confirmation a conversion actually happened? Slow or mismatched signals undercut everything downstream.
Who this matters most for
Google lists the sharpest cases as retailers managing cash flow through peak, travel and hospitality selling perishable inventory in time, financial services with strict CAC and compliance bounds, software firms planning pipeline and capacity, media businesses with finite launch windows, and CPG brands chasing shelf velocity and market share. Across our client base the common thread is simpler: anyone who scales spend seasonally feels this the moment they push budgets up. If your calendar has peaks, this change has your name on it.
What to do this week
- Pull a list of target-based campaigns that are budget-constrained — regularly hitting their cap — and over-delivering their target. These are the campaigns the change touches hardest.
- For each, record the actual ROAS or CPA over the last 30–60 days, not the stated target
- Reset the target to reflect the efficiency you're genuinely getting and can live with at scale.
- Then — and only then — uncap or raise the budget, and watch delivery closely for the first two weeks.
- On PMax and Demand Gen, note the current channel mix now, so you can tell whether post-update allocation is actually steadier.
- Consider demand-led budgets on your strongest campaigns so they can capture every profitable conversion at target rather than being throttled by an arbitrary cap.
- Reset expectations with whoever reads your reports. If a campaign's headline ROAS moves from an over-delivering 10x to a stable figure at target while volume climbs, that can mean more profit in absolute terms — but only if you've briefed it before the numbers move,
not after.
FAQ
Do I need to change anything before 17 August?
Not technically — the update applies automatically. But if you have budget constrained campaigns over-delivering their target, doing nothing risks handing back that efficiency as you scale. Audit and reset targets first.
Will my CPCs go up?
They can, and much of the industry frustration is about exactly this. If a campaign was over-delivering and now delivers at a looser target while you scale, average CPC and CPA can rise even though you're technically hitting target. That's precisely why resetting the target to your real efficiency matters.
Does this affect Maximise Conversions or Maximise Value without a target?
The change is specifically about strategies with a target set — Target ROAS and Target CPA, including the target-bearing versions of Maximise Conversions and Maximise Value. Strategies running with no target aren't the focus here.
23.png)
