In the early days of Performance Max, I was running paid search for The Cotswold Company — at that point one of the UK's fastest-growing furniture retailers. December is a brutal month for furniture. The market goes quiet before Christmas as consumers hold off, then swings hard the other way on Boxing Day. The only sensible way to manage it is to restrict budget through the slow period and release it the moment demand returns.
We did exactly that. What we didn't expect was what happened to the numbers during the restriction.
Efficiency went through the roof. ROAS was running at roughly three times what we'd been hitting with uncapped budgets. The algorithm, starved of room to spend freely, had become extraordinarily picky — only pursuing the cheapest, most certain conversions. The budget cap was doing efficiency work we hadn't asked it to do.
We spent a long time thinking about why. This wasn't something we'd seen behave the same way with Smart Shopping campaigns. It appeared to be a property specific to how Performance Max's unified auction interacted with budget constraints — a new emergent behaviour that the old campaign types hadn't produced in the same way. Eventually we concluded this wasn't a seasonal quirk. It was something more fundamental: when you cap a campaign's budget below what it wants to spend, Smart Bidding can't chase volume. It can only chase value.
When the budget is capped, the algorithm can only chase value. The constraint forces better decisions. We've been applying this deliberately ever since.
What Google is changing — and why it matters
From 17 August 2026, Google is updating how Smart Bidding behaves in budget-limited campaigns. Currently, when a campaign is budget-constrained, it beats its ROAS target because the algorithm can only buy the cheapest conversions. From August, it will instead bid all the way up to the target you've set.
The practical impact is straightforward. If your Target ROAS is set to 300% but your campaign has been delivering 600% because it's budget-limited, you are about to see performance drift back toward 300%. Same budget. Same target. Fewer conversions, higher cost per result.
Nothing will visibly break in the account. The campaign will still be "working". It will just be working less efficiently — and the drop will look like market softness or seasonal variation unless you know what to look for.
What Google wants you to do about it
Google's recommended response is to adopt "demand-led budgets" — remove your caps and let the system spend what it wants to spend. They have also built a Bid Target Adjustment Tool, launching 6 July 2026, to help advertisers reset their targets automatically ahead of the change.
We recommend ignoring both suggestions.
The "demand-led budgets" advice is the part that should give you pause. Removing budget caps means spending more. More spend means more revenue for Google. The recommendation is technically correct — uncapping does allow the algorithm to optimise more freely — but it is advice that happens to align perfectly with Google's commercial interests. Draw your own conclusions about whose side of the table that advice is coming from.
The automated adjustment tool carries a different risk. A single anomalous month — a viral moment, a competitor going offline, a flash sale — can skew 90-day averages significantly. An automated tool working from those averages may recommend a target that doesn't reflect your real sustainable performance. We are resetting targets manually for every client.
What to actually do before 17 August
Find every budget-limited campaign
In Google Ads, campaigns showing "Limited by budget" in the status column. These are the ones affected.
Check actual ROAS vs target ROAS
Pull performance over the last 90 days. Compare what the campaign is actually delivering against the target you've set. If actual is significantly higher, you have a cushion that is about to disappear.
Raise the target to match actual performance
Set the Target ROAS to what the campaign has genuinely been delivering — not what you originally set it at. This is your new baseline.
Do not use Google's Bid Target Adjustment Tool
Anomalous months skew automated recommendations. Reset targets manually using your own 90-day data.
Push TROAS up incrementally
Once you've set the target to actual performance, push it further — 10% at a time — monitoring conversion volume and cost per result as you go. Stop when the campaign stops being profitable. That number is your true floor. If you're not sure what profitable looks like for your specific business, our TROAS calculator will work it through for you.
The wider point
This change removes a genuine efficiency mechanism that sophisticated advertisers have been using intentionally for years. It is a property of PMax that Google arguably didn't design on purpose — and is now, just as advertisers have understood and applied it, switching off.
Google is replacing it with a system that is, by design, more amenable to increased spending. The businesses that will be least affected are the ones who know exactly what a profitable TROAS looks like and set their targets accordingly. The businesses that will be most affected are the ones running on auto-pilot, trusting that the platform is optimising in their interest.
It isn't. It never entirely was. That's worth remembering every time a Google rep calls with a recommendation.
Not sure how exposed your account is?
We're auditing every client account before 17 August. If you'd like us to take a look at yours — free, no obligation — get in touch.
Book a free audit →