In five days, Google is changing how its Smart Bidding system behaves when campaigns are limited by budget, and the eCommerce advertisers who stand to lose the most are, counterintuitively, the ones whose campaigns are currently performing the best.
Starting August 17, campaigns using Target CPA or Target ROAS bidding strategies that are flagged as "Limited by Budget" will optimise more consistently toward the merchant's set target. That sounds like an improvement. It isn't, at least not for the merchants who've been benefiting from the current behaviour.
What's actually changing
Here's the mechanism that most advertisers don't fully understand. Today, if you set a Target CPA of £100 but your campaign is limited by budget, Google's algorithm often delivers a lower actual CPA, say £60 or £70. This happens because the budget constraint forces the system to be more selective about which auctions to enter, effectively cherry-picking cheaper conversions. The result is that your campaign beats your target, which looks great in reporting.
After August 17, Google's algorithm will instead try to hit your target more precisely, even when budget-constrained. A campaign with a £100 Target CPA that's been delivering at £60 could see its actual CPA rise toward £100, because the system will pursue conversions it previously skipped, entering more expensive auctions to use the full budget more aggressively.
Google is framing this as delivering "more easily achieve consistent and predictable performance based on your goals." That's technically accurate. But for eCommerce advertisers who set conservative targets expecting the algorithm to outperform them, this is effectively a cost increase.
Google announced the change in mid-June, via Ads Product Liaison Ginny Marvin on LinkedIn and the Accelerate with Google blog, then followed up with direct clarifications after advertisers raised concerns about efficiency. The update applies to Target CPA, Target ROAS, and, specifically for Demand Gen campaigns, Target CPC, across Search, Shopping, Performance Max, Demand Gen, and Travel campaigns.
For eCommerce advertisers, that Shopping and Performance Max are both in scope matters enormously: these are the two campaign types most eCommerce budgets actually run through.
What isn't affected is worth noting too. Manual CPC and Target Impression Share strategies are untouched. App campaigns and Video reach and Video view campaigns keep their current behaviour. And the change doesn't interact with Smart Bidding Exploration, which is designed for unconstrained budgets in the first place.
Why this hits eCommerce disproportionately
eCommerce advertisers are particularly exposed to this change for three reasons.
1. Budget-constrained campaigns are common in eCommerce. Most merchants don't have unlimited advertising budgets. They set daily caps based on their margins and expected returns. If your campaigns are regularly hitting those daily caps, which many do, especially in competitive categories, you're affected by this change.
2. Conservative targets are standard practice. Experienced eCommerce advertisers typically set Target CPA or Target ROAS targets conservatively, expecting the algorithm to outperform. Setting a Target CPA of £50 when you know the algorithm will deliver at £35 is a widely used strategy for protecting margins. After August 17, that £35 CPA could drift toward £50, and your margin buffer disappears.
3. Peak season preparation amplifies the risk. This change takes effect five weeks before the critical Q4 trading period begins. Advertisers who don't adjust their targets now will enter September with campaigns that look identical but perform differently. By the time the impact shows up in weekly reporting, budgets may have already been misallocated.
What to do before August 17
1. Audit every budget-constrained campaign, and look back further. Run a report filtering for campaigns with the "Limited by Budget" status and Target CPA, Target ROAS, or Target CPC (Demand Gen) strategies. This is your at-risk list. Google's own notifications for this change look back 12 months, not 30 days, so a campaign that only hit its budget cap during last Q4's peak could still be flagged today, even if it's looked unconstrained since.
Once you have the list, compare each campaign's set target to its actual delivered CPA or ROAS. The bigger the gap, the more room that campaign has to deteriorate. A campaign delivering at 60% of its Target CPA has more room to worsen than one delivering at 90%. Prioritise your highest-volume, best-performing campaigns first: they're usually the ones with the most to lose, because they've benefited the most from the behaviour that's about to change.
2. Tighten your targets to match actual performance. If a campaign has been delivering at a £60 CPA against a £100 target, consider reducing your target to £65-£70. This tells the algorithm that you expect performance to stay near current levels. If you leave the target at £100, you're giving the algorithm permission to spend up to £100 per conversion.
3. Consider increasing budgets on your strongest campaigns. One reason campaigns are budget-constrained is that the budget is too low for the opportunity available. If a campaign is performing well and limited by budget, increasing the daily budget may remove the constraint entirely, taking it out of the affected cohort and potentially capturing more conversions at current costs.
4. Monitor daily for the first two weeks post-change. Don't wait for your weekly or monthly reporting cycle to reveal the impact. Set up daily monitoring for CPA, ROAS, and conversion volume on all affected campaigns. The first five to seven days after August 17 will tell you whether your adjustments were sufficient.
5. Prepare a reallocation plan. If campaign costs rise and conversion volume drops, you need a plan for where to redirect that budget. This might mean shifting spend to campaigns that aren't budget-constrained, testing alternative channels like ChatGPT Ads or Shop Campaigns, or simply banking the budget for the Q4 push.
The bigger picture
This change is part of a broader pattern where Google is tightening how its automated bidding systems behave. The separation of Target CPA and Target ROAS as standalone strategies, the upcoming AI Max auto-upgrades in September, and now this budget-constrained bidding adjustment all point in the same direction: Google wants advertisers to be more deliberate about the targets they set, because the algorithm will take those targets more literally.
For eCommerce merchants, the lesson is simple: the era of setting conservative targets and hoping for outperformance is ending. Your targets are becoming your outcomes. Set them based on what you actually want, not what you hope the algorithm will beat.
Five days isn't a lot of time. But it's enough to audit your campaigns, adjust your targets, and prepare for the change. The merchants who do this work now will maintain their margins. The ones who don't will discover the change in their September reporting, and by then, the budget will already have been spent.
On Tap's perspective
Google rarely announces these changes as loudly as the impact deserves. A LinkedIn post from a product liaison and a help centre update don't carry the same weight as a platform-wide email, even when the financial effect is comparable. That gap between how a change is communicated and how much it actually costs is where most advertisers get caught out.
What's easy to miss here is that this isn't really about Smart Bidding getting smarter. It's about Google closing a gap between what advertisers typed into a target field and what the algorithm was actually willing to deliver. For years, that gap quietly worked in merchants' favour. From August 17, it stops.
The accounts that come out ahead won't be the ones with the biggest budgets. They'll be the ones whose targets reflect what they actually want to pay today, not what they set eighteen months ago and never revisited. If it's been a while since anyone looked at yours, this is a good week to check. Get in touch with our team if you'd rather have a second pair of eyes on it.
