The performance cushion many Google Ads advertisers were quietly banking on disappeared in three weeks. Smarter Ecommerce (smec) has published one of the first aggregated before-and-after looks at Google’s August 17 bidding change for budget-limited Target CPA and Target ROAS campaigns, and the headline is clear: the ‘gradual’ adjustment was a cliff.
What Google changed
Google’s update said budget-limited campaigns using target-based bidding would perform more consistently against their target. The change rolled out across Search, Shopping, Performance Max, Demand Gen and Travel on August 17, after a June 15 disclosure, July notification emails and a Bid Target Adjustment Tool. Campaigns that were not budget-limited were supposedly untouched.
The smec data
In an October 20 episode of its Growing Ecommerce podcast, smec’s Mike Ryan and Christian Scharmüller walked through their own campaign data. Before August 17, the median budget-limited campaign beat its ROAS target by about 20%. By the week of August 24, that overachievement was around 6%, and by August 31 it was roughly zero.
That pace surprised even Ryan, who had earlier argued the update would not be a cliff. “It was not smooth. It was not communicated correctly,” Scharmüller said in the episode.
The distribution shifted too. Before the change, 55% of budget-limited campaigns overachieved, versus about a third of all smart bidding campaigns. In the first weeks after, around half landed on target, about a quarter overachieved and about a quarter missed low.
Why the surplus existed
The hosts point to Google’s explanation: when a campaign is budget-constrained, the system throttles bids to avoid hitting the budget cap. Lower bids stretch the daily budget, produce cheaper conversions, and create a ROAS surplus. smec’s data supports that diagnosis. Before August 17, only about 5% of budget-limited campaigns lost impression share mainly to budget; after, budget loss became the dominant reason, meaning campaigns were genuinely budget-limited.
A UK kitchen and homeware retailer in the case study saw weekly average CPC jump from about 17 pence to 27 pence after the update, while actual ROAS adhered closely to a newly raised target of about 13. “Predictable performance. This is predictable performance,” Scharmüller said.
What to check before Q4
- Recheck ROAS targets on campaigns that are genuinely budget-limited; the old overachievement is not coming back.
- Do not raise ROAS targets on non-limited campaigns out of fear; smec says many advertisers did and left cheap clicks on the table.
- Use impression share lost to budget versus rank as a diagnostic again.
- Treat targets, not budgets, as the new efficiency control.
Why this matters now
The timing compounds the effect. Google has positioned targets rather than budgets as the efficiency control, and Target CPA and Target ROAS now act as instructions in budget-capped campaigns. Retailers who set targets assuming Google would beat them lose that cushion just as CPCs rise on limited campaigns. Meanwhile, advertisers overreacting on unaffected campaigns may be bidding too conservatively and leaving volume to competitors.
The caveats
smec did not disclose sample size, markets or how ‘limited by budget’ was defined over time. The CPC percentages were not stated in the transcript, and the case study is a single unnamed account. Google did not respond in the material. Still, for retail advertisers heading into peak season, the practical message is to plan without the surplus.
Source: PPC Land
