Google promised advertisers a gradual adjustment when it changed how budget-limited Target CPA and Target ROAS campaigns bid on August 17, 2026. According to data presented by Smarter Ecommerce (smec) in an episode of its Growing Ecommerce podcast published on YouTube on October 20, 2026, the adjustment was anything but gradual: the median budget-limited campaign in smec's data went from beating its ROAS target by about 20% to roughly matching it within three weeks.

The episode, "Google's August 17 Bidding Update: Good or Bad? The Data", runs a claim-by-claim check of Google's rationale for the change. Its two hosts, Mike Ryan, Head of Ecommerce Insights at smec, and Christian Scharmüller, the company's Chief Client Officer and Managing Director, conclude that Google's diagnosis was correct, that the change did what Google said it would, and that it did so far faster than the word "smooth" implies. Scharmüller promoted the episode in a LinkedIn post on October 21. Every figure in the episode is smec's own analysis of campaigns it has access to. Google does not appear in the episode, and no response from Google is part of the material.

In Short

Google changed how it bids for advertisers whose campaigns run out of daily budget, so that they stop getting results much better than the goal they set. A company that manages Google Ads for retailers looked at its own data and found the extra performance disappeared in about three weeks, not slowly. If you relied on that extra performance to hit your numbers, it is gone, and the podcast hosts think some competitors overreacted in ways that left clicks on the table.

What Google changed

The background is well documented. Google disclosed the change on June 15, 2026, alongside promotion mode. Its Help Center documentation said budget-limited campaigns using a target-based bid strategy would perform more consistently toward the bid target, including after budget changes, and framed the result as more predictable performance. Notification emails followed on July 2, and a Bid Target Adjustment Tool arrived on July 6. Practitioners pushed back within weeks, and Google's Ads Product Liaison Ginny Marvin insisted the scope was narrow, describing the update as a clarification of existing controls.

On August 17 the rollout began across five campaign types: Search, Shopping, Performance Max, Demand Gen and Travel. Campaigns not limited by budget were not supposed to be affected. Google said it would not change targets or budgets automatically, and its guidance pointed to a window of mid-September to mid-October for the first reliable reads.

In the episode, Ryan recalls the announcement differently in one respect. "So basically in, I don't know, late June, I believe, buried at the bottom of a blog post, Google made an announcement that there would be updates to smart bidding in mid-August when the whole northern hemisphere is on vacation," he says. PPC Land's own record dates the disclosure to June 15, through a LinkedIn post by Marvin, a post on the Accelerate with Google blog and Help Center documentation.

Ryan's summary of the options is consistent with Google's documentation. Advertisers could stop being budget-limited by spending more, raise the ROAS target closer to what they had been getting, switch to Maximize Conversions or Maximize Conversion Value - strategies that "cannot be budget limited", in his words - or do nothing and accept the new outcome.

The mechanism Google described

Ryan says the most useful explanation he found came from a single Google slide. "What they say there is that when a campaign is budget constrained, the system automatically decreases bids to avoid hitting the budget gap," he says, according to the auto-generated captions; the word is almost certainly "cap". "So they throttle the bids on purpose." Lower bids stretched the budget across the day, "leading to cheaper conversions. And this is what caused the surplus and efficiency."

This is the same slide PPC Land reported on August 5, when Ryan shared a Google sales deck page explaining that budget-constrained campaigns can appear to overachieve, in a LinkedIn post that also argued August 17 would not be a cliff. That post said results should be judged 30 to 60 days after the change and promised data. The October 20 episode is, in effect, the delivery on that promise, after Ryan published his first charts on September 14.

Claim one: budget-limited campaigns really did overdeliver

The hosts start with the population of Smart Bidding campaigns as a whole. According to Ryan, smec has studied over-delivery against targets before. In campaigns with few monthly conversions, whether a campaign lands above, on or below its target is "very hit or miss." As conversion volume rises, the share of campaigns missing the target downwards shrinks and the share landing on target grows. The share overachieving, however, stays flat at about a third regardless of volume.

Ryan's reading of that pattern is that Google prefers to miss high. "If your ROAS is too high, you've theoretically left revenue on the table. But they know that people react better to a ROAS that is too high than too low," he says.

Budget-limited campaigns looked different. "Now, if we look at the limited by budget campaigns and how they compare to that, instead of having about a third of campaigns that overachieve, it's 55%, holy smokes," Ryan says. "And only about a third are actually on target."

The degree of overachievement was larger too. Looking back to June 1, Ryan says, the median budget-limited campaign beat its target by about 20%. The roughly one third of non-limited campaigns that overachieved did so "only by a median of 5%." On these figures the hosts mark Google's claim as true: there was a systemic over-delivery problem among budget-limited campaigns.

The episode does not say how many campaigns, accounts or markets the analysis covers, how "limited by budget" was defined for each campaign over time, or how "on target" was bounded. In the video description, smec describes itself as helping "350+ global retailers manage €680M+ in annual ad spend"; a July 2026 smec study covered by PPC Land described its data as primarily European retail. Neither figure is presented as the sample for this analysis.

Were the cheap clicks bad clicks?

Scharmüller raised a theoretical objection: lower bids should push spend toward the queries with the clearest buying intent, which could mean the surplus was partly a quality effect rather than an artefact. Ryan said older smec work on Shopping query sculpting found long-tail queries carried strong intent and cheaper CPCs, but added that he had not repeated such an analysis "in a long time." He described Google's bid landscape as "very stratified," with ceilings that mean "you can't just linearly scale a little bit more," before conceding, "now with Google's bidding and their targeting and everything, I just don't know."

The hosts settled on a metaphor. "You could argue that these limited by budget campaigns were acting as a sponge, a market level sponge that were soaking up good, cheap clicks," one of them says.

Claim two: the cause was bid throttling

To test whether lowered bids were behind the surplus, Ryan turned to impression share loss, which Google splits into two reasons: budget and ad rank. He argued the bid is now the dominant input to rank, "especially when we're talking about things like shopping."

Before August 17, according to smec's data as Ryan describes it, only a small share of budget-limited campaigns were losing impression share mainly to budget - "it's like about 5% or so". Rank was the dominant cause. "And after August 17th, those two lines cross," he says, which he reads as Google ceasing to rank-limit these campaigns to keep them from overspending. "Now they are genuinely budget limited."

Ryan's conclusion is that the budget-lost impression share metric has become a useful diagnostic again. Before the change, a campaign Google labelled as limited by budget was, by this measure, mostly losing auctions on rank.

Claim three: CPCs rose, but not everywhere

The next chart in the episode compares cost per click across the two groups. "CPCs were going up a pretty decent percentage, and they were actually getting a bit cheaper on the campaigns that are not limited by budget," Ryan says. He does not give the percentages in the transcript, and he flags a caveat: "There's a confounding thing here we should talk about."

That divergence is not how the LinkedIn post summarizing the episode reads. Scharmüller's post says the "algorithm change hit CPCs across the board - affecting both budget-limited campaigns and non-budget-limited campaigns alike." The post does not state a direction; the episode, and its chapter title "CPCs rose on budget-limited campaigns, fell elsewhere", describe opposite moves.

The hosts offer two explanations for cheaper clicks on unaffected campaigns. The first extends the sponge metaphor: once Google stopped holding bids down on budget-limited campaigns, the cheap but valuable clicks they had absorbed went back into the wider auction, where campaigns able to spend more picked them up. The second is behavioral. According to the hosts, smec saw advertisers whose campaigns were not limited by budget raise their ROAS targets as well, "which was not necessary because this change doesn't affect them," in the words of one host. A more restrictive target means bidding less aggressively, which would lower CPCs for those campaigns and leave volume available to others.

Both hypotheses are presented as untested. "They could both be true. They could both be wrong as well, who knows? But, only Google knows," one host says.

Scharmüller blames the target-raising on how the change was communicated. "I think a lot of people just didn't know what's going on here," he says. Ryan widens the blame to "all the influencers on LinkedIn, Twitter, Reddit, YouTube," adding that "a lot of people get clicks and engagement by being scary." He says a lot of those advertisers did not relax the targets afterwards and describes this as "the new temperature now." The hosts frame this as an opening for competitors that kept their targets, and Scharmüller makes the same point in his LinkedIn post: "The market reacted irrationally by being far too aggressive with its new ROAS targets." The episode does not quantify how many non-limited campaigns raised their targets, or by how much, beyond calling the share "very surprising."

Claim four: it was a cliff

The central finding concerns how fast the surplus disappeared. According to Ryan, the median overachievement of budget-limited campaigns dropped to about 11% in the week of August 17, to about 6% in the week of August 24, and "by August 31st it's at about zero." Non-limited campaigns showed what the hosts call a slight underperformance since the change, which Ryan says he is "not ready to draw conclusions about" because "there could still be some attribution effects in here."

The timing matches another data point. PPC Land reported on October 10 that the rollout was completed on August 27, a fact advertisers learned from a help page edit rather than an announcement. smec's weekly medians reach roughly zero in the first full week after that date. That alignment is PPC Land's observation, not something the hosts discuss.

The speed contradicts what Ryan himself had told the market. "I was putting message out there. I was like, guys, everyone's worried that August 17th is a cliff. It's not a cliff," he says. "But but actually the data shows that it was a [expletive] cliff." Scharmüller's reply is short: "It was. Yeah. So and so you were wrong."

Scharmüller puts the problem in commercial terms. "If I drop from the from a ROAS level of, let's say 25," he says, to "10 in 3 weeks. This is not smooth." Those figures are an illustration, not a reported case.

After the change, the distribution of budget-limited campaigns also shifted. Before, 55% overachieved. In the first weeks after, according to Ryan, "about 50% on target instead of over. And then it's about a quarter above and a quarter below." That means a larger share of budget-limited campaigns now miss their target on the low side than before. Ryan calls this "still kind of preliminary data" from the first weeks, and expects October or November figures to look more like the overall population.

The case study: a UK kitchen and homeware retailer

To move from aggregates to an account, the hosts present one advertiser, described only as "a UK retailer in kitchen and homeware." Ryan says, "It's not cherry picked or anything. I think it's quite representative, actually." The retailer is not named, and the account's spend, product range and campaign structure are not disclosed.

Before August 17, according to Ryan, its weekly average CPC sat around 17 pence and was "very consistent." In the week of the change it jumped to about 27 pence, and stayed there. Ryan describes this as "about 44% higher CPCs." On the rounded figures he gives, the increase works out to about 59%, a PPC Land calculation; the 44% may reflect unrounded weekly averages that are not in the material.

The ROAS chart, as described in the episode, showed the reverse pattern. Before the change, actual ROAS was spiky: one host says it reached up to 35, fell to about ten in other weeks, and "was often around 20." The target before the change is unclear in the transcript. Afterwards, the advertiser raised its target, Scharmüller says, "slightly to about 13," and "immediately the actual ROAS adhered to that line." Ryan describes the two lines as "married to each other."

Scharmüller treats this as Google delivering what it promised. "You know what Google said. Predictable performance. This is predictable performance," he says. The arithmetic the hosts draw out is that higher CPCs combined with stable ROAS means value per click rose in proportion. "It's not over proportionate anymore but it's proportionate," one host says.

Where the hosts land

The downside, in Scharmüller's account, is that many retailers had learned to set targets on the assumption that Google would beat them. For those businesses, he says, the over-delivered ROAS "is absolutely needed for the business to meet their goals as a business," and losing it in two to three weeks "is a serious thing." He limits his criticism of Google to two points: "It was not smooth. It was not communicated correctly."

Ryan raises a puzzle he says he cannot explain. If Google had been artificially holding bids down, he would have expected budget-limited CPCs to start below the baseline and rise to meet it. "But actually they were consistent with the baseline and they went up from there," he says. "And so in a sense, it does kind of look like punishment." He relays a question from an audience member at a webinar he gave with SMX and Search Engine Land the night before recording: "why is Google punishing us for being for being limited by budget." Ryan's own view: "this was bitter medicine and maybe needed for for the auction health."

The hosts close on the role of the target itself. "I just think the ROAS became even more important as a steering mechanism," Scharmüller says, because actual ROAS now tracks the target closely. The episode's description says the same, and also states the hosts' recommendations to advertisers; this article does not repeat them as guidance.

Reactions to the post

Under Scharmüller's LinkedIn post, freelance Google Ads consultant Joey Bidner, who criticized the change in July and again in October, wrote: "I was called crazy for raising the flag on CPC spikes after lifting troas on limited budget campaigns by Google documentation Bible readers." Scharmüller replied that "there is actually some crazy stuf going on," adding: "It is hard to understand why so many companies set more restrictive ROAS targets on their campaigns which are not limited by budget. They are not affected by the bidding change....." He said smec would share more data once it had it.

On YouTube, a commenter posting as Jason Waring wrote: "On heavily budget limited campaigns I found it better to get off of tROAS altogether. Preferring max conv value without a target in some cases." These are individual third-party comments and do not represent smec or Google.

Why this matters

For months the debate over August 17 ran on Google's assurances and practitioners' anecdotes. PPC Land's earlier coverage included Measured's argument that targets would become binding instructions, and a Reddit advertiser who said Shopping sales fell at a 300% tROAS after the rollout. The smec episode is one of the first presentations of aggregated before-and-after figures from a vendor with a large retail book. It supports Google's diagnosis - budget-limited campaigns overachieved, and lowered bids look like the reason - while contradicting the impression that the transition would unfold slowly.

The same Q4 timing compounds the effect. Google has positioned targets rather than budgets as the efficiency control, with Target CPA and Target ROAS now acting as instructions in budget-capped campaigns. Retailers that relied on beating their targets head into the holiday peak with that cushion removed, while smec's data suggests that some unaffected advertisers are bidding more conservatively than they need to.

What the material does not establish is equally important. The sample size, markets and selection of campaigns are not disclosed. The CPC percentages for the two groups are not stated in the transcript. The case study is a single unnamed account. The "irrational" target-raising is described but not counted. And the material contains no response from Google on any of the numbers.

A note on the transcript

The quotes in this article come from the auto-generated captions supplied with the video, which contain obvious errors. "ROAS" is variously captioned as "Roas", "rows", "rose", "rules" and "draws"; "bids" appears as "bits"; "ad rank" as "adrenaline"; "budget cap" as "budget gap"; "CPCs" as "keys"; and "below" as "blow". "Troy" in one host's question appears to be "Schadenfreude", and "Afro" is unclear. The opening teaser repeats lines from later in the episode with different captions: "there are clicks available for you" at 0:44 becomes "there are clickable before you" at 21:56. The case study's CPCs are captioned once as "£0.17" and then as "$0.17" and "$0.27", although Ryan says he means pence. The pre-change target ROAS is captioned as "82299 ish", which cannot be resolved, and a figure for how far actual ROAS ran above target appears as "3,040%", likely "30, 40%".

The chapter markers in the description also only loosely match the conversation. The chapter "Claim check: 55% beat their ROAS target" starts at 10:05, but the 55% figure is first stated around 13:07, after a discussion of bid stratification. Timestamps in the pasted transcript differ from the chapter list by up to six seconds.

Timeline

Summary

Who: Mike Ryan, Head of Ecommerce Insights, and Christian Scharmüller, Chief Client Officer and Managing Director, of Smarter Ecommerce (smec), hosts of the Growing Ecommerce podcast. Third-party comments come from Joey Bidner on LinkedIn and a YouTube commenter. Google is not part of the material.

What: A postmortem of Google's August 17, 2026 change to budget-limited Target CPA and Target ROAS campaigns, based on smec's own data. smec found 55% of budget-limited campaigns overachieved their ROAS target before the change, against about a third overall, by a median of about 20%; the median fell to about zero by the week of August 31. Budget-limited campaigns shifted from losing impression share mainly to rank to losing it to budget. CPCs rose on budget-limited campaigns and fell slightly elsewhere, and one unnamed UK homeware retailer saw CPCs rise from about 17p to about 27p in one week.

When: The episode was published on YouTube on October 20, 2026, and promoted on LinkedIn on October 21, 2026. The data covers June 1 through early October 2026.

Where: Google Ads campaigns in smec's dataset, whose markets are not specified; the case study is a UK account.

Why: Google said the change would make performance more predictable. smec's data suggests the diagnosis was right and the outcome is predictable, but that retailers lost their over-delivery in about three weeks rather than gradually, heading into the Q4 peak. The sample, method and Google's view are not in the material.