Another week of mature conversion data gives us a better test of Google’s promise: whether actual performance is beginning to operate more consistently around the targets advertisers set.
Two weeks ago, we published our first read on Google’s August 17 change to target-based bidding. With another week of data and more time for conversions to mature, we wanted to start answering the bigger question behind the update:
Is actual performance becoming more predictable relative to the Target CPA and Target ROAS advertisers set?
Two weeks in, the verdict is still out.
At the same time, one of the strongest findings from our first analysis has held: conversion rate continues to be a bigger performance driver than rising CPCs. And by comparing tCPA performance with Maximize Conversions campaigns competing in many of the same Search auctions, we can start separating what may be specific to the bidding change from what is happening more broadly across Search.
Jump To:
Key takeaways
Actual CPA is not consistently moving closer to Target CPA yet.
Across the broader tCPA Search cohort, actual CPA went from 4% below target in the final pre-change week to 14% above target in week one and 62% above target in week two. That is distance from the stated target, not a 62% increase in CPA.
Implication: The change has not yet produced the tighter relationship between actual CPA and Target CPA we would expect if performance is becoming more predictable. There is still time for that pattern to emerge as campaigns adjust and conversions mature, but it is one of the clearest measures we are watching.
Conversion rate is still the more pervasive performance challenge.
In week two, tCPA CPC was 9% below baseline and tROAS CPC was 14% lower, while conversion rates were 8% and 29% lower, respectively.
Implication: When CPA rises while CPC is flat or declining, the first question should be what changed in conversion rate. That points us toward traffic quality, query mix and landing-page performance before assuming bidding itself is the problem.
Interestingly, the CVR weakness also extends beyond strategies affected by the update.
Maximize Conversions was not part of this target-based bidding change, but week-two CPC was essentially flat while CVR was about 30% below baseline, pushing CPA roughly 42% higher.
Implication: Because Maximize Conversions was not affected by the August 17 update, this suggests at least part of the current conversion-rate weakness is broader than the target-bidding change. It does not rule out an effect from the update, but it makes it harder to attribute the full decline to it.
Higher-volume tCPA campaigns are beginning to look different.
Among budget-constrained tCPA campaigns generating 30 or more weekly conversions, actual CPA moved from 27% above target in week one to just 8% above target in week two. Lower-volume campaigns have not shown the same consistent pattern yet.
Implication: This is an early pattern worth following. Higher-volume campaigns may be settling around their targets faster, but two weeks is not enough to know whether conversion volume is actually the reason.
Is Google actually moving performance closer to target?
Before August 17, actual CPA was often below Target CPA across parts of our cohort.
That was not necessarily a problem. Advertisers often set Target CPA somewhat above current performance when they want to create room for additional scale. As the rollout approached, teams were also making measured adjustments where stated targets and actual performance had become too far apart.
That progression is visible in the broader tCPA Search cohort.
The important thing to read here is distance from target. A campaign that lands 10% below target and one that lands 10% above target are both 10% away. We are not expecting Google to hit the target exactly every week. Normal fluctuations will continue.
What we want to see over time is that distance narrowing and becoming less variable. So far, that has not happened consistently.
The budget-constrained subset, where Google’s new behavior is most relevant, shows a similar but less extreme pattern.
Actual CPA for these campaigns has not suddenly spiked. But it also hasn’t become more closely aligned to the target yet. That is the measure we’ll keep coming back to.
Conversion rate is still the story
The first week showed that the post-change Search story was not simply about rising click costs. The second week reinforces that.
CPCs are moving around. For tCPA, the week-one increase reversed in week two. CPA ultimately landed close to its pre-change baseline. But CVR moved from slightly above baseline to 8% below it.
For tROAS, clicks became cheaper in both post-change weeks. Conversion rate remained substantially below baseline, keeping CPA elevated.
The more consistent issue continues to be what happens after the click.
What Maximize Conversions tells us about the same auction
Maximize Conversions was not part of the August 17 target-based bidding change. That is exactly why it is useful to watch.
These campaigns continue competing in many of the same Search auctions as tCPA and tROAS campaigns. They encounter many of the same users, queries, competitors and changes in advertiser behavior, but they are not operating under the updated target-based bidding behavior.
It is not a perfect experimental control. Campaign goals and advertiser mix differ. But it gives us a useful comparison for what may be unique to the update versus what is happening more broadly across Search.
Maximize Conversions did not experience a meaningful CPC increase. Its higher CPA came primarily from weaker conversion rate.
That makes it much harder to attribute the broader conversion-rate weakness we are seeing entirely to the August 17 tCPA and tROAS change.
There are still plenty of possible contributors, including auction competitiveness, query mix, seasonality, user behavior and individual account performance. But the comparison tells us that conversion weakness is showing up outside the campaigns operating under the new target-based behavior.
Higher-volume campaigns may be settling faster
One of the more interesting early signals appears when we split budget-constrained tCPA campaigns by weekly conversion volume. The highest-volume group begins to look different.
For campaigns generating 30 or more weekly conversions, actual CPA moved from 27% above Target CPA in week one to 8% above target in week two. That is the kind of movement we would expect to see if performance begins settling more tightly around the target.
The lower-volume groups have not shown the same consistent pattern.
That does not mean conversion volume is the reason. The campaigns within each group have different advertisers, goals and economics. The question worth continuing to test is whether campaigns with more weekly conversions settle around their targets faster under the new behavior.
What advertisers should watch now
Break CPA into its components.
If CPC is stable or declining while CPA rises, look at conversion rate and traffic quality before assuming the bidding target is the problem.
Use better conversion data.
First-party data, qualified conversion actions, offline outcomes and appropriate exclusions can give Smart Bidding more useful information about which conversions matter to the business.
Stay close to Search queries.
Continue reviewing query quality, negatives and match-type behavior. Automation does not remove the need to understand what traffic you are buying.
Watch impression share, budget and top-of-page visibility together.
Search impression share, top impression share and absolute top impression share can help show whether an individual campaign is entering a different mix of auctions even when aggregate CPCs look relatively steady.
Make measured target changes.
If the goal is greater predictability around the target, repeatedly changing that target makes it harder for the system to settle and harder for advertisers to evaluate whether the new behavior is working.
What we’re watching next
The first two weeks have given us a better question than whether CPA simply went up or down: is Google actually making target-based bidding more predictable?
We don’t have enough evidence to say yes yet.
Over the next several weeks, we’ll keep measuring how far actual performance lands from target, the share of campaigns landing within reasonable ranges around that target, and whether that gap becomes smaller and more consistent.
We will also continue comparing target-based campaigns with Maximize Conversions and watching whether weekly conversion volume helps explain why some budget-constrained campaigns appear to settle faster than others.
The first week showed us how performance initially reacted. The next several weeks should tell us whether Google’s new bidding behavior actually delivers the predictability behind the change.
Methodology
We kept the original study windows fixed and refreshed the same periods after allowing additional time for conversions to mature: Jul. 27-Aug. 16 as the pre-change baseline, Aug. 17-23 as week one and Aug. 24-30 as week two.
The analysis uses the same account and campaign comparability rules and influence checks as the original study. Maximize Conversions is analyzed separately as a directional same-auction comparison and was not part of this specific target-based bidding update.
For actual-versus-target reporting, we are measuring distance from target in either direction, not percentage growth in CPA. Because advertisers also made measured target adjustments before and during the rollout, we will continue refining this analysis using campaign-level target history as the series develops.
Watch our latest webinar on demand, Earn Trust in AI Discovery with Search, Social, and Media, for practical ways to strengthen visibility as search behavior changes.
Want a closer look at your paid search performance? Connect with Amsive.
