Target ROAS Underperforming After a Google Update? The Test
Run this diagnostic to tell the Aug 17 auction change from a real problem, then fix the right thing.
Target ROAS underperforms after a Google update when your budget-limited campaign was already beating its target before August 17, 2026, and Google's new bidding rules pull it back toward the number you typed in. Three tells confirm the cause in minutes.
Quick Answer: Is the August 17 Change Behind Your Drop
Yes, a real Google change is behind this, but only if your campaign was budget-limited and already beating its target before August 17, 2026; if either is false, look elsewhere first.
What Changed on August 17, 2026
Google stopped letting budget-limited Target CPA and Target ROAS campaigns over-deliver: starting August 17, 2026, they optimize closer to what you typed in, not what you were quietly beating (Google Ads Help, 2026).
Google's example: "If your campaign's Target CPA is $10, but your recent actual CPA performance is $5, your campaign will deliver more closely to a $10 actual CPA starting August 17, 2026."
The same applies to Target ROAS and Target CPC on Demand Gen (Google Ads Help, 2026); the gap closes now, not gradually. That's the mechanism. Whether it actually touches your account comes down to two things: campaign type, and whether you were budget-limited to begin with.
Is Your Campaign Even the Type This Touches
This change only reaches budget-limited campaigns on Target CPA or Target ROAS.
Affected: Search, Shopping, Performance Max, Demand Gen, Travel (plus SA360 and DV360 for agencies).
Not affected: budget-unlimited campaigns of any type, App, Video Reach, Video View, Hotel, Display.
If your account sits entirely in that second list, treat this article as background reading, not urgent reading.
"Limited by budget" means your campaign could spend more if the daily cap were raised; check what "Limited by budget" looks like in your account if unsure. Performance Max confuses owners most, since how budget size affects Performance Max bidding covers underfunding, a budget too small for Smart Bidding's ~30-conversion threshold: the opposite problem. Different failure mode, same panic.
Maximize Conversions with no target, Display, and campaigns that never hit their cap are untouched: skip to "The Tells That Point Somewhere Else" if any describe your account.
What a Bid-Strategy Watcher Checks First
Three checks, run in order, tell you whether this is the Aug 17 change or something else entirely.
Buzz, kampaio's bid-strategy agent, runs this sequence on every account it watches; a human running it by hand gets the same answer, no agent required, just paper and patience.
Tell 1: budget-limited and beating target before August 17. The pre-change gap size roughly predicts the size of the hit now, per Google's $10/$5 example above.
Tell 2: CPC flat or down, but value per click fell. Amsive measured this exact signature in budget-constrained Target ROAS campaigns: they "delivered more clicks at a lower CPC, but those clicks produced less value and ROAS declined" (Amsive, 2026), Aug 17-23, 2026 vs. the average of Jul 27-Aug 16, 2026.
Tell 3: conversion rate dropped with no account-side edit in that window: no new ad, no landing page, no tracking change.
Best for held close to baseline
- CPA stayed near pre-change baseline
- Sharp ROAS decline observed
Best for more volatile this week
- Delivered more clicks at lower CPC
- Value per click fell, ROAS down 32% overall, Aug 17-23 vs. Jul 27-Aug 16
All three true: the Aug 17 change is likely your answer. Any one false: check the next section.
The Tells That Point Somewhere Else
Four signs point to a different cause. Chase the wrong one and you burn a week you could spend fixing the real problem.
Tell 1: the campaign had budget headroom, not budget-limited to begin with.
Tell 2: actual performance already tracked close to target before August 17, so there was no gap to correct.
Tell 3: the drop started on a date that isn't mid-August. Check the exact day, not just "this month."
Tell 4: something else changed at the same time: tracking, a product feed, a landing page, a paused ad, or an aggressive new competitor.
Daniel Feldman (@mrfeldmanSEM), July 14, 2026, before the Aug 17 change existed, described this trap: "ROAS fell off, CPCs looked normal, nothing in the account had been touched... They'd sold through their best-selling SKU, and google kept spending like it was in stock" (Daniel Feldman, 2026). The cause was a feed outage, not a bidding change. Same symptoms, completely different fix.
None of the four tells fit? Turn to if your ROAS dropped for a different reason: five other causes (tracking, signal starvation, pMax drift, new competitor, market shift), each with a 10-30% Target ROAS cut, for a cause not yet identified. Once the tells above confirm the Aug 17 change, reset once to your real number and hold, instead of shaving a percentage on a schedule.
What the First Week Actually Showed
Amsive measured the first week (Aug 17-23, 2026, vs. the average of Jul 27-Aug 16, 2026): Target ROAS and Target CPA moved in different directions. This is the data behind the tells above, broken down by vertical and match type.
drop across the comparable Target ROAS retail campaigns Amsive reviewed, Aug 17-23 2026 vs. the Jul 27-Aug 16 baseline.
Across comparable Target ROAS retail campaigns Amsive reviewed, ROAS declined roughly 29% to 35% over that Aug 17-23, 2026 vs. Jul 27-Aug 16 window (Amsive, 2026), an early signal from a small sample, not a retail-wide benchmark.
The overall Target ROAS/value-based group in Search (not ecommerce-specific) declined roughly 32% over that same Aug 17-23 vs. Jul 27-Aug 16 window; Target CPA Search stayed close to baseline.
Search overall, Aug 17-23, 2026 vs. Jul 27-Aug 16: CPC +4%, conversion rate -15%, CPA +21%, spend and clicks flat. Budget-constrained campaigns took close to 80% of post-change dollars in that same window.
Vertical CPA increases, Aug 17-23, 2026 vs. Jul 27-Aug 16, all approximate: financial services +30%, B2B technology +28-34%, healthcare +29%, home services +22%, senior living +8%, insurance +5%.
Match type used a narrower window, Aug 17-23 vs. Aug 10-16, 2026 (Amsive's methodology note): exact became more expensive; broad and phrase converted less often. That's the shape of the damage. The fix itself is simpler than the data above might suggest.
What to Do If It Is the Change
Reset the target to your pre-change actual number, not your old aspirational one, then leave it alone.
The order below matters almost as much as the number itself.
- Get your real pre-change numberPull your actual CPA/ROAS from the 30 days before Aug 17, 2026, not after. Add the Conv. value/cost column (x100 for ROAS%), or check Avg. target ROAS vs. Actual ROAS in your bid strategy report (Google Ads Help, 2026).
- Set the target to itSet the target to that real number, or slightly better, per Google's recommendation.
- Do CPA and ROAS separatelyReset Target CPA and Target ROAS separately. Amsive's data shows they moved differently.
- Then wait out a full cycleDon't touch it again for a full conversion cycle. See the next section.
Target ROAS needs at least 50 conversions in 30 days for a reliable read (Google Ads Help, 2026, via how Smart Bidding reacts to target changes); a target beyond what data supports restricts delivery either way. See AI bidding vs. setting the target yourself if you'd rather set it yourself.
No practitioner has published a completed reset with before/after numbers, only intentions. Treat this as Google's recommendation, not a guaranteed result.
How Long Should You Wait Before Judging
Wait a full conversion cycle before judging, not the seven days Amsive's first-look data covers.
Google's guidance: "When you change targets, the bidder will react immediately but will need some time to hit the new target (give it 1-2 conversion cycles)" (Google Ads Help, 2026). Google never defines a conversion cycle in days; our own plain-English translation is roughly two to four weeks for most SMB accounts.
Two calibration points matter here. First, the rollout ran from August 17 to August 27, 2026 (Google Ads Help, 2026), so Amsive's Aug 17-23 read straddles the rollout and is a mid-rollout snapshot, not a fully-live-state measurement. Second, if performance is still off after a full cycle, either the reset was wrong or the cause was never the Aug 17 change: go back to "The Tells That Point Somewhere Else."
Frequently Asked Questions
Is Target ROAS broken after the Google Ads update?
No. Google intentionally changed how budget-limited Target ROAS and Target CPA campaigns optimize from August 17, 2026: campaigns beating target now perform closer to the typed-in number. Some "is Target ROAS broken" discussion predates the rollout by two weeks: anticipation, not confirmed breakage.
Should I lower my Target ROAS after the August 17 change?
Only if the three tells confirm the change caused your drop: reset once to your actual pre-August-17 performance (via the Conv. value/cost column) and hold, rather than cutting an arbitrary percentage. If unconfirmed, the general percentage-cut playbook for other causes still applies.
How long do I wait before judging a Target ROAS change?
Google says wait 1-2 conversion cycles: roughly two to four weeks for most SMB accounts, our own estimate, not Google's number. Amsive's first-week data covers only seven days, an early signal, not a verdict.
What is a good Target ROAS for Google Ads?
There's no universal "good" number; a workable target reflects your account's own actual conversion value per cost, not an aspirational figure. Check it via the Conv. value/cost column before setting or resetting a target.
What is the Target ROAS formula?
Target ROAS equals conversion value divided by cost, multiplied by 100. Add the Conv. value/cost column to your performance table and multiply by 100.
How is Target CPA affected differently than Target ROAS by this change?
Amsive's first-week measurement (Aug 17-23, 2026, vs. the prior 3-week average): Target CPA Search stayed close to baseline; Target ROAS got more clicks at lower CPC but less value; the overall value-based group was down roughly 32%, ecommerce/retail down 29% to 35%.
Stop Guessing Which Cause It Is
Buzz checks budget-limited status, the target-vs-actual gap, and CPC-versus-value-per-click movement automatically, then flags whether your account is the Aug 17 bucket or something else, before you spend a week guessing. That's the whole diagnostic: run once, in order, before you touch a target field. See more google ads optimization guides, or start with B6.
Not sure whether it was the update or your account?
Connect Google Ads, and Buzz runs the three tells on every Target CPA and Target ROAS campaign you have, then tells you which bucket each one is in.
Run the Three TellsSources
- Google Ads Help, 2026: change mechanics. support.google.com/google-ads/answer/17061251
- Google Ads Help, 2026: rollout timeline. support.google.com/google-ads/answer/17125145
- Google Ads Help, 2026: Target ROAS mechanics. support.google.com/google-ads/answer/6268637
- Google Ads Help, 2026: Smart Bidding floor. support.google.com/google-ads/answer/7065882
- Amsive, 2026: first-week data. amsive.com/insights/digital-media/google-changed-the-rules-on-tcpa-and-troas-the-auction-is-still-catching-up
- Daniel Feldman (@mrfeldmanSEM), X, 2026: non-Google cause example. x.com/mrfeldmanSEM/status/2077169791517839447