Your shopping cart is empty!
Google Ads stops treating tCPA and tROAS targets as a ceiling
Publication date: 12.09.2026
Target CPA and Target ROAS in Google Ads are no longer a safety net. The target used to be a ceiling: a campaign with a $10 tCPA could happily deliver conversions at $5, and that counted as normal. Now the target is literal: the system steers the campaign towards conversions at roughly $10, not cheaper.
The upside and the downside
The upside is predictability. When you scale budget, you know what cost you will hit. The downside is just as clear: campaigns that outperformed their target for years lose that advantage.
| Before: target as a ceiling | Now: target as a goal | |
|---|---|---|
| System behaviour | finds conversions no dearer than the target | steers average cost to the target |
| A $10 tCPA campaign | could deliver at $5 | delivers around $10 |
| Scaling | unpredictable | predictable |
| Who wins | those who padded the target | those who set it from actuals |
The market has been here before
Search Engine Land quotes Reva Minkoff, founder of Digital4Startups, from an SMX Now webinar: in 2015–2016 tCPA worked exactly this way — Google described the strategy as setting bids so that "the average cost per conversion equals the target chosen by the advertiser". What changed is not the mechanics but the ecosystem around them: back then there was no Performance Max, no automated creative, and nothing like today's volume of signals feeding the algorithm.
Eight steps for the new target logic
- Decide the campaign's job: volume or efficiency. For volume use Maximize Conversions; for efficiency that caps spend use tCPA/tROAS.
- Set the target from actual numbers, not from thin air. If CPA is $30 today, start at $30, not $12.
- Lower the target in 10–20% steps and wait one or two conversion cycles. Minkoff's moving-services client cut CPA by 75% in two weeks this way: $10 → $7.50 → $5.
- Do not touch the target daily. Accumulate enough conversions first to see whether the system holds the target at all.
- If tCPA became a brake, step down: Maximize Conversions, then Maximize Clicks. That tells you whether the target is really the problem.
- Check conversion quality: primary actions must be real business outcomes, not micro-events.
- Split campaigns with different economics — brand and non-brand, new customers and returning. Their acceptable CPA differs.
- Watch adjacent metrics: search impression share, impressions lost to budget, volume trend and CPC. They show first that the target has strangled delivery.
Small and niche accounts have it harder
The fewer conversions a campaign gets, the longer those "one or two cycles" actually take. An account with three leads a week sees the real reaction to a target change in a month, not two days — which is exactly why daily edits hurt most here.
The second trait of the Ukrainian market is narrow niches where brand and non-brand differ in cost by multiples. If they still sit in one campaign, the new logic makes the blended cost worse than either: the system pulls cheap brand traffic up towards the shared target. Separating those campaigns is usually the first thing that surfaces in a contextual advertising audit.
Calling this "automated bidding broke" overstates it: what actually broke is the margin accounts lived on for years — the number in settings now finally matches the number in the report.
