Target CPA vs Target ROAS is a question about value, not which bidding strategy sounds more advanced. Use Target CPA when conversions are worth roughly the same amount or value tracking is weak. Use Target ROAS when conversion values vary materially, those values are accurate, and the account has enough history to predict which clicks produce more business value.
Both strategies set auction-time bids automatically. Target CPA optimizes selected actions around an average cost. Target ROAS optimizes selected conversion value around an average return. If the value data is wrong, Target ROAS automates the measurement problem.
Target CPA vs Target ROAS at a glance
Target CPA, or tCPA, focuses on the average cost per conversion. Target ROAS, or tROAS, focuses on conversion value relative to ad spend. Google began simplifying the displayed strategy names in June 2026, but says the underlying bidding behavior did not change.
| Decision factor | Target CPA | Target ROAS |
|---|---|---|
| Primary objective | Conversion volume at an average acquisition cost | Conversion value at an average return |
| Best fit | Similar-value actions or incomplete value tracking | Meaningfully different values with reliable uploads |
| Required input | Accurate conversion actions | Accurate conversion actions and values |
| Main reporting metric | Cost per conversion | Conversion value divided by cost |
| Common failure | Target set too low and volume collapses | Target set too high or values misrepresent the business |
| Typical use case | Flat-fee service, one product, qualified lead goal | Ecommerce, variable deal size, stage or revenue values |
Google’s Target CPA documentation says the strategy sets bids to generate as many conversions or customer actions as possible around the desired average CPA. Its Target ROAS documentation says the strategy predicts conversion value and sets bids to maximize that value while trying to hit the average return target.
Use Target CPA when conversion values are similar
Target CPA is the better choice when one conversion is economically close to another. If a business sells one product at a fixed price or every qualified appointment has a similar expected value, there may be little useful value variation for Target ROAS to exploit.
Our operating rule is to use Target CPA when purchase or lead values sit within roughly 20 percent of each other. This is not a platform requirement. Small differences may not justify the extra measurement complexity.
Target CPA also makes sense when:
- Revenue values are not being uploaded yet
- Offline values arrive too late for useful bidding feedback
- CRM stages are reliable but dollar values are not
- The campaign has one clear qualified action
- A few outlier deals would distort average revenue
- The team needs cost control more than value allocation
For lead generation, the action still has to represent real intent. A low cost per form fill is not useful when the sales team rejects most forms. Our guide to improving Google Ads lead quality explains how to pair the bidding signal with match-type breadth.
Use Target ROAS when outcomes have different values
Target ROAS becomes useful when Google should bid differently for conversions with different economic value. If one customer is worth three times another, treating both as equal conversions can send budget toward the easiest action rather than the best return.
Use Target ROAS when:
- Purchase values vary materially, often three times or more
- Lead stages carry defensible expected values
- Closed revenue or margin is uploaded consistently
- The same conversion action can produce very different deal sizes
- The account has enough history across the relevant campaigns
- Finance, CRM, and ad-platform values reconcile
The values must represent the business outcome you actually want. Assigning every MQL a guessed value of $50 and every SQL a guessed value of $250 can be useful only if those weights reflect real downstream conversion economics. Otherwise, the bidder is optimizing a scoring system that looks precise but has no financial basis.
Google recommends reporting values across relevant campaigns for four weeks or three conversion cycles, whichever is longer, before determining a Target ROAS and activating value-based bidding.
Three questions that decide the strategy
The fastest way to choose between Target CPA and Target ROAS is to answer three questions in order.
1. Do your conversion values vary enough to matter?
If purchases or qualified outcomes sit within roughly 20 percent of each other, start with Target CPA. There is not much value signal to optimize around.
If values vary three times or more, Target ROAS deserves consideration. A $1,000 order and a $10,000 order should not necessarily earn the same bid. The same is true for a lead-generation account where one service line or customer tier carries far more expected revenue.
2. Are you uploading accurate values?
If not, use Target CPA. Target ROAS without values is not value-based bidding. Target ROAS with invented values is worse because the strategy may confidently allocate spend around bad assumptions.
Accurate value can come from:
- Transaction revenue
- Gross profit or margin-adjusted revenue
- Verified contract value
- CRM stage values based on historical close rates
- Offline customer or revenue imports
If the values live in the CRM, connect them to the original ad interaction. Our Google Ads server-side conversion tracking guide covers click-ID capture, database events, and offline imports.
3. Does the account have enough conversion history?
Google says Target CPA can work with no conversion history. We still use operating floors because an account can run while remaining too sparse for stable decisions.
For Search, our practical floor is about 15 conversions in 30 days. Demand Gen often needs closer to 50. Campaign type, conversion delay, market size, and data quality all matter. Our conversion-action threshold guide shows how those requirements change across Search, PMax, and Demand Gen.
For Target ROAS, count alone is not enough. Fifteen conversions with accurate, varied values can be more useful than 100 conversions that all carry the same guessed value. On the other hand, three large deals can make historical ROAS look strong while providing very little repeatable evidence.
Worked example: high-ticket B2B lead generation
Assume a B2B company sells contracts from $20,000 to $300,000. The monthly funnel produces 40 booked calls, 18 SQLs, and 4 closed deals.
| Event | Monthly volume | Value quality | Bidding use |
|---|---|---|---|
| Booked call | 40 | Similar event, weak revenue proof | Possible Target CPA signal |
| SQL | 18 | Better qualification, value may vary | Strong Target CPA candidate |
| Closed deal | 4 | Real revenue, very sparse | Report and feed back, but risky alone |
If the CRM does not yet upload values, optimize with Target CPA toward SQLs or a carefully weighted combination of booked calls and SQLs. Keep closed deals visible as a downstream check.
Once the account reliably uploads contract values, it can test Target ROAS. But four deals per month may still be too sparse to let closed revenue steer the entire account by itself. A value model can combine verified stage probabilities with actual revenue, but it should be built from historical data and reviewed with finance.
The right answer is not “ROAS because deals vary.” It is “ROAS after the value signal is accurate and dense enough to guide bids.”
How to set the first Target CPA
Set the first Target CPA around recent historical performance, not the acquisition cost you wish the campaign had. Google’s recommendations use recent average CPA and account for conversion delay when historical data exists.
A safe operating sequence is:
- Pull the last four weeks of CPA.
- Exclude the most recent conversion-delay window.
- Confirm the selected conversion actions are the ones you want to buy.
- Start near the trailing CPA.
- Adjust in 5 to 10 percent steps.
- Wait one to two conversion cycles before judging.
Setting the target far below history can restrict traffic sharply. Our Target CPA scaling playbook explains how to use Search Lost IS from budget, down-funnel signals, and gradual target changes without choking volume.
How to set the first Target ROAS
Start from historical conversion value divided by cost, excluding the recent conversion-delay period. Google says to use the “Conv. value/cost” column as the historical reference and warns that a target set too high can limit traffic.
If the campaign produced $400,000 in verified conversion value from $100,000 in spend, historical ROAS is 4.0, or 400 percent. Starting at or below that historical level gives the bidder room to participate in auctions it has already shown it can win.
Do not set 600 percent only because the business needs 600 percent. Confirm the value tracking, funnel economics, and market can support it. A business requirement is not auction evidence.
How target changes affect scale
The direction of the controls is opposite:
| Goal | Target CPA action | Target ROAS action |
|---|---|---|
| Increase volume | Raise CPA target | Lower ROAS target |
| Increase efficiency | Lower CPA target | Raise ROAS target |
| Avoid a sharp reset | Change gradually | Change gradually |
| Evaluate fairly | Wait through conversion delay | Wait through conversion delay |
Google’s guide to Smart Bidding target adjustments says the bidder reacts quickly, but it can take one to two conversion cycles to achieve the new target because conversions arrive later than clicks.
Major target changes can produce major spend or volume changes. Do not change the conversion goal, target, budget, and campaign structure on the same day. Change one major input, let the data mature, then inspect the result.
When not to use Target ROAS
Do not use Target ROAS when:
- Values are missing, guessed, or duplicated
- Refunds and cancellations are not reflected
- A few large deals dominate the entire history
- The conversion cycle is longer than the review window
- Stage definitions change between sales reps
- Margin varies but only revenue is uploaded
- The account cannot explain where each value came from
In these cases, Target CPA with a qualified conversion can be the more honest strategy. A simpler model built on clean data beats a value model built on fiction.
The takeaway
Choose Target CPA when conversions are economically similar or value tracking is not ready. Choose Target ROAS when outcomes vary enough to matter, accurate values flow back to Google, and the account has enough history to make those differences useful.
Start both targets from historical performance after excluding conversion delay. Make gradual changes. Judge Target CPA on qualified acquisition cost and Target ROAS on verified business value, not the platform number alone.
TNT Growth manages Google Ads, CRM integration, and down-funnel conversion tracking for brands spending $75K+/mo. Review our Google Ads and tracking services or book a 30-minute ad audit to choose the bidding strategy your current data can actually support.
Frequently asked questions
What is the difference between Target CPA and Target ROAS?
Target CPA bids to generate as many conversions as possible around an average cost per action. Target ROAS bids to maximize conversion value while trying to reach an average return on ad spend. Target CPA treats selected conversions more equally. Target ROAS needs accurate values so it can favor higher-value outcomes.
When should you use Target CPA instead of Target ROAS?
Use Target CPA when conversions are worth roughly the same amount, reliable revenue values are not being uploaded, or the value data is too sparse or inconsistent. It is often the cleaner choice for a single product, flat-fee service, or lead-generation campaign that measures one qualified action.
When should you switch from Target CPA to Target ROAS?
Consider switching after conversion values have been reported consistently across relevant campaigns for at least four weeks or three conversion cycles, the business outcomes vary meaningfully in value, and historical value per cost gives you a defensible starting target. Do not switch because ROAS sounds closer to revenue.
How many conversions do you need for Target ROAS?
Google can run Smart Bidding at low volume, but stable value-based bidding still depends on enough accurate feedback. TNT Growth uses about 15 monthly Search conversions as an operating floor, then checks value coverage, conversion delay, and whether a few large deals distort the average before trusting Target ROAS.
What happens if a Target ROAS target is set too high?
A Target ROAS target set above what historical performance can support can restrict bids, traffic, conversion volume, and total conversion value. Start at or below recent historical ROAS after excluding the conversion-delay window, then adjust gradually and evaluate the result over one to two full conversion cycles.