tCPA vs tROAS in Google Ads: Which Smart Bidding Strategy Wins in 2027

Target CPA (tCPA) wins when every conversion is worth roughly the same, which describes most lead gen accounts. Target ROAS (tROAS) wins when conversion values vary and you pass real revenue back to Google Ads, which describes most ecommerce accounts. The right choice depends on your data volume and value signal, not on which strategy sounds more advanced.

Quick answer: Use tCPA when conversions have similar value, such as leads or booked calls, and you record about 30 or more conversions in 30 days. Use tROAS when basket values vary, you send accurate revenue, and you record about 50 or more conversions in 30 days. Below those levels, start with Maximize Conversions.

What is the difference between tCPA and tROAS?

tCPA tells Google to get as many conversions as possible at an average cost you set, while tROAS tells Google to get as much conversion value as possible at an average return you set.

With tCPA at USD 80, Google treats a USD 40 lead and a USD 4,000 lead as equal wins and bids to hit the USD 80 average. With tROAS at 400 percent, Google bids harder for searches likely to produce a USD 300 basket than a USD 30 one, aiming for USD 4 of tracked value per USD 1 spent. Google's Target CPA documentation and Google's Target ROAS documentation both describe these targets as averages, so individual auctions can land well above or below them.

The core trade-off: tCPA optimizes volume at a controlled cost, tROAS optimizes value at a controlled return. tROAS is only as smart as the values you feed it.

Key takeaway: tCPA counts conversions; tROAS weighs them. Pick the one that matches what your conversions are actually worth.

When does tCPA beat tROAS?

tCPA beats tROAS when your conversions have similar value, your volume is modest, or you cannot yet pass reliable values back to Google Ads.

A plumbing company paying USD 60 per booked call, a UK accountancy firm paying GBP 45 per consultation request, or an Australian clinic paying AUD 90 per booking all have one conversion type with roughly fixed worth. Inventing values for those leads gives tROAS noise to chase. In accounts we manage, lead gen campaigns between 30 and 80 conversions a month are steadier on tCPA, with less week-to-week CPA swing than the same campaigns on value-based bidding.

tCPA also suits newer accounts. It needs fewer data points to learn a single number than tROAS needs to learn a value distribution.

The catch is quality. tCPA happily buys cheap junk leads if the form fires on junk. Our guide to Performance Max for lead gen covers how to stop that with qualified conversion goals.

Key takeaway: For fixed-value leads at moderate volume, tCPA is the simpler and more stable choice.

When does tROAS beat tCPA?

tROAS beats tCPA when order values vary widely and you send accurate revenue, because it can pay more for a USD 400 basket than a USD 40 one.

Consider a store with baskets from USD 25 to USD 350. Under tCPA at USD 30, Google treats both as equal and drifts toward cheap, easy conversions. Under tROAS, it bids up queries tied to higher baskets. Typical ecommerce accounts we audit see AOV rise 10 to 25 percent after a clean switch to tROAS, with similar or slightly lower conversion counts.

What you need first:

Accurate conversion values

Pass actual transaction revenue, ideally net of tax and shipping, with deduplicated transaction IDs. A value bug of even 20 percent sends tROAS chasing the wrong traffic.

Enough value spread

If 90 percent of orders sit within USD 10 of each other, tROAS has little to optimize, and tCPA performs about the same with less risk.

Lead gen with value rules

Lead gen can use tROAS too, but only when you import downstream values from your CRM. Our offline conversion tracking guide shows how to send qualified and closed revenue back to Google.

Key takeaway: tROAS earns its keep only when values vary and the values you send are true.

How many conversions do you need before switching?

As a working guideline from our accounts, aim for roughly 30 conversions in 30 days per campaign before tCPA and 50 or more before tROAS.

These are not hard platform minimums. Google allows both strategies on lower volume, and portfolio strategies can pool data across campaigns. In practice, campaigns below 15 conversions a month on tCPA often swing 40 percent or more in CPA from week to week because each conversion carries too much weight.

If you are below threshold, run Maximize Conversions or Maximize Conversion Value without a target for 3 to 4 weeks, collect data, then add a target. Combining two thin campaigns into one portfolio strategy can also cross the line faster.

Key takeaway: Volume first, targets second. A target on thin data is a guess Google will chase.

How do you set the right target at launch?

Set your starting target from your trailing 30 day actual CPA or ROAS, then move it 10 to 15 percent per change, no more often than every 1 to 2 weeks.

Example in USD: a campaign spent USD 6,000 for 75 conversions over 30 days, an actual CPA of USD 80. Launch tCPA at USD 80, not at your wished-for USD 55. If performance holds for two weeks, tighten to USD 70, then USD 62. Example in GBP: GBP 12,000 spend returned GBP 42,000 revenue, a 350 percent ROAS. Launch tROAS at 350 percent and step toward 400 percent in 10 to 15 percent moves.

Jumping a target 30 percent at once is the most common cause of volume collapse we see. Google responds by bidding only on the safest auctions, and spend can drop by half overnight.

Key takeaway: Start where you are, not where you want to be, and climb in small steps.

What breaks smart bidding and how do you fix it?

Smart bidding breaks on bad conversion tracking, mixed conversion actions, and repeated learning resets, and all three are fixable within a week.

Bad or missing tracking

If tags drop 20 percent of conversions, Google thinks your CPA is 25 percent higher than it is. Set up enhanced conversions in Google Ads to recover conversions lost to cookie limits and cross-device journeys.

Mixed conversion actions

Counting a USD 2 newsletter sign-up as a primary goal next to a USD 400 purchase poisons both strategies. Keep only revenue-relevant actions as primary and set the rest to secondary.

Learning resets

Big changes to targets, budgets, or conversion settings restart learning, which usually takes 1 to 2 weeks. Batch your edits and avoid budget changes larger than about 20 percent at once.

Key takeaway: Most smart bidding failures are measurement failures. Fix the data before blaming the algorithm.

How do tCPA and tROAS work with Performance Max?

Performance Max uses the same targets, set as an optional target CPA on Maximize Conversions or an optional target ROAS on Maximize Conversion Value, and it competes with Search for some of the same traffic.

Where a Search campaign and Performance Max both qualify for a query, Google gives priority to an exact match Search keyword that matches the query. Everything else is open, so PMax with a loose USD 120 tCPA can outbid a Search campaign held at USD 80 and absorb its volume.

Our working rule: keep targets consistent across PMax and Search for the same product line, within about 10 percent, and review search term insights weekly. For ecommerce, see our services page for how we structure feed-led PMax alongside brand and non-brand Search.

Key takeaway: PMax does not ignore your Search targets; mismatched targets make the two campaigns fight.

Which strategy should you pick: a decision checklist

Pick tCPA if most answers below point to fixed-value conversions and modest volume, and tROAS if they point to variable values, clean revenue data, and higher volume.

1. Do your conversions vary in value by more than 2x? If yes, lean tROAS.
2. Do you record 50 or more conversions per campaign in 30 days? If yes, tROAS is viable; if 30 to 50, prefer tCPA.
3. Do you pass real revenue, net of tax and shipping, with transaction IDs? If no, use tCPA until you do.
4. Is your main goal leads or bookings with no CRM value import? Use tCPA.
5. Do you import qualified lead or closed deal values from a CRM? tROAS becomes an option for lead gen.
6. Is your primary conversion list limited to revenue-relevant actions? If no, fix that before either strategy.
7. Has the campaign had a stable target for at least 2 weeks? If no, wait before switching.
8. Can you afford a 4 to 6 week test with 50 percent of traffic? If no, delay the switch.

Key takeaway: The checklist almost always lands on the strategy that matches your data quality, not your ambition.

How do you test a bidding switch without wrecking the account?

Run a Google Ads experiment with a 50/50 traffic split for 4 to 6 weeks, and judge it on profit per conversion or revenue, not on the first week.

Set up the experiment from the existing campaign, keep everything identical except the bid strategy, and set the new target from trailing 30 day actuals. Ignore the first 7 to 14 days of learning. Declare a winner only when the test arm beats control on your primary metric with Google's reported confidence, for example 10 percent more conversion value at equal spend, or equal conversions at a 10 percent lower CPA.

At USD 300 a day, a 5 week test puts about USD 5,250 through each arm, which is usually enough for a lead gen campaign at USD 80 CPA to produce 60 or more conversions per side. WordStream's Google Ads benchmarks research is a useful sense check on whether your CPA range is typical for your industry before you test.

As a digital marketing agency for US brands, as well as UK and Australian ones, we run this test before any bid strategy change on spend above USD 5,000 a month.

Key takeaway: A split test costs a few weeks; a blind switch can cost a quarter.

FAQ

Is tCPA or tROAS better for lead generation?

tCPA is better for most lead generation because leads usually have similar value and accounts rarely pass revenue back to Google Ads. Switch to tROAS only after you import qualified or closed deal values from your CRM and record about 50 or more valued conversions per month. Otherwise tROAS has no meaningful value spread to learn from.

How long does smart bidding take to learn?

Most campaigns finish learning in about 1 to 2 weeks after a new strategy or a major target change, longer on low volume. Judge results over at least 3 to 4 weeks. Avoid changing targets by more than 10 to 15 percent at a time, because large moves can restart learning and cause sharp drops in volume.

What happens if my tCPA target is too low?

Google bids only on auctions it is most confident will convert cheaply, so impressions, clicks, and conversions often fall sharply, sometimes by half within days. Your CPA may look good on paper while total leads collapse. Reset the target to your trailing 30 day actual CPA and tighten in small steps every 1 to 2 weeks.