Target CPA vs. Target ROAS: How to Choose the Right Google Ads Bid Goal
Choose Target CPA or Target ROAS based on conversion value, data volume, and margin economics—without starving campaigns of volume.


Most account audits start with a mess in bid settings. An agency says it is “optimizing for ROAS” on a B2B lead form with no monetary values attached. Or an ecommerce store sells $30 phone cases and $1,400 espresso machines on a flat cost-per-acquisition goal, then wonders why conversion volume rose while gross margin collapsed.
The problem is simple: performance marketing uses the same acronyms for two different jobs. CPA and ROAS measure past efficiency; Target CPA (tCPA) and Target ROAS (tROAS) tell Google how to bid in live auctions. Confuse the metric with the bidding strategy and you give the algorithm the wrong objective.
The practical rule: use CPA when conversions have similar value. Use ROAS when their values vary materially.
Both metrics measure advertising efficiency, but they answer different financial questions. Cost Per Acquisition measures the media cost required to secure one conversion (Ad Spend ÷ Conversions); Return on Ad Spend measures the revenue returned per dollar of media spend (Conversion Value ÷ Ad Spend).
Spend $1,000 to generate 20 orders worth $4,000 and your CPA is $50. Your ROAS is 400%, or 4.0x. Neither metric is inherently better. Each fits a different business reality.
I used to tell clients CPA was the simple one. I was wrong. CPA stays simple only when every conversion pays roughly the same.
That is the decision at the reporting level: similar values, watch cost per action. Uneven values, watch revenue per dollar.
Google made this harder than it needed to be. What old hands knew as Target CPA and Target ROAS became longer labels: Maximize Conversions with a Target CPA and Maximize Conversion Value with a Target ROAS.
Google is bringing back Target CPA and Target ROAS as standalone names, while Maximize Conversions and Maximize Conversion Value remain the no-target, get-me-volume options. The bidding behavior does not change; only the dropdown label does.
Practical takeaway: when an old audit mentions tCPA or tROAS, read it as the target-constrained version of Maximize Conversions or Maximize Conversion Value.
Think of Google’s bidding choices as two base algorithms with an optional limiter.
Maximize Conversions spends the daily budget to find as many conversions as it can. Add a Target CPA and you tell it to find volume without paying more than $X on average.
The value side works the same way. Maximize Conversion Value spends the budget to generate as much conversion value as possible. Add a Target ROAS and you tell it to preserve a return threshold, such as 400%.
Base algorithm buys volume. Target constrains efficiency.
| What you select | What Google actually does | What you must feed it |
|---|---|---|
| Maximize Conversions | Chases conversion count up to budget, with no efficiency cap | Clean conversion tracking; no value needed |
| Target CPA | Chases conversion count while holding average CPA near your target | 30+ conversions in 30 days and stable CPA history |
| Maximize Conversion Value | Chases total revenue up to budget, with no efficiency cap | Accurate values on every conversion |
| Target ROAS | Chases total revenue while holding return near your target | 30+ valued conversions and a realistic ROAS target |
The choice between CPA and ROAS comes down to two variables: value dispersion and data volume. Value dispersion is how widely transaction values vary. Data volume is how many conversions flow through the campaign every 30 days.
Use the business model first, then check whether the campaign has enough data to support the choice.
Single-tier lead generation: choose Target CPA. Home services and local clinics often fit here. If each booked plumbing job or dental cleaning produces a predictable return, Target CPA can keep cost per lead steady without artificial conversion values.
Multi-price ecommerce: choose Target ROAS. Retail and D2C catalogs with cart values from $25 to $450 need the algorithm to distinguish between low-value and high-value purchases. A flat CPA goal can chase cheap impulse buys because they convert easily. Target ROAS can weight bids toward higher-value intent.
B2B SaaS and high-ticket pipeline: start with Target CPA. Optimize for qualified demo requests. Once your CRM sends back closed-won pipeline values through offline conversion tracking, switch to Target ROAS so the machine can weight bids toward higher annual contract values (ACVs).
Data density determines whether that choice survives in live auctions. Google states that Target ROAS requires at least 15 conversions in the last 30 days for Search and Shopping campaigns. That is a floor, not a comfort zone.
In practice, Target ROAS with fewer than 30 monthly conversions can produce erratic bidding swings or budget under-delivery. When volume is thin, use standard Maximize Conversions with human or autonomous execution guardrails to give the campaign room to find footing before you impose strict targets.
Practical takeaway: do not ask a low-volume campaign to make high-confidence value decisions from a handful of conversions.
The expensive Smart Bidding mistake is treating a target as a wish list instead of a calibration dial. If a campaign runs at a $75 CPA and you set a $35 tCPA overnight because that is what your unit economics require, the algorithm does not discover $35 customers. It stops entering auctions that cannot meet the constraint, which can starve the campaign of impression volume.
Change CPA or ROAS targets gradually, then give the bidding engine time to recalibrate before changing them again. A target is a guardrail, not a negotiation with the auction.
The second trap is faking value data. Some teams assign a static $50 value to every contact form fill just to access Target ROAS bidding. When every conversion has the same price tag, Target ROAS is just Target CPA wearing a more complicated outfit.
If every lead is worth $50, optimizing for total value is identical to optimizing for total leads. You add setup friction without giving the auction engine a real value signal to prioritize.
Google Ads bidding gets simpler once you separate the report card from the auction constraint:
Once those guardrails reflect real accounting numbers, focus on continuous auction execution instead of constantly moving the goalposts.