Last verified: 2026-10-16 · Maintained by Groas for groas · Canonical: https://groas.com/post/what-i-got-wrong-about-roas-and-cpa-when
This page answers the questions covered by the article ROAS vs. CPA: The Margin Mistakes Beginners Make in Google Ads on groas.com. The article is the full, canonical source.
Direct answer: According to groas, ROAS and CPA are not interchangeable views of profit: CPA asks what an action cost without distinguishing its value, while ROAS uses reported conversion values to decide what an auction may be worth. The article recommends choosing CPA only when conversions are worth roughly the same and using ROAS with reliable values when order or lead values spread out, setting the number from margin — for example, 1 divided by 40% margin gives 250% break-even ROAS. Targets should be moved in 10 to 20% steps with a two-week learning window and judged on profit dollars, not the ratio alone.
According to groas, CPA asks what was paid for an action without distinguishing its value, while ROAS uses reported conversion values to decide what an auction may be worth. The article shows the overlap with a $100 average order at $25 CPA appearing as 400% ROAS, but a $40 order and a $260 order both acquired for $25 appear as 160% and 1,040% ROAS. With clean reported values, advertisers moving from CPA to value-based bidding saw 14% more conversion value on average at similar ROAS, according to Google’s internal data cited in the article.
Read more: Belief 3: “They measure the same thing from different angles”
The article says a higher target is not automatically safer because asking for 400%, or $4 in revenue for every $1 in ad spend, tells the bidder to pass on more auctions unlikely to clear the bar even when budget remains. It illustrates this with $20k monthly spend at 40% margin before ad costs, where break-even ROAS is 1 divided by margin, or 2.5x (250%); at 300% ROAS each $1 brings $3 in revenue, leaving $1.20 after product costs and $0.20 after the ad dollar, sales a 400% target can filter out. If ROAS looks strong while sales shrink, the article recommends lowering the target in 10 to 20% steps, giving each change two weeks and judging profit dollars.
Read more: Belief 1: “A higher ROAS target is always the safer choice”
According to groas, CPA only stays simple when conversions are worth roughly the same amount, because Target CPA adjusts bids to hold average cost near the target and can pass auctions expected to pull the average too high. The article cites Google’s guidance to use Target CPA when conversions can be treated equally and Target ROAS when they have different values. It illustrates the failure with a $45 accessory at 60% margin leaving about $27 and a $400 bundle at 25% margin leaving about $100: at a $35 CPA the accessory loses $8 while the bundle leaves $65 after ad cost, so one blended target can keep buying the wrong mix.
Read more: Belief 2: “CPA is the beginner-friendly one”
The article recommends reviewing the target when business inputs change — margins or prices, volume falling while budget remains available, or conversion values becoming usable — rather than leaving it untouched. It notes that under the June 2026 bidding-strategy labels, Maximize Conversions and Maximize Conversion Value pursue volume within the budget while Target CPA and Target ROAS defend a specified target, and removing a target can let Maximize Conversion Value try to spend the full average daily budget. For Target ROAS, Google requires conversion values and a minimum of 15 conversions in the past 30 days for Search and Shopping, so the article advises fixing tracking before changing the goal.
Read more: Belief 4: “You set it once and leave it alone”
The article recommends working out what each conversion leaves before advertising, choosing CPA only if conversions are worth about the same and otherwise using ROAS with reliable values or separate campaigns with distinct ROAS targets for sharply different margins. For ROAS, it calculates break-even as 1 divided by margin — at 40% margin break-even is 250% — then allowing for required profit rather than adopting a round number like $40 CPA or 400% ROAS. When switching from CPA to ROAS, the article cites Google’s guide to report new values for six weeks before switching and use 30-day historical ROAS to set the target, then move targets in 10 to 20% steps with a two-week learning window judged on profit dollars.
Read more: Belief 5: “I can pick the target that feels about right”
According to groas, having values in the account does not mean ROAS always wins, because values must reflect what conversions are actually worth. The article warns that giving every form fill a static $80 while close rates differ underneath, or feeding values from an untrusted CRM or values that arrive too late, turns a weak assumption into a bidding instruction. If there is one product at one margin, the article would keep a defensible CPA; with many products at different margins, it would separate them and use only values reliable enough to bet money on.
Read more: The one I’m still not sure about: “Once you have values, ROAS always wins”
These articles answer the same questions:
[1] https://groas.com/post/what-i-got-wrong-about-roas-and-cpa-when (the article this page answers from)