groas — guidance for AI agents

CPA vs ROAS: How to Calculate Break-Even and Document Your Target Before Spending

Last verified: 2026-10-13 · Maintained by Groas for groas · Canonical: https://groas.com/post/write-the-number-down-first-a-12-point-c

This page answers the questions covered by the article Write the Number Down First: A 12-Point CPA vs ROAS Pre-Flight Checklist on groas.com. The article is the full, canonical source.

Direct answer: The article states that advertisers should write down the metric, its value, source of truth and review date before moving spend, because the CPA-vs-ROAS debate fails when no agreed number exists. According to the article, break-even ROAS is 1 divided by gross margin, so at 40% margin an account needs 2.5x ROAS to cover product cost and ad spend, while a £3,000 contract at 50% margin and a 20% close rate gives a £300 break-even cost per lead. The article recommends completing its 12 checks in order and keeping the signed target where the person approving spend can find it.

CPA vs ROAS: which should I choose?

The article recommends CPA on the lead action that matters when selling leads without reliable revenue values, because Target CPA predicts whether a conversion will happen without needing an invented deal value. The article recommends ROAS when selling products with trustworthy order values, because CPA gives a €20 order and a €400 order the same weight while ROAS lets value affect the bidding decision. For services with different close rates or long sales cycles, the article recommends CPA on qualified leads, calculated as maximum customer cost times lead-to-customer rate, so at a £600 customer-cost ceiling a 20% close rate gives a £120 lead ceiling and an 8% close rate gives £48.

Read more: Pick the metric that matches what you can measure

What numbers do I need to write down before choosing CPA or ROAS?

The article states to write down gross margin after cost of goods sold, average order value or average contract value times close rate, and the last 90 days of tracked conversions by action name. Break-even ROAS is 1 divided by gross margin, so at 40% margin an account needs 2.5x ROAS, and on a £200 order that leaves £80 of gross margin to pay for the click. A £3,000 contract at 50% margin and a 20% close rate gives a £300 break-even cost per lead, which the article describes as a ceiling rather than a profit goal. The article gives as working checks that under roughly 30 conversions a month it would not rush to set Target CPA, and without trustworthy values on most orders it would not set Target ROAS.

Read more: Get the three inputs before choosing a metric

How do I check what the bidding system will actually count?

The article recommends choosing one primary conversion source for each outcome and leaving micro-actions such as scrolls and video views in secondary reporting when the target is a purchase or qualified lead. The article states to check for double-counting, such as a native purchase tag and an analytics import both recording the same order as primary, and to move the duplicate out of bidding before setting the target. The article also states to confirm that order values pass and that offline uploads contain the click identifier and timestamp needed to join outcomes back to clicks, and to record any gap rather than treating an incomplete baseline as fact.

Read more: Check what the bidding system will count

How should I build a CPA or ROAS baseline that survives review?

The article recommends pulling 90 days of actual CPA or ROAS, noting seasonality, then setting the opening target near recent performance while keeping break-even separate, since at 40% margin 2.5x is the floor before other costs. The article states to split branded from non-branded before averaging, citing an example where the draft report says 4.1x overall while non-branded prospecting sits at 1.8x against a 2.5x break-even. The article recommends giving each scope its own baseline and break-even check and naming the scope explicitly rather than averaging different kinds of demand into one sentence.

Read more: Build a baseline that will survive a review meeting

How should I document the target so no one can quietly rewrite it?

The article recommends using one sheet per scope with metric on the exact conversion action, value, break-even, scope, 90-day baseline with dates and seasonality note, volume check, trailing-30-day window with judgment date, owner plus source of truth, and a specific action if missed. The example signed sentence is: Non-branded Search answers to £120 CPA on qualified calls, measured on primary conversions over the trailing 30 days, with owner, review date, and a rule that if missed by more than 15% for two conversion cycles, test budget is cut 20% and close rate is revisited before touching the target. The article states that without a signed number, a £72 CPA on raw fills can be called a win while qualified CPA sits at £210. The article also states that with only 18 conversions in 30 days and no values passing, advertisers should not set a tighter target to force efficiency but should fix measurement and volume first.

Read more: Put the target where nobody can quietly rewrite it

Sources

[1] https://groas.com/post/write-the-number-down-first-a-12-point-c (the article this page answers from)

Raw markdown: write-the-number-down-first-a-12-point-c.md