ROAS, CPA, CAC and LTV: which number actually decides
Most arguments about PPC results are arguments about the metric, not the performance. A ROAS of 6 can be a win or a loss. Here is the converter between all of them, the ceiling each implies from your margin, and one table you can rewrite with your own number.
- Cost of sale and ROAS are the same thing inverted. Neither on its own tells you whether you are in profit.
- The only valid ceiling comes from your gross margin: break-even ROAS = 1 ÷ margin. At a 30 % margin that is 3.33.
- For enquiries the key difference is between CPA (cost of an enquiry) and CAC (cost of a customer). They differ by the close rate.
Nine out of ten arguments about whether PPC is working are arguments about the metric. The agency shows a ROAS of 6, the client says nothing is left over, and both are right — because they are measuring different things. This article is the converter. Five metrics, how they translate into each other, where each has its ceiling, and which to use for which decision.
Quick converter
| Metric | Calculation | Answers | Main blind spot |
|---|---|---|---|
| CPA cost per acquisition | spend ÷ conversions | What one conversion costs | Does not distinguish a €20 order from a €2,000 one |
| CAC customer acquisition cost | spend ÷ customers won | What one customer costs | For enquiries it is a multiple of CPA; the two get confused |
| ROAS return on ad spend | revenue ÷ spend | What revenue a unit of spend produced | Ignores margin. ROAS 6 is a loss in electronics, a win in cosmetics |
| Cost of sale | spend ÷ revenue × 100 | The same as ROAS, inverted | Same as ROAS. Not comparable across industries |
| LTV lifetime value | sum of margins from all a customer’s purchases | What a customer earns over their whole life | Easily inflated. The budget is paid today, the LTV arrives in years |
The relationship between cost of sale and ROAS: cost of sale = 1 ÷ ROAS. A ROAS of 5 is a 20 % cost of sale. A ROAS of 10 is 10 %. One number, two ways of saying it.
The only ceiling that holds: the one from your margin
Industry benchmarks of the “a good cost of sale is 10 %” kind are useless, because they do not know your cost structure. The valid ceiling is calculated like this:
Break-even ROAS = 1 ÷ gross margin. Break-even cost of sale = gross margin.
In other words: at a gross margin of 30 % you reach zero at a ROAS of 3.33, that is a 30 % cost of sale. Everything above that is profit from advertising, everything below is a loss. And because a company also has fixed costs, the real target has to sit above break-even.
| Gross margin | Break-even ROAS | Break-even cost of sale | Typical field |
|---|---|---|---|
| 8 % | 12.5 | 8 % | Electronics, large appliances |
| 15 % | 6.67 | 15 % | Travel, intermediation |
| 25 % | 4.0 | 25 % | Furniture, sport, hobby |
| 40 % | 2.5 | 40 % | Fashion, accessories |
| 60 % | 1.67 | 60 % | Cosmetics, supplements |
| 80 % | 1.25 | 80 % | Software, digital products |
First: “gross margin” has to be after deducting shipping, packaging, payment fees and returns. In fashion returns easily run from five to twenty per cent of orders, and a margin calculated without them is fiction.
Second: break-even is not a target. It is the line you must not fall below. The target has to sit high enough that advertising also pays for its own management, the tools and a share of fixed costs.
CPA versus CAC: the costliest misunderstanding in lead generation
For a shop the conversion is an order and there is no difference between CPA and CAC. For services the conversion is an enquiry — and there the difference is often the whole problem.
An example from construction. A campaign brings 20 enquiries a month at a CPA of €48, so €960 in total. It sounds excellent. Except that of twenty enquiries, two turn into jobs. The cost of a customer is therefore €480, not €48. Whether that is a good deal depends on the margin on the job — and at a €16,000 job with an 18 % margin it is a good deal. The point is that the decision was made on a number ten times different from the one being reported.
CAC = CPA ÷ the share of enquiries that close. If you do not know that share, you do not know the cost of a customer — only the cost of a contact.
The practical consequence: in lead generation the highest-return work is often entirely outside the ad account. Raising the close rate from 10 to 12 per cent cuts CAC by 17 per cent, and no amount of bid optimisation will do that. We cover it in B2B PPC: generating leads sales will actually take.
LTV: useful, but easily abused
The argument “we can afford a higher CPA because the customer will come back” is legitimate and often correct. But it has three conditions, without which it is just a way of approving a bigger budget:
- Repeat purchase is measured, not estimated. From your own data, by cohort of first-purchase month. Not from an industry average.
- LTV is counted in margin, not revenue. A customer who spends €1,200 over three years at a 20 % margin has earned you €240.
- You have the cash flow for it. The budget is paid today, the second purchase arrives in eight months. Companies that missed that went bust with a beautiful LTV in a spreadsheet.
The sensible compromise we recommend: use the margin on the first purchase when calculating the ceiling for advertising, and treat repeat purchases as a bonus rather than a plan. If you do want to count LTV, cap the horizon at twelve months — anything longer is guesswork.
What none of them can do, and the two numbers that belong alongside
All five metrics share one weakness: they take the ad platform’s attribution as given. That is, they assume a conversion the platform credited to itself was actually caused by it. Often it was not, and most of all in brand and remarketing, where the platform takes credit for people who would have arrived anyway.
So two more figures belong next to the metrics:
- Share of non-brand. How much of the result comes from queries that are not your own name. That is the part genuinely bringing new customers. See brand vs. non-brand.
- Incremental ROAS. How much business would not have happened without the ads. Measured by a geo experiment, not by attribution. How far it diverges from platform ROAS is shown in the article on incrementality — in one public set of 225 geo tests, brand search came out at a median incremental ROAS of 0.70×, below the line of profitability.
Which number for which decision
| Decision | Use | Do not use |
|---|---|---|
| Should this campaign keep running? | ROAS or cost of sale against break-even from margin | CTR, click volume, impression share |
| Add budget? | Marginal CPA against margin per conversion | Average CPA, overall cost of sale |
| What may I pay for an enquiry? | CAC = CPA ÷ close rate | CPA on its own |
| What target for the bidding strategy? | A target just above break-even, set per category | One target for the whole account |
| Is the brand campaign worth it? | Incremental ROAS from a geo test | Platform ROAS of the brand campaign |
| Can we pay more for a first customer? | Measured LTV in margin, 12-month horizon | LTV in revenue, horizon “forever” |
What to take away
A metric is not a neutral instrument; it determines which decisions you will even make. Run an account on ROAS and you will optimise revenue. Run it on marginal CPA against margin and you will optimise profit. Those are two different accounts, and you cannot tell them apart from the report at first glance.
The minimum set worth watching monthly: cost of sale against break-even from margin, share of non-brand, CAC for enquiries, and for larger budgets an incrementality test once a year. Five numbers, not a dashboard with thirty-five charts. In the calculator you can turn your own margin into the ceiling it implies.
Common questions
What is a good cost of sale or ROAS?
One that sits below your gross margin with enough headroom for fixed costs. Break-even cost of sale equals gross margin, so at a 30 per cent margin the zero point is a 30 per cent cost of sale, that is a ROAS of 3.33. Industry benchmarks of the “a good cost of sale is 10 per cent” kind are not usable, because they do not know your margin, shipping costs or return rate. A shop on an 8 per cent margin loses money at a 10 per cent cost of sale; a shop on 55 per cent makes very good money at the same figure.
What is the difference between CPA and CAC?
CPA is the cost of one conversion as the ad system counts it, most often an order or an enquiry. CAC is the cost of a customer actually won. For an online shop, where the conversion is the purchase itself, the two are equal. For services and B2B the gap is large: CAC equals CPA divided by the share of enquiries that turn into business. At a 10 per cent close rate the cost of a customer is therefore ten times the cost of an enquiry.
Should I target ROAS or profit?
Profit, but ROAS is a usable route to it provided the target is derived from margin and set per category where margins differ. A single ROAS target for a whole account sends budget to the categories that sell most easily rather than those that earn most. For a decision about raising budget, ROAS is not enough at all — there what counts is the marginal, not the average return.
- How to Run a Google Ads Incrementality Study, StellaSource of the median incremental ROAS of 0.70× for brand search from 225 geo tests (DTC brands, August 2024 to December 2025). A single vendor dataset, DTC only; it cannot be transferred directly to B2B.
- Explaining incrementality, PPC LandExplanation of the difference between platform-reported and incremental return.