Three numbers, one decision
The question founders actually ask is simple: should we increase media spend, keep it flat, or cut it? Three numbers answer it, and they must be defined precisely or they lie. CAC measures what a new customer costs. ROAS measures what advertising returned relative to its cost. Contribution margin measures what the business keeps from each sale before fixed costs. Spend decisions follow the relationship between the first two and the third.
Definitions that survive scrutiny
Ambiguous definitions are the most common reason two teams disagree about the same number. Use these, and write down the exact definition next to the dashboard so it cannot drift.
- Media CAC = media spend divided by the number of new customers acquired in the same period. The media-only figure is the right input for spend decisions because it is comparable with the contribution each customer generates; a fully loaded acquisition cost that also includes team and tooling is a separate efficiency view, useful quarterly but misleading inside weekly decisions. New means new: a repeat purchase from an existing customer is not a new customer.
- ROAS (return on ad spend) = revenue attributed to advertising divided by the ad spend that produced it. It is a revenue ratio, not a profit ratio: it says nothing about what the revenue cost to produce. The word attributed matters: ROAS is only as good as the attribution model behind it.
- POAS (profit on ad spend) = contribution after media spend divided by media spend. In the example below it is 20,000 / 22,000, about 0.91. Contribution before media over spend would be 42,000 / 22,000, about 1.91, but that is a different ratio and it is not POAS as defined here. Neither figure is net company profit: fixed costs and taxes sit outside contribution and must be funded from what remains.
- MER (marketing efficiency ratio) = total revenue divided by total media spend across all channels, including organic and existing-customer sales in the revenue. It is a store-wide health check. It only equals an attributed ROAS when every sale in the store is attributed to the media, as in the deliberately simplified example below. In a real store with repeat customers and organic demand, MER and ROAS answer different questions and will differ.
A worked example in euros
Take a hypothetical single-channel cohort. Assume the store acquired 500 genuinely new customers in one month, that all 22,000 euros of media spend is attributed to acquiring them, and that these 500 customers generated all 100,000 euros of net sales in the period. This is a teaching scenario: it excludes existing customers, repeat purchases and organic-only orders, which in real operations are measured separately. All figures are monthly and exclude VAT, which is a simplifying assumption here: VAT treatment and recovery vary by jurisdiction and by the business's status, so do not copy these figures into a tax calculation.
| Line | Amount |
|---|---|
| Net sales from the 500 new customers (already net of refunded revenue) | 100,000 |
| Variable product cost | −42,000 |
| Fulfillment and shipping | −9,000 |
| Platform and payment fees | −4,000 |
| Returns handling, reverse logistics and unrecoverable product | −3,000 |
| Contribution before media | 42,000 |
| Media spend | −22,000 |
| Contribution after media | 20,000 (20% of net sales) |
Reading the example
The arithmetic, step by step: 100,000 / 500 = 200 euros of net sales per new customer. 42,000 / 500 = 84 euros of contribution before media per new customer. 22,000 / 500 = 44 euros of media CAC. 20,000 / 500 = 40 euros of contribution after media per new customer. The attributed ROAS is 100,000 / 22,000, about 4.55. The MER in this scenario is the same 4.55, but only because every sale is attributed to the media by construction.
The average media CAC of 44 euros sits below the 84 euro pre-media contribution per customer. That comparison alone does not justify more spend. The decision needs the marginal CAC of the next euro of budget, the conversion capacity of the page, and the stock and operational capacity to absorb more volume without degrading delivery. An average below the ceiling with no marginal headroom is a plateau, not an opportunity.
The formula behind these numbers, and how to build it per SKU and per channel, is in Contribution margin: the number that runs your business.
The decision rules
With the three numbers defined and computed honestly, the decision stops being a debate about ambition and becomes a comparison against a ceiling.
- Increase spend while the marginal CAC stays below the maximum defensible media CAC and the operation can fulfil the additional volume without degrading delivery or stock cover.
- Hold spend when CAC is close to the ceiling and the path to a higher ceiling is a margin or retention improvement, not more budget. Fix the ceiling before testing the limits.
- Cut or redirect spend when CAC exceeds the ceiling and the cause is structural: the product's contribution is too thin, the returns are too high, or the audience only responds at a price that destroys margin.
- Judge cohorts, not months. A first-order CAC above the ceiling can be rational when the cohort's repeat rate proves a higher lifetime ceiling. That proof comes from data, not from hope.
How the model breaks
Four failure modes recur in brands that have the definitions right but the conclusions wrong. Assumed retention: counting repeat purchases that have not happened. Blended averages: one aggregate CAC hiding a losing channel or product. Attribution windows: platform-reported ROAS measured over different windows than the contribution data, so the two numbers never describe the same sales. And the quiet one: returns and failed deliveries recognised late, so the month that looked profitable is restated a quarter later.
The defence is boring and effective: one definition document, one data source per number, and a monthly reconciliation where marketing and finance look at the same table before the budget conversation happens. If you want a second pair of eyes on that table, our 30, 60 and 90-day programmes start exactly there.
Direct answers
What is a good ROAS?
There is no universal good ROAS. The good ROAS is the one that keeps the media CAC below the maximum defensible media CAC given your contribution before media. A brand with 70% contribution can profit at a ROAS of 2. A brand with 15% contribution may lose money at a ROAS of 6. Compute the ceiling first, then judge the ratio.
Should CAC include salaries of the marketing team?
The media CAC used for spend decisions includes media spend only, so it can be compared with the contribution ceiling. A fully loaded acquisition cost, including salaries and tooling, is a useful quarterly view of efficiency, but mixing it into weekly decisions makes the ceiling move with hiring and confuses the question of whether the media itself works. Media CAC is not the only CAC: it is the version relevant to this decision.
Official references
- Google Ads: attribution models Official Google Ads help page on attribution models; checked October 2026.
Attribution windows and reporting defaults differ per platform and change over time. Align your reporting windows with your contribution data before comparing numbers across tools.