Marketing ROI is measured against the margin a customer leaves, not the revenue a platform claims. That means four numbers: blended customer acquisition cost, contribution margin after acquisition, payback period, and lifetime value against CAC. Platform ROAS is a steering tool inside a channel, and a poor scoreboard for the business. This covers each number, how to calculate it honestly, and the attribution traps that make reporting look better than the bank account.
Why platform ROAS is not ROI
Return on ad spend compares revenue to the money spent on ads. It leaves out what the product cost to make, what it cost to ship, the payment fees, the returns, and everything you spend on marketing that is not media. Two brands with identical ROAS can have completely different outcomes, because one keeps half of every order and the other keeps a third.
The correction is to set targets on margin. Your breakeven ROAS is one divided by the share of an order that survives product, fulfillment, fees and returns. At 50%, you break even at 2x. At 35%, breakeven sits close to 3x, which means the 3x that felt like a win was the business working for free. Any ROAS target that was not derived from your own margin was borrowed from someone else’s business. Measuring against margin is what makes performance marketing answer for profit rather than for activity.
Blended CAC: the number that matches the bank account
Blended customer acquisition cost is total marketing spend divided by new customers acquired, over the same period. Total means everything: media, agency or team cost, creative production, the tools. New customers means new, because counting repeat orders in the denominator makes acquisition look cheaper than it is and hides a retention problem behind an acquisition metric.
This number is unglamorous and hard to argue with. It ignores which channel deserves credit and asks a simpler question: for every euro that left the company, how many new customers arrived. When blended CAC and platform-reported CAC disagree, blended is the one reconciled against reality. Channel-level numbers still matter for steering inside a channel, and they answer a different question from the one the finance team is asking.
Margin after acquisition, layer by layer
The decisive number is what a customer leaves behind once everything is paid. Revenue minus product cost gives the first layer. Subtract shipping, fulfillment, payment fees and returns for the second. Subtract acquisition cost for the third, which is the one that decides whether scaling builds profit or losses. We work through the layers with an example in what is contribution margin (CM1, CM2, CM3).
Reading it per customer rather than in aggregate is what makes it actionable. Aggregate profit tells you the business worked last month. Margin per customer tells you whether the next customer is worth buying, which is the decision you actually face every day in an ad account.
Payback: how long the money is gone
Payback period is how long it takes a customer to return what you spent acquiring them. It is the number that governs how fast you can grow without running out of cash, and it is invisible in ROAS. A brand with a healthy lifetime value and a nine-month payback can be profitable on paper and still fail, because growth consumes cash faster than it returns it.
Shorter payback means you can recycle the same euro more times a year, which compounds. This is why first-order profitability matters so much for DTC brands without deep funding: if the first purchase covers acquisition, growth funds itself. If it does not, you are financing every new customer and betting on a second purchase that may never come.
Lifetime value against CAC
Lifetime value tells you what a customer is worth in margin over their whole relationship with the brand, and comparing it to CAC tells you whether the model works at all. The trap is calculating LTV on revenue, which inflates it, or on optimistic assumptions about repeat purchases that the cohort data does not support.
The honest version uses margin, and it uses cohorts: take customers acquired in a given month and follow what they actually spent over the following months. Cohort curves tell you whether retention is improving or decaying over time, which an average LTV number cannot. They also tell you when to stop trusting a projection. If your model assumes four purchases a year and the cohort data shows one and a half, you are not measuring, you are hoping. This is where retention stops being a separate topic and becomes the input that decides how much you can afford to spend.
The attribution traps
Every platform reports on its own conversions and each claims a sale it touched, so the totals overlap. Sum what the platforms say they delivered and you will exceed the orders in the shop, because the customer who saw a social ad, searched the brand, then bought is counted more than once. Nobody is lying: each is answering “did I contribute” rather than “was I the reason”.
Retargeting flatters itself most, because it takes credit for people already on their way to buying. The way to know is to withhold: hold out a region or an audience, run the channel everywhere else, and compare the difference. The practical version most brands can run is simpler. Watch what blended cost per new customer does when a channel scales up or turns off. If nothing moves, that channel was collecting credit rather than creating demand. None of this works if the tracking underneath is broken, which is its own problem and worth solving before you argue about attribution models.
What to actually put on the dashboard
A short list beats a wall of charts. Blended CAC. Margin after acquisition per customer. Payback period. LTV to CAC by cohort. Marketing efficiency ratio, meaning total revenue divided by total marketing spend, as a fast blended sanity check. Then, inside each channel, the platform numbers you use for steering, clearly labelled as such so nobody confuses them with the business result.
Review on a rhythm rather than continuously. A monthly cycle of what you predicted, what happened, and what changes as a result produces better decisions than watching a live dashboard and reacting to noise. Daily numbers are for spotting breakage, not for judging performance.
Where measurement goes wrong
Optimizing to ROAS without knowing the margin behind it. Counting repeat customers in CAC. Adding up platform-reported conversions as if they were orders. Calculating LTV on revenue and on assumptions the cohorts do not support. Ignoring payback until cash gets tight. Measuring channels in isolation and never checking the blended number that finance sees. Each one makes reporting more comfortable and decisions worse.
The deeper mistake is measuring acquisition carefully while never asking whether acquisition is the constraint. If the business leaks at conversion, pricing or retention, better measurement of ad spend just documents the problem more precisely. That question comes first, through a diagnosis, and the measurement follows it.
Frequently asked questions
What is blended CAC and how do I calculate it?
Total marketing spend divided by new customers acquired in the same period. Total means media, team or agency cost, creative and tools. Count new customers only, since including repeat orders makes acquisition look cheaper than it is.
Is ROAS a good measure of marketing ROI?
It is useful for steering inside a channel and weak as a business scoreboard, because it ignores product cost, fulfillment, fees, returns and non-media marketing cost. Targets should be derived from your own margin.
Why do the ad platforms report more conversions than I have orders?
Each platform claims sales it touched, so the same customer gets counted by more than one. Each is answering whether it contributed rather than whether it was the reason. Blended numbers are what reconcile with the shop.
How do I know if a channel is actually incremental?
Hold out a region or audience and compare against everywhere else, or watch what blended cost per new customer does when the channel scales up or switches off. If nothing moves, the channel was taking credit rather than creating demand.
Not sure what a customer really costs you?
We rebuild the numbers so decisions rest on margin rather than platform reporting. Request a strategy call and we will look at your real acquisition math.
