How unit economics and cohort analysis expose what topline revenue hides in FDD — contribution per customer, retention curves, payback, LTV/CAC and cohort decay.
A target can grow revenue 40% a year and still be a value trap. That sentence should sit at the front of your mind on every consumer or subscription deal you ever work. Aggregate growth tells you the business is getting bigger; it tells you nothing about whether each customer is worth acquiring, whether last year's customers are quietly walking out of the back door, or whether the next euro of marketing spend earns its keep. Unit economics and cohort analysis are the two tools that let a Financial Due Diligence (FDD) team look underneath the topline and decide whether a growth story is real or borrowed. For subscription and consumer businesses they are no longer a nice-to-have — they are the heart of the diligence, and increasingly the first thing a sharp investment committee asks about.
Unit economics break the business down to its smallest repeatable transaction — a customer, a subscription, a basket, a delivered order — and ask a single blunt question: does that unit make money? The starting point is contribution per unit: the revenue from the unit, less the variable costs directly attributable to serving it. For an e-commerce order that means product cost, payment processing, packaging, outbound shipping, and the slice of customer-service and returns cost the order genuinely causes. For a software subscription it means hosting, third-party licences consumed, and support.
Contribution is not gross margin, and it is emphatically not EBITDA. It deliberately strips out fixed overhead so you can see the marginal economics. This distinction matters enormously in diligence. A business with negative contribution per order does not become profitable by scaling — it loses money faster, because every incremental unit adds to the loss. A business with healthy contribution but thin EBITDA has simply not yet covered its fixed base, which is a very different, and far more fundable, problem. One is broken at the unit level; the other just needs volume. Confusing the two is the single most expensive mistake an inexperienced analyst makes on a growth deal.
The practitioner's job is to rebuild contribution from the raw data room, not to accept management's version. Costs get misallocated with remarkable consistency: "one-off" launch discounts turn out to be permanent, free shipping hides inside a marketing cost centre, and a chunk of support headcount that scales directly with order volume is parked in fixed overhead. The discipline is to attribute every variable cost back to the unit that caused it, then ask whether that allocation would still hold if volumes doubled or halved. This work sits right alongside revenue quality analysis and feeds straight into how you read margins and the wider quality of earnings.
Numbers make this concrete. Take a direct-to-consumer subscription box that management describes as "growing fast and highly profitable at the unit level". Here is the average order rebuilt from the transaction data:
| Per-order economics | Management view | Rebuilt in diligence |
|---|---|---|
| Average order value | £48.00 | £48.00 |
| Product cost | (£19.00) | (£19.00) |
| Payment processing | (£1.40) | (£1.40) |
| Fulfilment & packaging | (£4.00) | (£5.60) |
| Outbound shipping | (£0.00)* | (£4.20) |
| Returns & support | (£1.10) | (£3.10) |
| Contribution per order | £22.50 (47%) | £14.70 (31%) |
*Management treated shipping as a marketing "acquisition cost", not a cost of the order.
The rebuild does not make the unit unprofitable — £14.70 of contribution is real — but it deflates the story. Management's 47% contribution margin was closer to 31% once shipping and the true returns rate were pulled back to the order that caused them. That 16-point gap changes the CAC payback maths, the fundability of the growth, and ultimately the multiple. The number you can defend line by line is worth ten times the number in the management deck.
A cohort is a group of customers bound together by when they were acquired — the January 2024 cohort, the Q3 2023 cohort. Cohort analysis tracks each group across its life: how much they spend, how many remain, and how the economics evolve month by month after acquisition.
This is the single most powerful technique for seeing through aggregate growth. Consider a business adding customers fast. The topline rises smoothly. But split revenue by cohort and you might find that each successive cohort spends less over its lifetime than the one before, or that customers churn heavily after month three. The apparent growth is entirely a function of ever-larger acquisition spend masking a leaky bucket. Stop the spend and revenue falls off a cliff — which is exactly what a buyer discovers eighteen months after completion if the diligence team missed it.
| What aggregate revenue shows | What cohort analysis reveals |
|---|---|
| Topline up 30% | New cohorts smaller in lifetime value than old ones |
| Stable blended churn rate | Churn front-loaded; survivors sticky, newcomers not |
| Rising ARPU | Driven by a shrinking, premium base — not broad health |
| "Recurring" revenue | A third of it lapses within twelve months |
The blended average is where growth stories go to hide. If a management pack shows you only company-wide churn, ARPU and revenue, assume it is because the cohort view is less flattering — and ask for it explicitly.
Plot the percentage of a cohort still active against months since acquisition and you get a retention curve. Its shape is diagnostic, and reading it is a skill worth practising until it is instinctive:
The flattening point is what links cohort work to recurring revenue analysis: it tells you how much of today's revenue base will still be there next year with no new sales at all. A team that can quantify that plateau, and defend the assumption behind it, has done the most valuable single piece of work on the deal.
Two ratios summarise the acquisition engine, and both must be handled with care.
CAC payback is the number of months of contribution it takes to recover the cost of acquiring a customer. A short payback — say under twelve months — means the business largely self-funds its growth: acquire a customer, earn the cost back inside a year, recycle the cash into the next cohort. A long payback means the business must fund growth from outside, which makes it fragile to a downturn, a rise in acquisition costs, or a financing squeeze. Payback is the ratio that tells you how fundable the growth is.
LTV/CAC compares the lifetime value of a customer to the cost of acquiring them. It is a useful sanity check and the single most abused metric in any data room. Lifetime value depends entirely on assumed retention and assumed contribution margin — flex either and you can make LTV/CAC say almost anything you like. A diligence team should never repeat a management LTV figure without rebuilding it from the actual cohort data and stress-testing the retention tail. Treat a suspiciously round, unsupported LTV/CAC of "3x" as a prompt to dig, not a reassurance — it is exactly the kind of thing that belongs on a sensible list of red flags.
Strong blended unit economics can conceal a dangerous truth: a handful of customers carrying the entire cohort. Always decompose contribution by customer within a cohort, not just the average. If the top 5% of customers generate the bulk of lifetime value, the business is really a customer concentration story dressed up as a scalable one. The retention curve looks healthy because the whales stay put — but lose one whale and the cohort economics collapse, and no amount of new small customers replaces it. This decomposition is quick to run and repeatedly catches problems that the averages smooth over.
Pulling it together, the deal thesis usually rests on a single claim: this business compounds. Unit economics and cohorts let you test that claim directly rather than take it on trust.
This is most acute in subscription and consumer businesses, where revenue is assembled from many small, repeating units and the bridge from customer behaviour to enterprise value is short and visible. In a B2B software target the cohorts are fewer and larger, so gross and net revenue retention dominate; in a consumer brand the cohorts are vast and noisy, so the focus shifts to repeat-purchase rates and contribution after delivery and returns. The technique is identical; only the emphasis follows the unit.
Interviewers use this topic to separate candidates who memorise definitions from those who think like investors. Expect a deliberately open prompt: "A company has grown revenue 50% this year. What would worry you?"
A strong answer goes straight to the engine rather than reciting textbook metrics:
"Topline growth on its own tells me almost nothing, so my first move would be to break it down by cohort. If newer cohorts have lower lifetime value than older ones, or if retention decays without ever flattening, then the growth is just acquisition spend papering over churn — stop spending and revenue falls away. So I'd rebuild contribution per customer to confirm the unit is genuinely profitable before fixed costs, being careful to pull shipping and returns back onto the order rather than leaving them in marketing. Then I'd look at CAC payback to see whether the growth self-funds or leans on external capital, because a long payback makes the whole thing fragile. I'd also decompose each cohort by customer, because a strong average can hide two or three accounts carrying everything. If the cohorts hold up and retention plateaus at a meaningful level, I'd treat the growth as durable and quality earnings; if they don't, I'd flag it as a quality-of-earnings risk and it should come off the multiple."
That answer signals you understand the difference between a business that is growing and a business that is compounding — which is exactly what the work is about, and exactly what an interviewer is listening for.
Aggregate revenue is a headline; cohorts are the story underneath it. A business that grows because each new customer is worth having and each old customer keeps coming back is a compounding machine and deserves its multiple. A business that grows only because it spends ever more to replace customers leaking out of the back door is a treadmill, and the day the spending stops is the day the revenue does too. Unit economics tell you whether the unit is worth having; cohorts tell you whether the customer stays. Master both, and you will never again mistake a leaky bucket for a growth story — which, on a consumer or subscription deal, is the difference between a good investment and an expensive lesson.
The Transaction Services Interview Programme (€119.99, one-time) includes a dedicated module on unit economics and cohort analysis, with a worked contribution rebuild, annotated retention curves, and a framework for stress-testing management's LTV/CAC and payback claims live in a data room. Enrol today.
Hundreds of candidates prepared their interviews with this programme. Those who landed the role have one thing in common: they worked the cases before walking into the room.
