A company can grow revenue every month, raise round after round, and still be quietly unprofitable on every order it ships, because it is paying more to acquire and serve each customer than that customer is worth. Unit economics is the lens that catches this early, while it is still a spreadsheet problem and not an obituary.

The quick version

  • Unit economics is the profit (or loss) on a single unit, usually one customer, once you count what it costs to win and serve them.
  • The three numbers that matter: CAC (what you spend to acquire a customer), contribution margin (what's left from their revenue after variable costs), and LTV (the lifetime margin you expect from them).
  • The well-known rule of thumb is LTV at least 3× CAC, with acquisition cost recovered in under a year, but it was written for mature subscription businesses, not for every company at every stage.
  • The number is only as honest as its assumptions. Lengthen the customer lifetime in a model and you can justify almost any acquisition cost, on paper.

The idea in depth

Unit economics is deceptively simple: pick the "unit" your business repeats, a customer, an order, a ride, a subscription, and work out whether that unit makes or loses money on its own terms. If a unit is profitable and you have a way to make more of them, growth compounds value. If a unit loses money, growth just compounds the loss. That is the whole game, and most of the sophistication is in measuring it honestly.

Three quantities carry the weight. Customer Acquisition Cost (CAC) is the total sales and marketing spend over a period divided by the customers it won. Contribution margin is the revenue a customer generates minus the variable costs of serving them, the costs that rise with each extra unit (cost of goods, payment processing, support, hosting), as distinct from fixed overhead like rent or your core engineering team. Lifetime Value (LTV) projects that contribution forward across the whole relationship, not just the first sale.

flowchart LR
  A(["Revenue from one customer"]) --> B(["Minus variable cost to serve"])
  B --> C(["= Contribution margin"])
  C --> D(["× expected lifetime"])
  D --> E(["= Lifetime value (LTV)"])
  F(["Sales & marketing spend ÷ customers won"]) --> G(["= CAC"])
  E --> H{"LTV vs CAC"}
  G --> H
  H --> I(["Profitable unit if LTV comfortably > CAC"])
The unit-economics chain, from one customer's revenue to the LTV:CAC comparison. Leaders Loop

Why the LTV:CAC ratio became the headline number

The reason unit economics has a standard scorecard is largely one body of work: David Skok's SaaS Metrics 2.0, published on the For Entrepreneurs blog. Skok set out two guidelines that have become near-universal. First, LTV should be at least 3× CAC, "The best SaaS businesses have a LTV to CAC ratio that is higher than 3, sometimes as high as 7 or 8." Second, you should recover the cost of acquisition quickly: many of the strongest businesses, he notes, "recover their CAC in 5–7 months." Crucially, Skok also insists LTV be calculated on margin, not revenue, in his detailed definitions he writes that "to truly get an accurate picture of LTV, it is important to also take Gross Margin into consideration." A customer's top-line revenue is not the prize; the margin you keep is.

Investors lean on the same ratio because it predicts how efficiently a company turns marketing dollars into durable value. Andreessen Horowitz frames 3× as "a rough benchmark of a consumer company's financial health," and argues the payoff is real: in their analysis, "improving your LTV:CAC from 2× to 3× can nearly triple your valuation" (Sullivan & Immerman, 2023). The logic behind the 3 is tidy: roughly a third of lifetime value pays back acquisition, a third covers the cost of serving the customer, and a third is profit you can reinvest.

So the move is: stop reporting customers and revenue as if they were health. Build the smallest honest version of this chain for your business, pick the unit, subtract variable costs to get contribution margin, divide your real sales-and-marketing spend by customers won, and compare. If you sell to several segments, do it per segment; the blended average almost always hides a segment that is bleeding and one that is carrying it.

Where the model breaks down

The honest limitation is that LTV is a forecast wearing the costume of a fact. Its most sensitive input, how long a customer stays, is the one you know least about, especially early on. Bill Gurley made this the centre of his essay "The Dangerous Seduction of the Lifetime Value (LTV) Formula" (2012). The formula is seductive, he argues, precisely because of "its simplicity and certainty", and that certainty is borrowed. Relax one assumption (customers will surely stay longer; churn will surely fall) and the model cheerfully tells you to spend more on acquisition. A healthy LTV, Gurley adds, does not by itself "create sustainable competitive advantage", it describes a business that retains and monetises customers today, not one that will keep winning when a rival floods the same acquisition channels.

The 3× rule has a second catch the headline hides: it was derived from mature, steady-state subscription companies with stable churn and multi-year lifetimes. Applied to a pre-product-market-fit startup with three months of data, the "lifetime" in LTV is a guess, and a flattering one. Treat the ratio as a question to interrogate ("what would have to be true about retention for this to hold?"), not a target to hit.

So the move is: never quote an LTV without quoting the churn or retention assumption underneath it, and pressure-test it. Re-run the numbers with a pessimistic lifetime and a realistic discount on future cash. If the unit only works on the optimistic assumptions, you don't have unit economics, you have a hope with a decimal point.

A worked example

Illustrative figures, a stand-in business, not a benchmark. Imagine a small subscription app, "Tideline," charging £20 a month. Of each £20, payment fees, hosting and support eat £6, so the contribution margin is £14 per month, or a 70% margin. Marketing spends £30,000 in a quarter and wins 1,000 customers, so CAC is £30.

The first read looks alarming: CAC (£30) is more than the first month's margin (£14). But the customer recurs. At £14 of margin a month, Tideline recovers the £30 acquisition cost in just over two months, comfortably inside the under-12-months guideline. If the average customer stays 24 months, lifetime contribution is 24 × £14 = £336, giving an LTV:CAC of about 11×. Wonderful, if they stay 24 months.

Now apply the limitation. Suppose new data shows the average customer actually leaves after 8 months. Lifetime contribution falls to 8 × £14 = £112, and LTV:CAC drops to about 3.7×, still defensible, but a third of the rosy story. Halve retention again to 4 months and you're at £56, an LTV:CAC of 1.9×: the unit no longer pays for itself once you account for the cost of serving and the cost of capital. Same product, same price, same CAC, the entire verdict turned on one assumption you didn't control. That is the lesson worth carrying into your next board deck.

flowchart TD
  S(["Same product: £14/mo margin, £30 CAC"]) --> A(["Lifetime 24 mo → LTV £336 → 11× CAC"])
  S --> B(["Lifetime 8 mo → LTV £112 → 3.7× CAC"])
  S --> C(["Lifetime 4 mo → LTV £56 → 1.9× CAC"])
  A --> V{"Verdict swings on one assumption: how long they stay"}
  B --> V
  C --> V
The retention assumption does the heavy lifting: three plausible lifetimes, three very different verdicts on the same unit. Leaders Loop

Frequently asked questions

Is a higher LTV:CAC ratio always better?

No, counter-intuitively, a very high ratio can be a warning. An unusually high LTV:CAC often means you are under-investing in growth: you have found profitable customers but are not acquiring enough of them. A 10× ratio with flat growth may be a business leaving customers (and market) on the table. The aim is profitable units and a channel to make more of them, not the largest possible number.

Should LTV use revenue or margin?

Margin. Skok is explicit that gross margin must be folded into LTV; a £100/month customer who costs £90/month to serve is not a £100 customer. Use contribution margin, revenue minus the variable cost to serve, so the number reflects money you actually keep.

What's a good CAC payback period?

The widely-cited guideline is recovering CAC in under 12 months, with the strongest subscription businesses landing around 5–7 months. Faster payback means less working capital tied up and less exposure if churn rises. But like the 3× rule, this came from mature SaaS; a business with cheap capital and very long customer lifetimes can rationally run a longer payback.

Does unit economics apply to non-subscription businesses?

Yes, though the "unit" changes. For e-commerce it might be one order's contribution margin against the cost to acquire that order; for a marketplace, the margin per transaction net of incentives. The discipline is identical: define the repeating unit, count only the costs that vary with it, and ask whether it pays for the cost of creating it.

When should a startup start worrying about this?

Track it early, judge it late. Before product-market fit your retention data is too thin for LTV to mean much, so measure CAC and contribution margin honestly but hold the ratio loosely. The danger isn't ignoring unit economics early, it's scaling spend on a unit you've never proven, then discovering the loss at volume.

Related in the Toolkit

Where to go next