The Numbers That Matter: Reading Unit Economics in Diligence

Once a deal has cleared the judgement questions, demand, team, market and real traction, it earns the part most people think of first: the numbers. But reading a startup's metrics well is not about knowing more ratios. It is about knowing the few that matter at this stage and, more importantly, seeing through the way they are presented to the unit economics underneath. The same company can look fundable or fragile depending only on which version of its numbers you accept.

A few numbers, not sixteen

There is no shortage of metrics to track; Andreessen Horowitz's well-known list runs to sixteen, and a full diligence pack can hold dozens. But at the early stage, where most of the numbers are still small and noisy, a handful carry almost all the signal, and they are the same ones a founder reaches for in a B2B raise: how fast revenue is growing, whether existing customers expand, how much of each dollar survives delivery, how long it takes to earn back the cost of a customer, and how much cash the company burns to produce its growth. Master those, and you can read most early B2B companies without drowning in the rest.

The difference between a careful investor and a careless one is not which metrics they know, but whether they reconstruct each one from the raw data instead of accepting the headline. Every number on the list below has a clean version a founder will present and a messier version that is closer to the truth, and the gap between them is exactly where the diligence happens.

Where the numbers get dressed up

Three adjustments hide more weak businesses than any others, and a disciplined investor checks for each by default. The first is revenue quality: a company with heavy setup or services work can present a healthy-looking ARR figure that is partly one-time money, which does not recur and should not be valued as if it does. The second is gross margin: when hosting, onboarding or support costs are quietly placed below the gross-margin line, a services-like business can wear software-like economics, and the multiple you would pay should not follow. The third is the word 'adjusted', which can be reasonable or can be where inconvenient costs go to be forgotten; the right move is always to ask to see the unadjusted version next to it.

The CAC line hides the most at the early stage, because a 'blended' number averages cheap word-of-mouth customers together with expensive paid ones and makes acquisition look efficient. The question that cuts through is simple: what does it cost to acquire the next customer, the marginal one, through the channel the company actually plans to scale. If that number is much worse than the blended one, the growth on the chart is being bought, not earned, and the burn multiple is the place that truth eventually shows up.

What the numbers are really for

It helps to remember what you are trying to learn. The numbers are not a grade; they are evidence about one question, whether this company turns each dollar of investment into more than a dollar of durable, efficient growth. Growth and retention tell you there is real demand. Gross margin and CAC payback tell you the economics can work. The burn multiple ties them together into a single read on efficiency. An early company will rarely be strong on all of them, and that is fine; what you are looking for is a coherent story where the weak numbers have a credible reason and a plan, not a polished surface with an unexamined hole underneath.

The takeaway

Reading the numbers well is less about knowing more metrics than about reconstructing a few of them honestly from the raw data. Hold to the handful that matter early, growth, retention, gross margin, CAC payback and the burn multiple, and check each for the standard dressing-up: one-time money in ARR, costs hidden below the gross-margin line, and blended CAC masking expensive acquisition. The headline number is what a founder chooses to show. The version you rebuild yourself is the one you can underwrite, and the discipline of rebuilding it is what separates an investor who reads the numbers from one who is reassured by them.

It is the read EvoScale Capital runs on every B2B submission, rebuilding the unit economics rather than accepting the deck, alongside people who have lived these cost lines from the inside and know where the real ones hide. Once the numbers are honest, the next questions are about price: what the company is worth and what you give up to own a piece of it.