Unit Economics

Company performance used to be read only at the company level. Revenue up and losses shrinking meant things were going well. That reading has a gap in it, though, because company-level numbers are easy to move. Cut advertising and defer hiring this quarter and the loss halves. The cost arrives next quarter, when no new customers do.

The same blind spot hides the opposite case: a company growing revenue 20 percent a month while losing money on every customer it adds. Spend $46 acquiring an account, charge $7 a month, and watch the average account leave after six months, and every account is a loss. Yet the revenue chart keeps climbing, because all that requires is more people arriving than leaving. While there's investor money in the bank the company looks healthy, and the structure only surfaces in the month the money runs low.

Unit economics moves the calculation from the company level down to one customer. Look at what one customer costs to acquire and what that one customer returns, and neither of those distortions survives. Growth is what you accelerate after this calculation works; accelerate before it does and the losses scale by the same multiple. Which is why it's the first number an investor asks about.

Subtract what one customer costs you from what that customer pays you. Three figures are enough.

1. What one customer pays. A Pagedeck account is $7 a month.
2. What that customer costs in variable expense. Document storage, payment processing fees, and a share of support handling add up to $1.50 per account. Salaries and rent don't belong here, because they don't rise when one more customer signs up.
3. What it cost to acquire them. This is customer acquisition cost (CAC) — advertising plus sales salaries divided by the customers who arrived in that period. Here it's $46 per account.

One minus two leaves $5.50 a month. Recovering $46 takes about eight and a half months, and that stretch has a name: payback period. That's the arithmetic; the verdict needs one more figure, the average length of stay. At an average of twenty-four months, an account returns $132, and $86 of that survives the acquisition cost. At six months it returns $33 against $46 spent, so every account loses $13.

The same calculation can run on a transaction instead of a customer. A marketplace looks at the fee left on one order against what shipping and payments take out of it. What the unit should be comes down to the smallest thing that repeats in that business — an account for a subscription product, an order for a marketplace, a delivery for a logistics company. Pick a unit that only occurs once and there is nothing to average, and the numbers stop being usable for planning.

The first decision it settles is whether to put more money into growth. With a payback period of eight months and an average stay of twenty-four, raising ad spend is sound: the money comes back in eight months and earns after that. When payback runs longer than the average stay, raising ad spend is a decision to lose more. What that situation calls for is revisiting pricing or reducing the share of customers who leave, and neither of those is an advertising task.

It also gets read by customer segment, because one blended average hides the segments that don't work. At Pagedeck, teams under twenty people pay back in four months while teams of five or fewer take fourteen. The blended figure of eight and a half months looks acceptable, but if the advertising is mostly bringing in teams of five or fewer, the structure is quietly degrading. Split it and the question of which segment to advertise to answers itself.

It informs product decisions too — specifically the ones touching the variable side of the cost structure, the breakdown of outgoing money into fixed and variable. Change how documents are stored so per-account cost drops from $1.50 to $0.90 and the margin per account goes from $5.50 to $6.10, which pulls nearly a month off the payback period.

"The unit economics don't work yet. Doubling ad spend doubles the loss."

Don't accelerate before the structure is sound.

"Teams of five or fewer don't pencil out. I'm cutting the ads aimed at them."

The read has to be per segment, not on the blended average.

"Don't put salaries in the cost figure. They don't go up when customers do."

Setting the standard for what counts as variable cost.

At Pagedeck, Grace redoes this calculation every quarter. The table splits into four customer segments, and each cell holds acquisition cost, monthly revenue, variable cost and average stay. Last quarter, teams above twenty people paid back in five months and teams of five or fewer took fourteen.

She reviews that table with Ravi, who uses it to decide which segments get ad budget. Last quarter the spend aimed at teams of five or fewer came down and moved to teams above twenty. Lead volume dropped, and the average payback period fell from eight and a half months to seven. Both of those movements were expected, which is the point of running the table before the budget rather than after it.

The most common mistake is reading working unit economics as profitability. If each customer earns, the sum should earn. But this calculation contains no fixed costs. At $5.50 per account, a company spending $300,000 a month on salaries and rent needs about 55,000 accounts to break even. So the number tells you whether the structure is worth scaling, not whether the company makes money. Conflate the two and the structure becomes the justification for hiring, and fixed costs grow first.

The other recurring problem is a loose definition of what goes into the cost figure. It rarely comes from anyone gaming the number; the boundary of what one customer causes is genuinely ambiguous. Support salaries behave like a variable cost, since doubling customers means hiring more support. Engineering salaries don't move with customer count, so they're fixed. Draw that line differently each time and quarter-over-quarter comparison becomes impossible. Set it once, write it down, and when it changes, recompute the previous quarter alongside it.