Skip to content
Build80 minutesUnit 4 of 8 · Lesson 2 of 3

Does this feature make money?

A still from the film for Does this feature make money?
The assistant leaves beta next month. It costs about ₹9 a month per user in model calls, and the subscription is ₹199. Finance wants a number, not an opinion.

Four numbers carry it

  1. Four numbers carry it
    01

    Four numbers carry it

    What a customer costs to win, what they pay, what they cost to serve, and how long they stay. Everything else is decoration.

  2. Months to pay back
    02

    Months to pay back

    The question executives ask first: how many months until a customer repays what it cost to win them? Under twelve, most subscription businesses breathe easy.

  3. AI broke the old rule
    03

    AI broke the old rule

    Software used to cost almost nothing to serve. Now a feature people love can lose money on every user, every month.

  4. Stress it
    04

    Stress it

    What if the model gets cheaper? What if people use it twice as much? A call that survives both is one you can defend.

Try the concept

What does one customer contribute?

Interactive teaching example · All numbers and tickets are fictional.

Monthly revenue: ₹300Coral: variable cost · Teal: positive contribution₹180 contribution · 5.0 months simple payback

Payback = ₹900 acquisition cost ÷ monthly contribution, only when contribution is positive. Fixed costs, churn, taxes, and cash timing are excluded from this teaching model. The numbers are made up, so they will not match Sona.

How Dropbox handled it

Dropbox
Public commentary. No affiliation.
  1. The situation

    Dropbox was paying far more to win a customer through search ads than that customer paid in a year.

  2. What they did

    It replaced much of that spend with a referral programme that gave both people free storage. Extra storage cost Dropbox very little, so paying in product was cheap where paying in cash was ruinous. Signups grew sharply.

  3. Why it matters to you

    The insight was not that referrals are good. It was that serving one more user cost almost nothing. The same tactic is brilliant at one company and fatal at another.

Doing it with AI, and where it breaks

The move

Build the model in a sheet, then have a model stress test it: which assumption, if wrong by 20 percent, changes the conclusion?

The trap

Never let a chat window do the arithmetic. Make it write the formulas and compute them yourself in the sheet.

Prompt worth stealing

Here is my unit economics sheet with its formulas. Which single assumption, if it were wrong by 20 percent, would change the go or no go call? Name the cell and the formula. Do not compute any results yourself.

Build it

The artefact

A unit economics sheet with a go or no go recommendation.

Google Sheets

0/4

The worked solution

Stuck, or want to compare?

There is a worked solution. Try the build first. Reading it before you have attempted anything is the fastest way to learn nothing.

Checkpoint

If you did the build, these take two minutes. If you cannot answer one of them, that is the part to go back to.

Moving on marks this lesson complete. Finish the build first, it is the part that counts.

Previous lessonWhere the money actually comes from