Skip to content
Tutorials Center
My learning

Topics

Lesson 36 of 38 ยท Projects and Practice

Project: Build a Unit Economics Model

A hands-on project: model CAC, LTV, payback and profit by channel, run what-if scenarios and find the budget split that keeps growth profitable.

Marketing decisions are easier when you can see the money flow in one place. In this project you build a unit economics model for a fictional online subscription business, "FreshBox", which sells a monthly box of cooking ingredients. You will compute margins, CAC, LTV and payback by channel, test scenarios and decide where the next budget unit should go. All numbers are invented.

The inputs

Collect or estimate these, and write down the source for each:

  • Price and margin: monthly price, cost of goods, delivery and payment fees.
  • Retention: share of subscribers who cancel each month (churn).
  • Channel costs: spend, clicks, sign-up conversion rate and any extra costs by channel.
  • Fixed marketing costs: tools, team time, agency fees (for a fully loaded CAC).

Definitions are in budgets, bidding and ROAS; keep them handy.

Step 1: Contribution per customer per month

JS Monthly contribution and LTV

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Notice how sensitive LTV is to churn. Lowering churn from 8 to 5 percent raises LTV by 60 percent, often a cheaper win than buying cheaper traffic.

Step 2: CAC by channel

Each channel has its own cost per sign-up. Include non-media costs allocated to the channel.

JS Channel CAC, LTV and payback

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Read the table: some channels pay back quickly, others take over a year. A channel with a poor ratio may still be worth it if it feeds the others (awareness that later converts via search); check this with attribution thinking rather than cutting it automatically.

Step 3: Scenario analysis

Ask "what if?" Build a function that recomputes the blended results under changed assumptions and compare scenarios.

JS What-if scenarios

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The last scenario shows a classic trap: a higher price may increase churn enough to cancel the gain. Test price changes carefully with experiments (A/B testing).

Step 4: Marginal returns and the budget decision

Average CAC hides that each extra dollar buys less. Model diminishing returns for a channel with a simple curve: customers = k x spend^0.7. Then compare the marginal CAC with LTV.

JS Marginal CAC with diminishing returns

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The rule "keep scaling while marginal CAC stays under the limit" is a practical budget decision tool. The curve and exponent are assumptions; with real data you would fit them from spend experiments.

Step 5: Write the recommendation

Turn the numbers into a one-page memo:

  1. Where we stand: blended CAC, LTV/CAC and payback.
  2. What is driving it: the best and worst channels and why (include caveats about attribution).
  3. Biggest levers: for example churn reduction (onboarding emails via automation), conversion rate gains on the sign-up page, price tests.
  4. Recommendation: budget moves, tests to run, and the metric that will tell you if you were right.
  5. Risks and assumptions: cohort data is young, churn may differ by channel, discounts distort LTV.

Present the numbers with ranges, not false precision, and update the model monthly using real cohort data.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Which change usually raises LTV the most for the same traffic cost?

    Lower churn lengthens the customer lifetime and raises LTV without extra acquisition cost.
  2. 2. A channel has CAC 180 and monthly contribution 18. What is its payback period?

    Payback = CAC / monthly contribution = 180 / 18 = 10.
  3. 3. Why look at marginal CAC when scaling?

    Diminishing returns mean average CAC understates the cost of additional customers.
  4. 4. Why should a model be updated with cohort data?

    Real cohorts replace assumptions and show channel differences.

Summary

  • Start from contribution per customer, not revenue.
  • Compute CAC, LTV/CAC and payback per channel, fully loaded.
  • Use scenarios to see which assumption matters most.
  • Compare marginal CAC to a limit before scaling.
  • Finish with a short, honest recommendation with its assumptions.

Next: Project: Plan, Run and Report an Experiment