What is cohort economics?

Cohort economics groups customers by the calendar month of their first order and follows each group's behaviour separately: how many they are, how many came back, and how much contribution margin each customer has generated. Lifetime averages blur everyone together; cohorts answer the question those averages can't — are the customers we acquire now better or worse than the ones we acquired a year ago?

Formula

cohort            = customers whose first order fell in month M
repeat rate       = cohort customers who ordered again ÷ cohort size
LTV per customer  = cohort's total contribution margin ÷ cohort size

Worked example

The January cohort: 220 customers, repeat rate 34%, LTV €96 per customer. The June cohort: 410 customers, repeat rate 19%, LTV €41. Alarming — until you note June is only three months old and January has had eight months to mature. Compare June against what January looked like at three months, and the honest questions appear: is the bigger June cohort (cheaper acquisition?) tracking to the same maturity curve, or did scaling spend buy structurally worse customers?

How Saldo Metrics computes it

v_cohort_economics classifies every order as first (order date equals the customer's first order date) or repeat, groups by the first order's calendar month, and aggregates: cohort size counts distinct first-order customers, repeat customers are those with any later order, and the margin figures apply the order-grain cost rule — an order with any uncosted line contributes unknown margin, not zero. cohort_ltv_margin_base is the cohort's total realized contribution margin (first and repeat orders both); ltv_per_customer_base divides it by cohort size. All figures are lifetime-to-date per cohort and grow as cohorts age.

Why it matters

Cohorts are where acquisition quality becomes visible. Average LTV mixes five-year-old customers with last month's; the cohort table separates them, which is what lets you see whether a channel shift, a pricing change or a scaling push changed the kind of customer arriving — usually a year before the blended averages move.

Common mistakes

• Comparing cohorts of different ages at face value. Every young cohort loses to every mature one; compare at matched age or against the older cohort's history.
• Judging acquisition on first-order economics. A cohort can be loss-making on first orders and excellent at twelve months; the payback story needs the repeat columns.
• Ignoring cohort size. A 30-customer cohort's repeat rate swings wildly; read small cohorts as anecdotes, not trends.

Where you see this in the app

The Cohort Economics dashboard widget, one row per first-order month.