What is Cohort LTV?

How it differs from a single LTV figure

A single "customer lifetime value" figure is usually a forecast: average order value × orders per year × years retained. It mixes customers acquired years ago with last month's, and it assumes a lifetime that may never happen.

Cohort LTV is measured, not forecast. It follows one group of customers, all acquired in the same month, and records what they have actually earned at fixed ages: three months, six months, twelve months. Each cohort is compared with others at the same age, so you can see whether newer customers are better or worse than older ones.

How it is worked out

Cohort LTV at month m = first-order contribution + repeat-order contribution earned up to month m, per customer in the cohort.

• Use contribution (CM2), not revenue, so the figure can be set against acquisition cost.
• Divide by all customers in the cohort, including those who never ordered again. Dividing by repeat buyers alone overstates it.
• Count repeat orders after cancellations, RTOs and returns.

What it tells you

• LTV to CAC. Cohort LTV at twelve months divided by blended CAC shows how many times over the cohort has repaid its acquisition cost.
• Payback. The month cohort LTV crosses blended CAC is the CAC payback period.
• Trend. If each new cohort's six-month LTV is lower than the last, the brand is acquiring weaker customers, often because a discount, a new channel or a broader audience brought in one-time buyers.

What goes wrong

• Projecting from the first few months. Repeat behaviour in a product's first season may not last. Show measured points and label any projection as one.
• Mixing channels. Marketplace customers usually cannot be identified for repeat orders the way website customers can. Measure cohort LTV where the brand owns the customer data, mainly its own website and app.
• Ignoring the cost of repeat orders. Retention discounts, loyalty points and remarketing reduce the contribution of a repeat order. Deduct them.