Customer Lifetime Value Explained

Customer lifetime value (LTV or CLV) is an estimate of how much a customer is worth to the business over a defined period. In e-commerce, the useful version is usually contribution LTV: cash you keep after variable costs, not GMV.

A simple model:

LTV ≈ AOV × purchase frequency × gross margin × lifespan

Better:

LTV ≈ (contribution per order) × (orders per customer in the horizon)

Use the customer lifetime value calculator for the arithmetic. The quality of LTV is entirely in the inputs. Treat the output as a range, not a budget you can spend in full on ads.

Compare it to CAC with the CAC calculator. If you compare LTV-revenue to CAC, you will overbid.

Simple average vs cohorts

Storewide average (“customers spend $X per year”) mixes:

• One-time bargain hunters and loyal buyers.
• Old cohorts acquired on cheap brand terms and new cohorts from cold ads.
• Categories with different return rates.

Cohorts (customers whose first order was in the same month) show whether repeat actually happens for this acquisition mix. Plot orders and contribution at 30/90/180/365 days. If 90-day contribution never covers paid CAC, a pretty 12-month LTV built from historical whales is not a license to scale.

Approach	What it answers	Main lie
Historical average LTV	What past customers did, blended	Mix has changed
Cohort LTV	What a class of new customers did over time	Small samples; incomplete horizons
Predictive LTV	What a model thinks will happen	Overfit; untested on new channels

Young stores should prefer observed 90-day contribution over a 3-year prediction. You can still model a longer tail; do not bid as if it were cash in the bank.

Contribution, not revenue

Average order value is an input, not LTV. Subtract:

• Product and inbound landed cost.
• Payment fees.
• Outbound shipping you do not recover.
• Expected returns / RTO (especially with COD).
• Variable marketplace fees if that is the channel.

That is the same contribution idea as profit margin. LTV built on list price will disagree with the P&L.

Refunds after the first order belong in the frequency and contribution terms, not as a surprise at year end.

Repeat rate and lifespan

Purchase frequency is not a personality trait. It is stock availability, replenishment cycle, and whether the first order was a good product.

• Consumables can support a higher honest frequency.
• Durable goods often have a long gap; LTV is mostly first order plus accessories.
• Subscriptions need churn and failed-payment (dunning) in the model, not a copied D2C frequency.

Lifespan in the simple formula is often “we assume 2 years.” If you have not observed 2 years, say “LTV-12m” instead. Precision to the cent on a made-up lifespan is worse than a coarse 12-month number.

Do not overclaim

LTV is used to justify:

• Higher CAC.
• Loyalty programs.
• Raising AOV with bundles.

All of those can be rational. None are justified by a single slide that says “LTV is $480.” Ask:

• Which cohort?
• Contribution or revenue?
• What horizon?
• What happens if repeat drops 20%?

ROAS on the first order can look poor while LTV looks fine—or the reverse if repeat was a one-time restock. Put both on the metrics that matter dashboard.

A practical operating model

1. Compute first-order contribution.
2. Compute 90-day contribution LTV by acquisition channel if you can.
3. Set CAC payback rules (e.g. first-order contribution covers X% of CAC; 90-day covers the rest) that match cash.
4. Recompute when mix, returns, or AOV shift—not once a year.

Payback, not a single ratio

A common operating rule is CAC payback: how many days of contribution until acquisition spend is recovered. That uses the same contribution LTV building blocks without pretending you know year five.

Example shape (illustrative): paid CAC $28, first-order contribution $18, 90-day extra contribution $16. You recovered CAC inside 90 days in that cohort. If the next cohort’s 90-day extra is $4, the same CAC is a different decision. Storewide LTV would have hidden that.

Do not fund payback with unpaid invoices or COD that never remits. Cash timing is part of whether LTV is spendable.

Discount-driven repeat is not loyalty

If repeat orders only happen when you send 25% off, the simple frequency input is a coupon habit. Contribution LTV should use discounted tickets and a lower margin, or you will bid as if full-price repeat were the base case.

Subscriptions have explicit churn. One-off catalogs have implicit churn (they never come back). Both belong in the model as a rate, not as “customers love us.”

New vs returning mix in the acquisition channel also changes LTV. A cohort acquired on a steep first-order discount is not the same as one acquired on a replenishment SKU. Split LTV by first-SKU category if the catalog is wide.

Marketplace customers you do not own (no email) should not be blended into D2C LTV without a footnote. You cannot email them a replenishment the same way.

Returns after day 30 still belong in contribution LTV if they are common in your category. Cutting the horizon to look healthier is how the model lies.

The e-commerce analytics guide is the reporting home. LTV is a planning tool. It is not a guarantee of future orders, and it is not a reason to ignore conversion and product quality on order one.