Customer Lifetime Value
Date: 2026-08-16
What a customer is worth over their relationship with you. Several definitions circulate and they differ by a factor of three — so the first question about any LTV figure is which one it is, and the second is whether it’s contribution or revenue.
What it is
Customer lifetime value (LTV or CLV) is the total value a customer generates across their lifetime.
The definitions that circulate, on the running model — £50 average order, £15 contribution, 1.6 orders per customer over 24 months:
revenue LTV 1.6 × £50 = £80.00
gross margin LTV 1.6 × £22.50 = £36.00
contribution LTV 1.6 × £15 = £24.00 ← the one comparable to CAC
Three-and-a-third times between the top and the bottom. A business quoting £80 LTV against a £22.50 CAC looks excellent; the same business quoting £24 looks marginal. Both figures are “LTV”.
Only contribution LTV is comparable to CAC, because CAC is money actually spent and revenue isn’t money kept — Contribution Margin, LTV to CAC Ratio.
The simple formula
For subscriptions, the equivalent expressed through churn:
At £15 monthly contribution and 5% monthly churn: £15 ÷ 0.05 = £300. Note how violently that moves — churn at 4% gives £375, at 6% gives £250. LTV is extremely sensitive to the churn estimate, which is why subscription LTV figures should always come with the churn assumption attached.
The horizon problem
“Lifetime” is unbounded, which makes the number unfalsifiable. A customer acquired today might buy for a decade — you cannot know, and modelling it forward compounds every assumption.
Use a bounded horizon and say so. 12-month or 24-month value is measurable, checkable, and comparable across cohorts:
LTV(12m) £18 ← observed, verifiable
LTV(24m) £24 ← observed for older cohorts
LTV(∞) £41 ← modelled, and load-bearing assumptions
A 24-month figure derived from cohorts that are actually 24 months old is a fact. An infinite-horizon figure is a forecast, and treating it as a fact is how acquisition budgets get set too high.
Where it’s fragile
- It’s an average over a heavily skewed distribution. A small share of customers generate most of the value, so the mean LTV describes almost nobody. Report the median alongside, and segment — Skewed and Heavy-Tailed Distributions
- It requires durable identity. Anonymous traffic has no LTV. Which means LTV is computed on identified customers only — a self-selected, better-performing group — Anonymous and Identified Users, Identity Stitching
- It varies enormously by acquisition channel. Discount-led acquisition brings customers with lower repeat rates. A blended LTV justifying a channel’s CAC can be wrong in both directions — Channel Mix
- Survivorship Bias. Computing LTV on customers who are still active excludes those who left, inflating it
- Returns and refunds. Deducted from LTV, or it’s overstated by the return rate — Return Rate and Reverse Logistics
Using it well
- Contribution-based, bounded horizon, segmented by acquisition channel and cohort. All four qualifications matter
- Compute it from cohorts, not from a blended average. Cohort LTV curves show whether it’s improving; a blended figure moves with acquisition mix
- Compare against Payback Period as well as CAC. A high LTV realised over three years doesn’t help if you can’t fund the gap
- Recompute quarterly. Contribution margin moves, repeat rates move, and an LTV figure nobody has revisited is a fiction the business is spending against
In plain terms: LTV is the most assumption-laden number in commerce and it’s used to justify the largest spending decisions. Ask what horizon, what margin basis, and which customers — before you use anyone’s figure, including your own.