Quick definition: Customer lifetime value (CLV or LTV) is the estimated net economic value a defined customer or account generates over its relationship with a business.
What is customer lifetime value?
Customer lifetime value summarizes the value of a customer relationship over time. It is used to set acquisition budgets, evaluate customer segments, prioritize retention work, model pricing, and assess whether an experiment creates durable value. “Lifetime” is an estimate unless the relationship has ended; it is not a promise that every customer will remain forever.
CLV and LTV usually mean the same concept, although organizations should write their convention down. It is not revenue, annual recurring revenue, average order value, or customer acquisition cost. Revenue measures a period’s sales; AOV measures order size; CAC measures the cost of acquiring a customer. CLV estimates value across future and past relationship periods. A company may compare expected CLV with CAC, but only when the customer definition, costs, timing, and margin basis are consistent.
A credible CLV model identifies the unit—individual, household, account, or workspace—and the economic basis. Gross-revenue LTV is useful for topline planning but can exaggerate money available to fund acquisition. Contribution-margin CLV subtracts variable costs such as payment processing, fulfillment, support, commissions, and service delivery. A discounted cash-flow model additionally reflects that cash received later is worth less than cash received now.
CLV formulas and denominator choices
A simple retrospective calculation for a completed cohort is:
historical CLV = total net revenue or contribution margin from cohort / customers in cohort
A common steady-state subscription approximation is:
CLV ≈ average monthly gross margin per customer / monthly churn rate
This approximation assumes a stable churn process and is unreliable for new products, highly variable contracts, non-geometric retention, and customer groups with different behavior. A cohort model is often better: estimate each cohort’s retention and expected margin in each period, multiply them, and sum the results. A discounted version is CLV = Σ(expected period margin × probability retained in period / (1 + discount rate)^period).
Denominators change the story. CLV per acquired customer includes customers who never pay; CLV per payer excludes them; account CLV may combine many user seats. For acquisition decisions, use the same acquisition unit as CAC. Do not divide account revenue by active users and call it customer lifetime value.
CLV in A/B testing
CLV is the ideal business outcome for pricing, onboarding, loyalty, and retention changes, but it is usually slow to observe directly. An experiment can measure near-term mechanisms—activation, conversion, margin, early retention, refunds, and support burden—then use a preplanned model only if its assumptions are credible. A short-run revenue lift can reduce CLV if it creates poor-fit customers, greater discount dependence, or later churn.
When testing, calculate outcomes for all eligible units assigned to each variant, not only customers who purchase or renew. Conditioning on payment or engagement can hide an effect of the variant on who reaches that stage. Pair a leading primary metric with downstream CLV components as guardrails, and keep a holdout when the decision is high stakes. Primary and guardrail metrics provides the planning framework.
Do not announce “LTV lift” from a model recalibrated after seeing treatment data. Fix the cohort horizon, retention definition, margin components, discount rate, and any prediction model before the read. Show observed outcomes separately from forecasts, with intervals or sensitivity ranges. Revenue metrics are skewed and can be driven by a few accounts; our revenue-metric guide discusses analysis choices.
Worked CLV scenario
A subscription software product acquires a cohort of 1,000 accounts. Average monthly subscription revenue is $100. Variable costs are $35 per active account each month, so monthly contribution margin is $65. Historical monthly churn for comparable accounts is 5%.
simple CLV estimate = $65 / 0.05 = $1,300 per acquired account
This is an approximation, not a forecast guarantee. Suppose an onboarding variant costs $5 more per account to deliver but reduces comparable monthly churn from 5.0% to 4.5%, with margin otherwise unchanged. The simple values are $1,300 and approximately $1,444; after the extra $5 service cost, the modeled difference is about $139. The team must validate whether churn truly changes over a mature horizon and whether the added support cost persists. It should not multiply a noisy early estimate across the entire customer base without uncertainty analysis.
CLV data-quality caveats
CLV joins billing, refunds, usage, support, marketing, and customer-identity data. Duplicated accounts, missing upgrades, late refunds, currency conversion, tax treatment, and contract migrations can bias it materially. Reconcile revenue to finance records; use stable account keys; retain history when accounts merge; and version the model’s assumptions. If margin costs are allocated, document the allocation rule rather than implying they are directly observed.
Survivorship bias is common. Averaging only customers still active today overstates expected CLV, because customers who already left disappear. Cohort analysis includes the original population and explicitly treats loss. Forecasts can also be biased by acquisition-channel mix, seasonality, sales-led versus self-serve customers, and selective offers. Segment when business logic supports it, but avoid fishing across many segments for a favorable estimate.
Privacy and governance matter because CLV can influence targeting and service. Use only lawful, necessary data; limit access to sensitive attributes; and test models for unfair performance across groups. A high predicted value should not justify withholding essential support or deceptive pricing from other customers.
Common CLV mistakes
- Calling gross revenue “value”: margin and service costs can change the decision.
- Using a churn shortcut as a universal formula: its stability assumptions often fail.
- Mixing customer and account units: CLV and CAC then cannot be compared.
- Ignoring acquisition and retention cohorts: newer or different channels may behave differently.
- Rebuilding the model after the test: this turns modeling choices into outcome-driven flexibility.
- Treating a forecast as observed fact: report assumptions and sensitivity.
Frequently asked questions
Are CLV and LTV the same?
Usually yes. The organization should define its preferred term and the exact economic calculation behind it.
Should CLV use revenue or profit?
Use the basis relevant to the decision. Contribution-margin CLV is generally more useful for acquisition and investment decisions than gross revenue alone.
How does CLV relate to CAC?
CLV estimates relationship value; CAC estimates acquisition cost. Compare matched units and timing, and avoid simplistic ratios when payback timing or uncertainty matters.
Can CLV be negative?
Yes. Refunds, acquisition-related servicing, and variable costs can exceed revenue or margin for a cohort.
Can an A/B test measure CLV directly?
Only when follow-up is long enough. More often it measures observed components and uses a pre-specified, validated forecast cautiously.
Summary
CLV estimates the lifetime economic contribution of a defined customer or account. It requires explicit units, revenue or margin rules, retention assumptions, and time value of money where relevant. Use cohort-based evidence, distinguish observed data from forecasts, and evaluate experiments with all assigned units plus mature retention and cost guardrails.
Sources
- Federal Reserve: Customer lifetime value research
- Recurly: SaaS metrics guide
- U.S. SEC: Annual-report filing example