How Customer Lifetime Value Calculator works

1. Scope

The CLV Calculator estimates the total gross profit a business can expect from an average customer relationship. It supports two modes: a subscription model (ARPU / churn) and a transactional model (purchase frequency × AOV × customer lifespan × margin). It does not perform cohort-level CLV from raw purchase logs, and it does not implement Pareto/NBD or BG/NBD probabilistic CLV models.

2. Inputs and outputs

Subscription inputs: ARPU, gross margin, monthly churn rate, optional discount rate for DCF-style NPV.

Transactional inputs: average order value, purchase frequency per year, retention rate, gross margin, optional customer lifespan override.

Outputs: CLV (gross profit form), CLV:CAC ratio when a CAC is supplied, implied average customer lifespan in months, and an NPV variant when the discount rate is enabled.

Engine source: src/lib/customer-lifetime-value-calculator/engine.ts.

3. Formula / scoring logic

# Subscription — memoryless churn, gross-profit form
CLV = (ARPU * gross_margin) / monthly_churn

# Transactional — classic Reichheld / Dwyer form
customer_lifespan_years = 1 / (1 - annual_retention_rate)
CLV = purchase_frequency * AOV * gross_margin * customer_lifespan_years

# NPV variant (when discount rate r is provided)
CLV_npv = sum_{t=1..T} (monthly_gross_profit / (1 + r)^t)

4. Assumptions

• Memoryless churn (subscription mode). The probability of churn in any given month is constant. Real retention curves often follow a declining hazard — the first 30 days see the highest drop-off.
• Gross margin is variable-cost margin. Include inference, hosting, payment-processing, fulfilment. Exclude fixed overhead.
• Transactional lifespan is 1 / (1 − retention rate). Equivalent to the geometric-series expected value of a memoryless retention process.
• No expansion revenue. Upsells, cross-sells, and price-tier migrations would need to be bundled into ARPU at period-weighted average.
• No seasonality. Q4-heavy retailers will under-estimate CLV for an average customer whose first purchase falls in Q1.

5. Data sources

Churn rate sources used as contextual reference (not baked into the formula):

• OpenView SaaS Benchmarks 2024 — gross and net churn by stage.
• Paddle SaaS Benchmarks 2024 — B2B vs B2C churn medians.
• SaaS Capital Annual Survey — retention norms for bootstrapped SaaS.

Probabilistic CLV model references (for readers who need cohort-level CLV):

• Schmittlein, Morrison, Colombo — Counting Your Customers: Who Are They and What Will They Do Next? Management Science, 1987 (Pareto/NBD).
• Fader, Hardie, Lee — "Counting Your Customers" the Easy Way: An Alternative to the Pareto/NBD Model, Marketing Science, 2005 (BG/NBD).

6. Known limitations

• Aggregate, not cohort-level. The formulas treat all customers as equivalent. For a business with heterogeneous cohorts (different channels, tiers, regions), aggregate CLV misleads — compute per-cohort CLV from timestamped data.
• Memoryless assumption is wrong at the edges. Retention curves are typically steep in the first 30 days and flatten afterward. The aggregate CLV under-states the sticky tail.
• No customer-heterogeneity modelling. Probabilistic CLV (Pareto/NBD, BG/NBD) accommodates heterogeneous purchase rates and drop-out — out of scope here.
• The Reichheld "5% retention lift = 25–95% profit boost" claim is context-dependent. It applies to businesses where retention is a major profit driver and cost-to-serve declines with tenure. We do not use it as a benchmark.

7. Reproducibility

8. Change log

• 2026-04-24methodology page first published. Subscription and transactional modes documented with limits.