Lifetime value

LTV estimates the total gross profit you expect from a customer over their life. The classic steady-state formula for a subscription with constant monthly churn:

Expected lifetime (months) = 1 / monthly_churn
LTV = (ARPA × gross margin) / monthly_churn

Same thing: monthly contribution × expected months. Always use margin-adjusted LTV when comparing to CAC. Revenue LTV inflates the number by 1 / GM.

The big idea

If 2% of customers leave each month, the average life is about 50 months. Multiply that life by the monthly leftover after COGS and you have LTV.

The analogy of a fixed lifespan breaks the moment expansion, mixed segments, or changing churn show up. Early startups often use 3–5 years of data to invent a 50-month life. Treat LTV as a scenario, not a fact. Prefer payback and cohort contribution curves when the company is young. That is Bill Gurley's warning — see LTV/CAC.

Walking through it

Classic SaaS sketch:

• ARPA = $1,000 / month
• Gross margin = 80% → monthly contribution = $800
• Logo churn = 2% / month

Lifetime ≈ 1 / 0.02 = 50 months
LTV = 800 / 0.02 = $40,000

Annualized: churn ≈ 2% × 12 = 24%/year → lifetime ≈ 1/0.24 ≈ 4.2 years; LTV ≈ 9,600 / 0.24 = $40,000. Same math, annual units.

If you expand revenue over time, use net retention in a cohort model rather than stuffing expansion into a single ARPA. Flat ARPA + gross churn understates good SaaS and overstates bad SaaS that only looks stable because big logos expand while small ones die.

Mixed segments break the blend. SMB: ARPA $400, GM 80%, churn 4%/mo → LTV = (400 × 0.8) / 0.04 = $8,000. Enterprise: ARPA $4,000, GM 80%, churn 0.8%/mo → LTV = (4,000 × 0.8) / 0.008 = $400,000. If CAC is $15k blended, SMB is underwater (0.53×) while enterprise is 27×. Scaling a blended "2.7×" motion that is actually mostly SMB kills the company.

Negative churn (expansion exceeds gross churn) makes 1/churn explode. Switch to cohort NPV of contribution. Do not feed −2% "churn" into the formula.

Finance-grade LTV discounts future contribution. A 10% annual discount on a 50-month life trims LTV by roughly mid-teens percent. Startup operating LTV usually skips this; if you include it, say so.

A 10% relative drop in monthly churn (2.0% → 1.8%) lifts LTV by about 11% in the simple model (1/0.018 vs 1/0.02). That is often worth more than a clever growth hack.

Common mistakes

• Revenue LTV compared to CAC. On 80% GM, $50k of revenue is $40k of gross profit.
• One LTV for mixed segments. The blend hides the underwater half.
• Feeding negative churn into 1/churn. The formula is not defined for that world.
• A 50-month life from 14 months of data. Scenario, not fact.

Check yourself

$1,000 ARPA, 80% GM, 2% monthly churn. LTV?

$40,000 of gross profit.

SMB LTV $8k and enterprise LTV $400k, blended CAC $15k. What do you do?

Stop scaling the SMB motion. The blend is a lie.

Key takeaways

• LTV ≈ (ARPA × GM) / churn. The classic case is $40k.
• Use margin-adjusted LTV against CAC.
• Segment. Blended LTV hides the half that does not work.
• When expansion exceeds churn, abandon 1/churn and use cohorts.

Further learning

The Dangerous Seduction of the Lifetime Value Formula (Bill Gurley)Many consumer Internet business executives are loyalists of the Lifetime Value model, often referred to as the LTV model or formula. Lifetime value is the net present value of the profit stream of a customer. This con…Above the Crowd

Next: CAC payback.