Skip to content

LTV — Customer Lifetime Value

ARPU divided by churn rate. Two inputs, and an unusually large amount of room to disagree about both.

The idea

If customers pay ARPU per month and a fraction \(c\) of them leave each month, the average customer stays \(1/c\) months and is worth \(\text{ARPU}/c\). That is the whole model.

Its fragility follows directly: LTV is a ratio of two estimates, one in the denominator. At low churn rates small absolute changes swing LTV enormously — 1% monthly churn implies a 100-month lifetime, 2% implies 50. The metric is least stable exactly where the business is healthiest.

Cohort LTV sidesteps the model entirely by measuring actual cumulative revenue per customer within a cohort. Slower to mature, but it is an observation rather than a projection.

Why it matters

LTV is only meaningful against CAC. The LTV:CAC ratio — with > 3x the conventional healthy benchmark — is the standard test of whether growth spend is sustainable.

Tidemill computes LTV today but not CAC, which remains P1 and unimplemented. CAC requires acquisition spend, which is why expense analytics via QuickBooks Online is a prerequisite rather than a side quest.

How Tidemill computes it

LTV is ARPU over logo churn rate, where ARPU is MRR over active customer count. Cohort LTV sums all paid invoices per customer in base currency, divided by cohort size.

Authoritative formulas: definitions.md — LTV. Implementation: tidemill/metrics/ltv/.

Known documentation drift

definitions.md files LTV under "Planned (P1) — designed but not yet implemented", but tidemill/metrics/ltv/ ships and is registered. The formulas there are current; the status label is stale. Recorded in the wiki log on 2026-07-23.

Contested ground

  • Churn denominator window. Tidemill uses the current-period churn rate. ChartMogul uses a 6-month trailing average to damp the volatility described above. ChartMogul's is arguably the better estimator; Tidemill's is the more legible one, since "which six months?" has no visible answer. A configurable lookback is a candidate improvement.
  • Revenue vs. margin. LTV computed on revenue ignores cost of service. For usage-heavy products with real COGS this overstates value materially.
  • Logo vs. revenue churn as denominator. Using logo churn implicitly assumes a departing customer was average-sized.

Sources