Test 03
The CLTV Reality Test
How much of your customer lifetime value is measured, and how much is an assumption compounded forward?
What everyone does instead
CLTV is almost universally computed as ARPA multiplied by gross margin divided by churn rate. That formula assumes churn is constant for the life of the customer, which it never is, hazard declines with tenure. The resulting figure is then used to justify acquisition spend and to compare acquisition against retention at the margin.
What to compute
- The textbook figure, as currently reported.
- Measured realised gross margin per customer over 24 months, from billing, with no extrapolation.
- A defensible lifetime figure using an observed declining hazard rather than a constant rate.
- The CLTV-to-CAC ratio under each, and how the marginal allocation decision changes between them.
The arithmetic
CAC $38,230 Churn rate used in the textbook formula 12% ARPA x gross margin, implied $66,300 Textbook $66,300 / 0.12 = $552,500 14.5x CAC Measured 24-month realised GM = $127,920 3.3x CAC Declining hazard estimate = $420,522 11.0x CAC Implied expected life $420,522 / $66,300 6.34 years $552,500 / $420,522 1.31 textbook overstates by 31%
Worked example
Against a CAC of $38,230: the textbook CLTV is $552,500 (14.5x), the measured 24-month figure is $127,920 (3.3x), and a declining-hazard estimate is $420,522 (11.0x). An inflated CLTV makes an acquisition dollar look better than a retention dollar.
Illustrative figures. The relationship shown is the hypothesis the test exists to check.
What changes if this is true
An inflated lifetime value makes the next acquisition dollar look better than the next retention dollar, every time the comparison is made. Correct the figure and the marginal comparison can reverse. Nothing about the business changed. The number you were deciding with did.
What you need and how long
Contract start, ACV, renewals, expansion, churn dates
- Where it lives
- Billing
- Days
- 1
Sales and marketing spend, new customer count
- Where it lives
- Finance
- Days
- 0.5
Hazard estimation and lifetime construction
- Where it lives
- Analysis
- Days
- 1.5
Every test runs on the same two extracts, forecast history from the CRM and contract history from billing, pulled by a RevOps analyst in about an hour. Everything is measured against your own outcomes: no benchmark, no composite, no assumed uplift.