Estimate Customer Lifetime Value by Segment
Build a transparent CLV model with named assumptions, sensitivity analysis, and a one-line per-segment recommendation.
When to use this
When you need a defensible CLV number per segment — for pricing, marketing budget, or board decks — and don't want a single fragile point estimate.
The prompt
You are an analyst who shows the math and the assumptions, because point estimates are fragile.
Inputs:
- **Business model**: [subscription / transactional / freemium / hybrid]
- **What I'm using CLV for** (the decision): [...]
- **My segments** (by acquisition channel, plan tier, geography, or cohort): [...]
- **Per-segment data I have** (any subset is fine): average revenue per customer per period, gross margin %, retention rate or churn, period length, observed customer-lifetime so far: [...]
- **Discount rate** I want to apply: [or "use industry standard for SaaS / consumer / B2B"]
Build a CLV model:
1. **The formula** I'm using and why — show it. State whether you're using simple, residual lifetime, or probabilistic.
2. **Per-segment computation** — for each segment, show:
- Inputs (with units)
- Computed CLV (one number)
- Confidence (high/medium/low) given data quality
3. **Sensitivity** — for each segment, which input swings the answer most? If retention is the lever, say so.
4. **The 6-month vs. 3-year view** — many CLVs look great over 3 years and break in the first 6 months. Show both.
5. **What this CLV is NOT** — name the things it doesn't capture: viral acquisition, support cost, payment failures, segment migration.
6. **One-line per-segment takeaway** — what should change in how we treat each segment, based on the math?
If I'm missing data for a number, ask for it. Don't invent.
What you'll get back
A stated formula, per-segment CLV computations with confidence, sensitivity callouts, dual time horizons, named limitations, and a one-line strategic implication per segment.
How this is structured in English
Notice the English patterns this prompt uses — they're worth borrowing for your own requests.
- Point estimates are fragile Statistical principle in three words. A single number hides the uncertainty; the spread around it tells the real story. Worth remembering.