How do I estimate LTV when I have only three months of data and no idea what my real churn or retention curve is yet?
The short answer
With three months of data you cannot compute a trustworthy LTV, so stop trying to project lifetime and instead track a shorter window you can actually observe, like 3 month or 6 month revenue per customer. Use a conservative retention assumption, model a range (pessimistic, base, optimistic), and revisit as real cohorts age. As a starting point, an honest 6 month revenue number beats a confident 3 year LTV built on a guessed churn rate.
Go deeper, your way
3 hand-picked resources, 2 link-checked. Pick how you want to dig in.
▶️ Video
✓ Link checkedFreeBeginner
Why we picked it
If you have never actually built a retention curve, reading about them only gets you so far, so this is a visual walkthrough of how a cohort table becomes a retention curve and how investors read it. It goes from raw user retention to net dollar retention step by step, which is the mental model you need before you try to bound LTV from three months of data. Watch it once, then go build your own tiny version in the template below.
Why we picked it
This is the piece that made cohort-based LTV the default way serious investors think, and it is exactly the right medicine when you have three months of data and no settled curve. Hsu argues you should read empirically realized cohort LTV instead of plugging numbers into a formula, and he shows why young cohorts should stay short lines (you genuinely do not know their LTV yet) rather than being extrapolated into a confident-looking figure. Treat it as a way to bound and shape your estimate, not to manufacture one.
Prefer empirically realized cohort LTV over a formula: with thin history, a single LTV number is a guess dressed up as math.
Young cohorts are short lines on purpose, do not extrapolate them; watch the shape (flat, sub-linear, linear, super-linear) instead of chasing one figure.
What you are really looking for early is evidence of linear or super-linear LTV in at least some cohorts, which tells you retention is holding, not just decaying.
Why we picked it
This is the template to stop guessing a single LTV number and plot your actual cohorts instead. You type in only the basics (how many customers you acquired each month and how many stayed each following month) and it works out retention, churn, MRR, and per-cohort lifetime value for you, so three months of data becomes three honest cohort rows rather than one made-up figure. It is a founder-friendly starting point, not a forecasting engine, which is exactly what you want this early.
