How would you model a 10% Day-1 retention improvement's impact on LTV?

WHAT THIS TESTS: This tests whether you model retention as a time-series survival curve or as a single average churn rate. Interviewers want to see if you understand that Day-1 retention is just one point on a cohort decay curve, and that translating an early-lift experiment into LTV requires explicit assumptions about how that lift propagates to Day 7, Day 30, and beyond. It also checks if you know the standard LTV components from unit economics: ARPU, gross margin, and the integral of expected retention over time.

A GOOD ANSWER COVERS: First, clarify whether the 10% improvement is relative or absolute percentage points, because moving Day-1 retention from 40% to 44% is very different from moving it to 50%. Second, define the cohort retention model, such as a power law or exponential decay curve fit to historical data. Third, state the propagation assumption: the most defensible approach is using historical elasticity, for example if a 10% relative lift at Day 1 has historically correlated with a 7% lift at Day 30, apply that decay factor rather than assuming uniform lift. Fourth, compute LTV by summing the product of projected survival probability, ARPU per period, and gross margin across the expected lifespan, or by integrating under the retention curve. Fifth, stress-test the result by running a sensitivity analysis on the curve shape, because if the improvement is only a first-day novelty effect, the LTV impact could be under 2% even though the Day-1 metric moved 10%.

COMMON WRONG ANSWERS: The biggest red flag is asserting that a 10% Day-1 retention gain automatically equals a 10% LTV gain. Another mistake is using the simplified formula LTV equals ARPU times margin divided by churn without acknowledging that churn is not constant in early life; this formula breaks down when retention is heavily front-loaded. Candidates also err by ignoring the distinction between correlation and causation, using historical LTV averages rather than building a forward projection from the modified retention curve.

LIKELY FOLLOW-UPS: How would you validate that a Day-1 retention experiment actually shifts the whole curve and not just the first day? If the improvement decays by half every week, what is the net present value impact at a 10% discount rate? How does this change your payback period or acquisition spend ceiling? What would the model look like if the product has a natural usage cycle, such as weekly or monthly, instead of daily?

ONE CONCRETE EXAMPLE: Suppose current Day-1 retention is 40%, Day-7 is 25%, and Day-30 is 15%. A 10% relative lift takes Day-1 to 44%. Historical data shows that 70% of the Day-1 lift typically survives to Day-7 and 50% to Day-30, so the new curve becomes roughly 44%, 27.3%, and 17.25%. If weekly ARPU is five dollars at 80% gross margin, the old LTV over 12 weeks sums to roughly 36 dollars while the new LTV sums to roughly 41 dollars, yielding a 14% LTV improvement. If instead the lift were isolated to Day-1 only, the LTV improvement drops to about 3%.

Source: statsig.com