How it works
Take a cohort of users who installed in the same period, add up all the revenue they have produced by day N and divide by the number of users in the cohort. The result is the day-N LTV, sometimes called cumulative ARPU. Split by campaign, country or platform, it shows which sources bring valuable users. Longer horizons are usually predicted: the shape of the early curve, together with retention, is used to estimate what the cohort will be worth later.
Why it matters for an app team
LTV sets the ceiling on what you can pay for a user. If the LTV you can count on is above the cost per install, with room for other costs, the channel can scale. If not, it loses money on every install however good the volume looks.
In practice
Example with made-up numbers: 4,000 users install in March. By day 30 they have generated $2,800, so D30 LTV is $0.70. By day 90 the total is $4,400, a D90 LTV of $1.10. With a cost per install of $0.90, the cohort pays back between day 30 and day 90.
Common confusion
LTV is an average, and a few heavy spenders can carry it, so look at the distribution as well. It is not the same as ARPDAU, which is revenue per active user per day. State whether a figure is measured or predicted, and whether revenue is before or after store fees. A predicted LTV is a forecast and should be checked against what cohorts later deliver.