LTV (lifetime value)

How it is calculated

In practice nobody waits for a user's whole lifetime. You take a group of users who installed in the same period, add up the revenue they have produced so far and divide by the number of users in the group. The result is written with its age: day-7 LTV, day-30 LTV, day-180 LTV. Revenue can include purchases, subscription payments and ad revenue, and you should say which ones you counted.

The group is a cohort, and the calculation only makes sense per cohort. Mixing users of different ages produces a figure that moves whenever install volume changes.

Predicted LTV

Waiting six months to learn whether a campaign paid off is too slow for most teams. Predicted LTV, often written pLTV, uses a cohort's first days of behaviour to estimate its later revenue. Treat it as a forecast: useful for early decisions, and worth checking against actual revenue once that arrives. Airbridge lists predictive LTV for a period of up to 180 days.^[1]

When it matters

LTV matters most when revenue arrives slowly. Subscription apps earn through renewals, and content apps earn through repeated sessions, so day-one revenue tells them little.

How it relates to attribution

Attribution is what lets you split LTV by source. Without it you have one LTV for the whole app. With it you have an LTV for each channel, campaign and creative, which you can set against what each one cost. That comparison is ROAS when expressed as a ratio.