Understanding how much money your customers spend is the core of marketing strategy: you can’t acquire users if you don’t know how much they’re worth. Lifetime Customer Value (LCV, although I’ve seen it represented as CLV and LTV) as a dashboard-level metric provides critical insight into how a service monetizes.
I speak of LCV in the context of the freemium model: a free product that produces revenue through add-on features. LCV is much simpler to calculate for fixed-price products that only produce revenue at one point (ignoring virality), and a little simpler to calculate for subscription products (where the crux of LCV is customer lifetime). Calculating LCV for freemium products requires insight into both customer lifetime and spending patterns.
The first component of LCV is what I call duration: I pirated this term from finance (although it’s an imperfect fit), and I use it to describe the number of days in which a user will engage with a service as opposed to the number of calendar days over which they’ll remain in the service. As an example, if a user downloaded my game from the app store on January 1st, played again on January 10th, and then deleted my app on January 30th, I’d calculate her duration as 2 days, not 30.
I like thinking about customer lifetime from the perspective of duration because a user then resembles a perpetuity bond — a financial instrument that distributes payments indefinitely. While obviously there are differences between a fixed-payment set of cash flows and the variable-payment cash flows of user purchases, at a high level, these two things are functionally similar. Conceptually, this framework forces budgetary discipline on the user acquisition process: if we have a concrete precedent for quantifying the value of a user, then the marketing budget will be driven by that (and not the other way around).
To calculate duration, you’ll need to first build a retention profile of your users. If you have the key retention benchmark data (1-, 7-, and 30-day), you can extrapolate your users’ retention curve out to some point — I usually forecast the curve to 365 days. Each point on the curve represents the likelihood that a user will interact with your app on that day; summing the values will provide you with duration.
The second component of LCV is customer value — that is, how much money users spend. This is fairly simple to calculate; I generally take a trailing average mean after some sort of classification algorithm has removed outliers. Since massive spenders are (hopefully) part of the userbase spending distribution, I don’t remove them without first validating that their spending patterns within the context of a high-spender are improbable or irregular. It’s important to measure average spend on a daily, per-user basis not in calendar days but in engagement days — otherwise it is incompatible with duration.
I use a trailing average because changes to the app’s functionality (content pushes, bug fixes, introducing new features, etc.), changes to the app ecosystem (competitor apps), and changes to userbase dynamics can cause spending patterns to shift. A historical average would not provide appropriate weight to these changes; LCV should tell you the expected income from a user acquired today. Historical spending averages are irrelevant information; ideally you’re constantly testing and improving upon monetization funnels and content, and you want your LCV metric to reflect those improvements.
LCV is not a magic elixir
Understanding how your users monetize and retain is important, and getting a dashboard-level LCV metric as accurate as possible can facilitate many decisions. But LCV is, first and foremost, a marketing stricture: it sets a limit on per-user marketing spend. It can also be controversial in organizations where product management and marketing do not interface regularly, or where product management has historically held carte blanche in budgeting marketing campaigns. LCV is not a product management measurement; it’s not a signal that a product is “good” or “bad”. It simply prescribes a target for acquisition budgeting, and is especially useful when aggregated by user characteristics such as acquisition channel, geography, and demographic group (e.g. male users aged 18-25 from the US acquired via Google AdWords have an LCV of X).