Calculating CLV: A Visual Walk-Through
In this sequence, we are charting a single customer as she spends business with our company over time. The plot below shows her cumulative money spent (revenue) with us at each time period (such as a quarter). In her first time period, Period 0, she spent $80. Then in Period 1, she spent another $80 to drive the cumulative sum up to $160. That process has continued at different periodic spending amounts through Period 4, where the total has reached $280.A customer’s value is not from gross revenue, but instead from the money that flows into the company net of variable costs. Below is what this customer’s curve looks like once we subtract variable costs.Next comes a prediction about what the customer will spend in the future. There are various ways to think about calculating this prediction, from simple linear extrapolation to Bayesian probabilistic models to machine learning predictions. In future posts in this series, we will go into some of the more popular methods.Since we are projecting future cash flows, the concept of "time value of money" tells us to apply some discount rate so we can express CLV in today’s dollars.CLV is defined as the dollar value where the curve flattens out. The effect of the discount rate pulls all such curves to a horizontal shape eventually.The CLV can be broken out into its components. Historical Customer Value is what we have already empirically observed. Residual Customer Value (RCV) is the amount that we expect to receive in the future (again, expressed in today’s dollars). For decision making about existing customers, we want to use RCV because it represents our expectations of the future. For future customers, yet to be acquired, the entire CLV is a prediction. In other words, CLV = RCV.When trying to predict the future, it is often helpful to calculate a 95% confidence interval to quantify the range of possible future scenarios. Below is what our confidence interval might look like.