Customer Lifetime Value Prediction with the BG/NBD Model

What is the BG/NBD Model?

The BG/NBD (Beta-Geometric/Nonhomogeneous Poisson Process) model is a statistical tool used in marketing to predict customer behavior, specifically focusing on purchase frequency and recency. It helps businesses understand how likely it is that a customer will make future purchases and the total value of these transactions over their lifetime.

Developed by Peter Fader and his colleagues at Wharton, this model combines two components: the Beta Geometric distribution for modeling the probability of a customer's first repeat purchase (dropout rate) and the Nonhomogeneous Poisson Process to model the frequency of purchases.

How Does the BG/NBD Model Work?

The BG/NBD model uses historical transaction data, such as the time since a customer's last purchase (recency) and their total number of transactions (frequency), to estimate parameters that describe the underlying distribution of these behaviors. These parameters are then used to predict future purchases and calculate the expected lifetime value of each customer.

The key parameters include the average frequency of purchases, the average recency of repeat purchases, and the probability of a customer becoming inactive after their last purchase. By updating these estimates as new data becomes available, businesses can refine their predictions in real-time.

Why Does It Matter?

Understanding customer lifetime value (CLV) is crucial for optimizing marketing strategies and allocating resources efficiently. The BG/NBD model helps companies identify high-value customers who are more likely to continue purchasing, allowing them to tailor their marketing efforts and retention strategies accordingly.

By accurately predicting future purchases, businesses can also better plan inventory, forecast revenue, and make data-driven decisions about customer acquisition and loyalty programs.

Real-World Applications

The BG/NBD model is widely used in e-commerce, retail, and subscription-based services. For instance, an online retailer might use it to identify which customers are most likely to make a repeat purchase and then offer personalized promotions or discounts to increase their CLV.

In the context of streaming services, the model can help predict how long subscribers will remain active based on their viewing habits, enabling targeted retention campaigns.

Frequently asked questions

Try it live

Everything above runs in your browser — open Customer Lifetime Value Predictor — BG/NBD Live and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.