Proportional Hazards Model for Customer Lifetime Value

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Estimating the hazard function for individual customers to calculate their customer lifetime value (CLV) is cumbersome and inefficient, as it requires determining a unique hazard function for each customer, which is challenging with existing methods.

Innovation Solution

A proportional hazards model is used, where each customer's hazard function is expressed as a product of a baseline hazard function and a coefficient value, allowing for efficient calculation of CLV by determining the baseline hazard function through historical data analysis and applying it to individual customer tenures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a unique hazard function is determined for each customer using existing methods, then the customer lifetime value calculation can be performed, but the process becomes cumbersome and inefficient

Engineering Contradiction:

Improvehazard function estimation accuracyVSAvoidcalculation efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The hazard function for each customer is segmented into two components: a baseline hazard function (common to all customers) and a customer-specific coefficient. This segmentation allows the complex individual hazard function to be broken down into a shared baseline part and individual adjustment factors, improving both estimation accuracy and computational efficiency.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The baseline hazard function serves as a universal component that applies to all customers, capturing common churn patterns across the entire customer base. This universal baseline can be estimated once from historical data and then reused for all individual customer CLV calculations, significantly improving productivity while maintaining precision through customer-specific coefficient adjustments.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If a unique hazard function is determined for each customer, then individualized CLV calculation is possible, but the complexity of the process increases

Engineering Contradiction:

Improveindividualized CLV calculationVSAvoidhazard function determination complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The hazard function determination process is segmented into estimating a common baseline hazard function from historical data and then applying customer-specific coefficients. This reduces the overall complexity by separating the complex historical analysis (done once) from the simpler individual customer applications.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of determining a completely unique hazard function for each customer from scratch, the method copies the baseline hazard function structure for all customers and then applies individualized coefficient adjustments. This copying approach maintains adaptability for individualized calculations while dramatically reducing the complexity of the determination process.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS20240020714A1Estimating a customer lifetime value based on a churn model

Publication Date: 2024.01.18 AFINITI AI LTD

AI summary

A method of determining a customer lifetime value (CLV) associated with a customer identifier (CID) is provided. The method comprises determining a hazard function associated with the CID based on a baseline hazard function and a coefficient value associated with CID; and calculating the CLV associated with the CID based on the determined hazard function. The determined hazard function is for calculating a probability that a customer associated with the CID will churn during a time interval.