Churn Risk Scoring via In-Memory Call Network Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Telecommunication providers face challenges in identifying and retaining customers at risk of churn, as existing methods lack efficient tools for predicting customer attrition and targeting retention initiatives effectively.

Innovation Solution

A computer-implemented system and method using a multi-variable churn risk model to analyze customer and account characteristics, generating churn risk scores, and presenting them visually through an interactive 3D interface, incorporating social network analysis and data visualization to identify high-risk customers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional customer attrition analysis methods are used, then customer churn risk can be identified, but the identification process is inefficient and lacks real-time capability

Engineering Contradiction:

Improvecustomer churn identification efficiencyVSAvoidtime delay in churn risk identification

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent replaces traditional mechanical data processing systems with an in-memory computing platform that uses advanced algorithms and social network analysis to process customer data in real-time, dramatically improving identification efficiency and eliminating time delays

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs preliminary analysis of customer behavior patterns, call data records, and account characteristics continuously in the background, so that when churn risk assessment is needed, the information is already prepared and immediately available, eliminating waiting time

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If comprehensive customer data analysis is performed to accurately predict churn, then prediction accuracy improves, but system complexity increases

Engineering Contradiction:

Improvechurn prediction accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the complex analysis system into distinct functional modules: social network analysis module, behavior pattern recognition module, predictive modeling module, and visualization module. Each module handles specific aspects of data processing, making the overall complex system manageable and maintainable while preserving high prediction accuracy

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary in-memory platform that acts as a mediator between raw data sources and analysis algorithms. This platform pre-processes and structures data from multiple sources (call records, account data, behavior logs), simplifying the input for predictive models and reducing the complexity burden on individual analysis components

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If detailed customer behavior analysis is conducted to identify high-risk customers, then retention targeting accuracy improves, but data processing requirements increase

Engineering Contradiction:

Improveretention targeting accuracyVSAvoiddata processing resource consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system continuously performs preliminary analysis of customer behavior patterns, call data, and account characteristics in the background, maintaining pre-computed metrics and risk indicators in memory. This allows detailed analysis to be performed on-demand without intensive real-time processing, reducing energy consumption while maintaining high targeting accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies different levels of analysis depth to different customer segments based on their risk profiles. High-risk customers receive comprehensive multi-dimensional analysis, while low-risk customers receive simplified assessment, optimizing resource allocation and reducing overall data processing requirements while maintaining accurate retention targeting

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS10306013B2Churn risk scoring using call network analysis

Publication Date: 2019.05.28 SAP SE

AI summary

Customer churn risk scores are based on a multi-variable churn risk model relating customer and customer account characteristics to a risk of customer churn. A computer-implemented method of generating and presenting churn risk scores of customers of a telecommunication provider involves analyzing, on an in-memory database platform, customer call data records and customer records to calculate a churn likelihood value, an influence factor value, and an average spend value for each customer. The method assigns a churn risk score to each customer according to the model using the calculated churn likelihood value, the calculated influence factor value, and the calculated average spend value as input to the model. The churn risk scores for one or more customers are displayed visually on an interactive computer-user interface (UI).