Customer Churn Prediction Using Service and Device Usage Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting customer churn in servicing businesses are limited by insufficient data insights, making it difficult and costly to retain customers, as they only reveal obvious causes of dissatisfaction.
Innovation Solution
Employing big data and artificial intelligence, including machine learning, to systematically collect and analyze various data sources from MFPs to identify correlations between events and attrition, generating alerts and suggesting remedial actions when satisfaction levels drop below a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If customer feedback and surveys are used to collect data, then customer satisfaction insights can be gathered, but the data is limited and only reveals obvious causes of dissatisfaction
Solution Approach 1:
The patent transitions from traditional one-dimensional survey feedback to multi-dimensional data collection by integrating device usage metrics, service interaction patterns, and operational data. This dimensional expansion enables the system to detect subtle churn indicators that surveys alone cannot capture, such as changing usage patterns or service response times.
Solution Approach 2:
The patent introduces an intermediary data processing layer that aggregates and analyzes diverse data sources (device logs, service calls, customer interactions) before presenting insights. This intermediary layer transforms raw, disparate data into meaningful patterns, revealing hidden causes of dissatisfaction that would otherwise remain obscured in the complexity of multi-source data.
2Productivity
If traditional customer feedback methods are used, then obvious causes of dissatisfaction are revealed, but it is expensive and difficult to acquire new customers compared to retaining existing ones
Solution Approach 1:
The patent implements preliminary action by continuously monitoring device usage and service interactions in real-time, identifying at-risk customers before they actually churn. The system proactively detects deteriorating satisfaction patterns and triggers retention interventions early, rather than reacting after customers have already decided to leave. This timing advantage significantly improves retention efficiency.
Solution Approach 2:
The patent establishes a closed-loop feedback system where customer behavior data continuously informs retention strategies. The system analyzes real-time data, identifies churn risks, and automatically triggers appropriate retention actions, then measures the impact of those actions. This iterative feedback loop enables continuous optimization of retention efforts, making the process increasingly efficient over time.
Data Source
AI summary
A system and method for minimizing customer churn in device service businesses commences with execution of a customer service contract. Ongoing customer data capture is made for each contract. Customer data includes contract events, environmental events, service events, device usage analytics and personnel events. Machine learning is applied to captured customer data, which machine learning is based on a state of customer data at the time of contract determination. Customer data is assigned weights, and aggregate data for each customer is compared to a preselected threshold level. Customers above a threshold are deemed happy and customers below the threshold are deemed to be at risk. Remedial measures relative to at risk customer data generates levels of automated remediation followed by remedial measure suggestions to an administrator when not sufficiently successful.