Call Routing via Predictive Value Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional call center routing methods often lead to high business value callers being subjected to long waits, while low business value calls are prioritized, due to inefficiencies in identifying and routing inbound callers based on their value, leading to dissatisfaction and inefficient allocation of resources.
Innovation Solution
The implementation of a predictive machine learning model that uses customer identifiers such as name, address, and phone number to analyze and route inbound callers, employing n-gram modeling and approximate string matching techniques to prioritize calls based on predicted value, thereby optimizing resource allocation and reducing wait times for high business value callers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional first-come-first-served routing is used, then all callers are treated equally, but high business value callers are subjected to long waits while low business value calls are answered more promptly
Solution Approach 1:
The system changes the routing parameter from simple queue position to predicted business value score. By calculating a value score based on multiple factors (customer demographics, purchase history, call purpose, etc.), the system reprioritizes callers dynamically, allowing high-value callers to bypass long queues and be connected to agents more quickly.
Solution Approach 2:
The system performs preliminary analysis of caller information before routing occurs. By pre-calculating the business value score using machine learning models and available data (ANI, caller ID, CRM data), the system prepares routing decisions in advance, enabling faster connection of high-value callers without increasing overall system complexity.
2Productivity
If automatic call routing based on customer identification is implemented, then high business value callers can be identified and prioritized, but errors in customer identification may cause errors in retrieving data and routing inefficiencies
Solution Approach 1:
The system introduces an intermediary matching layer between customer identification and data retrieval. Instead of direct matching, the system uses fuzzy matching algorithms that compare multiple data points (name variations, address formats, phone number patterns) and calculate similarity scores. This intermediary process reduces identification errors by accounting for data quality issues and formatting variations.
Solution Approach 2:
The system prepares for potential identification errors by implementing fallback mechanisms. When exact matches fail, the system has pre-configured alternative matching strategies (partial matches, phonetic matching, pattern recognition) ready to engage. This cushioning approach ensures routing continues reliably even when initial identification attempts fail or produce false positives.
3Measurement precision
If multiple data sources are queried for customer information, then more accurate customer identification and valuation is achieved, but data processing time and system complexity increase
Solution Approach 1:
The system segments the data collection and processing into distinct hierarchical layers: first gathering readily available data (ANI, caller ID), then querying internal systems (CRM, call history), and finally accessing external data sources (demographic databases, credit bureaus) only when needed. This segmentation allows the system to achieve high accuracy while managing complexity by activating additional data sources progressively rather than simultaneously.
Solution Approach 2:
The system develops a universal data integration framework that handles multiple data sources through a common interface and standardized processing pipeline. By creating a multi-functional architecture that can ingest, validate, and process data from various sources (internal databases, external APIs, real-time inputs) using the same mechanisms, the system reduces overall complexity while maintaining comprehensive data collection capabilities.
Data Source
AI summary
A call management system of a call center identifies an inbound caller based upon computer analysis of customer identifiers, which may include at least two of customer name, street address, and zip code. Approximate string matching analysis matches n-grams generated from strings within customer identifiers, with n-grams generated from customer identification fields while searching one or more databases. Approximate string matching can incorporate a closeness metric based on Jaccard distance, and a Gaussian mixture model of best matches. In one embodiment, a polymr search engine analyzes customer identifiers of inbound callers to retrieve customer data, such as customer demographic data, matched to the customer identifiers. In another embodiment, the polymr search engine analyzes customer identifiers of inbound callers to identify repeat callers and retrieve previously collected customer data. Retrieved customer data is used in predictive modeling and scoring value of the inbound call, and in routing the scored inbound call.