Service Time Probability Distribution for Contact Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current contact center routing algorithms inaccurately estimate wait times due to their failure to account for individual agent skills and external factors, leading to inefficiencies and inaccuracies in resource allocation.
Innovation Solution
The development of service time probability distributions based on historical data for each agent and skill type, allowing for more accurate estimation of service times by using a histogram and probability distribution function, which can be updated dynamically to reflect changing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If average contact historical length is used to estimate service time, then the calculation is simple and easy to implement, but the accuracy of wait time estimation deteriorates
Solution Approach 1:
The patent changes the parameter from a single average value to a probability distribution function that captures the variability and shape of service times. This allows the system to account for different contact types, agent skills, and historical patterns, thereby improving estimation accuracy while maintaining computational feasibility through pre-computed distributions.
Solution Approach 2:
The system performs preliminary computation by pre-calculating and storing service time probability distributions based on historical data. These pre-computed distributions are then queried during real-time routing decisions, avoiding the need for complex real-time calculations while providing accurate estimates.
2Device complexity
If all agents are treated with the same average service time, then the routing algorithm is simple, but the reliability of resource allocation deteriorates
Solution Approach 1:
The patent applies local quality by creating agent-specific and skill-specific probability distributions rather than using a uniform average for all agents. Each agent's historical service time data is used to generate a customized distribution, allowing the routing algorithm to account for individual agent capabilities and contact types, thereby improving allocation reliability.
Solution Approach 2:
The system segments the homogeneous group of agents into distinct categories based on skill sets and historical performance. By dividing agents into segments with different service time characteristics, the routing algorithm can make more precise allocations while maintaining manageable complexity through structured classification.
3Reliability
If contacts are queued to maintain order, then customer requests are not lost, but customer wait time increases
Solution Approach 1:
The system performs preliminary estimation of service times using probability distributions before contacts enter the queue. This allows the routing algorithm to predict wait times and make informed decisions about contact assignment, enabling customers to be served more efficiently while maintaining queue integrity.
Solution Approach 2:
The system continuously refines service time estimates by incorporating actual service time data into the probability distributions. This feedback mechanism allows the system to learn from historical performance and improve its predictions, thereby reducing actual wait times while maintaining reliable contact handling.
Data Source
AI summary
Systems and methods are provided to estimate the amount of service time left in a contact with a contact center. A system records metadata factors associated with contacts and the service time for the contacts. This information is incorporated into one or more service time probability distributions. A service time probability distribution is a function describing how much time a contact, having certain, defined characteristics, took historically. The distribution can subsequently be used to estimate service time for future contacts that have the same or similar characteristics.