A Framework for Call Center Decongestion Using Sequential Pattern Analysis

摘要

In an effort to improve its customer service, mobile telecommunication companies implemented various customer service channels like call center hotlines, text messaging, email or web self-service where subscribers conduct various after-sales transactions. Nowadays, a hotline call center via a customer service representative is the top choice alternative preferred by subscribers. However, it is noted that the cost of each call when transacting on a hotline is much greater than the cost of the other channels. Furthermore, subscribers get easily irritated when they need to wait for a long time to avail the services. In order to address the problem of reducing hotline calls, as well as to reduce the cost of customer service transactions, subscriber call transactions are analyzed in this paper to predict the next type of call that the subscriber will transact. A sequential pattern analysis methodology is applied and frequent sequences of calls are collected. Given the frequent sequences, the sequences of transaction calls are identified and a corresponding campaign is introduced to intercept the new calls and divert the transaction to less costly customer service channels.