How BODEX improved customer satisfaction of a Call Center by 28% in a year!
Problem Statement
A Call Center typically has hundreds of thousands of call recordings that contain significant untapped intel that can potentially be used to improve efficiency and customer satisfaction. To be able to extract actionable intel, patterns from the stored assets (call recordings), was a huge task and a massive undertaking if done manually. In the absence of such a quality control process, Call Centers are working without appropriate feedback and are ignoring efficiency optimization opportunities.
Solution
We were approached by a software company that builds Call Center automation software; they described the situation where there were thousands of Call Center recordings that needed to be audited. We understood the problem and proactively suggested additional capabilities that we could build into the software in addition to just the audit functionality. We took a step back and looked at the problem from a holistic perspective.
We created a cloud-based software application that processes the incoming call recordings, automatically extracts the transcripts, and provides a browser-based interface for auditing. We came up with a solution for this problem by breaking it up into two different major sections:
Part one involves managing assets and includes ingestion of assets (call audio recordings) and
Part two involves building a queuing mechanism that allows supervisors to find, listen to and rate recordings, and an automated way to extract the transcripts from given audio files.
Technically, it has 3 main parts:
Ingestion Engine and Assets management:
All day-to-day agents’ recorded calls (we call them CCR-Customer Call Recordings) along with metadata that was stored in MongoDB (for storing recordings of a specific agent) and MYSQL (for storing metadata of a particular record by a specific agent) via Nodejs scripts which is an automated process (we call it Ingestion Engine) which works on CRON scheduler.
Admin Panel:
It is comprised of Result sets UI (contains all metadata fetched from MYSQL) and an embedded graphical audio player having great functionalities like search, filter (by agent name, date-wise, etc.), voice splitter through which admin can hear individual voices of agent and customer separately, 1 sec skip while hearing recordings, rating the particular agent’s activeness based on recording, etc.
Creating transcript from metadata files:
When the ingestion engine ingests the Customer Call Recordings (CCRs), it additionally extracts the transcripts from given audio files and stores them in MySQL
As a result, we now have a multitenant, SAAS-based, cloud-based software application that solves the above problem for Call Centers. Its objective is to assess and rate the quality of a Call Center agent’s correspondence with the caller. The underlined objective of this app is to host an Audio Files Repository in conjunction with a Browser-based User Interface that provides a queuing mechanism to track and listen to call recordings so that Call Center supervisors can run audits and Quality Control on their client calls.
Duration
3+ Month
Estimated Working Hour
480+ hrs
