THE PROBLEM
Client, a multi-national telecom services company, conducts telephonic surveys to obtain key business insights such as brand perception, competitor benchmarking, and purchase drivers. However, the process to extract insights from these audio recordings was manual and non-standardized. In addition, the slowness of the process was hindering them from acting upon insights related to churn drivers in a timely manner. Hence, the client partnered with Inxite Out to automate the extraction of insights from these outbound call recordings in a fast, scalable manner.
INXITE OUT APPROACH
An end-to-end solution was developed leveraging MEGHNAD, Inxite Out’s Conversation Intelligence Framework. The solution consisted of the following steps:
Brand Perception Understanding using Meghnad
Transcription
MEGHNAD’s multi-lingual transcription engine was used to transcribe the call recordings with a significant degree of background noises. The transcribed text files generated had the following elements making them amenable for downstream processing:
-
- Speaker-Segmentation: Clearly demarcates the utterances of the caller from the responder.
- Confidence Score: Quantitative indication of the goodness of transcription in the presence of noise – enables the filtering of files with very high degree of disturbance.
Primary Insights Extraction
Leveraged MEGHNAD’s Generative AI-based reading comprehension and extractive QA capabilities supported by business context-specific prompt templates to automatically extract three types of information from each such transcribed text file:
-
- Various demographic information (e.g., name, age, city, gender, nationality etc.) of the respondent,
- Usage information (e.g., current service provider, prepaid/postpaid, typical monthly expenditure distribution across local/international / internet etc.),
- Perception by service attribute by brand, as expressed in the call.
Synthesis of Augmented Insights
Primary Insights extracted from each call were aggregated and summarized to obtain key business insights such as brand perception and competitor benchmarking across purchase drivers, churn drivers etc. Identified key patterns by slicing and dicing the data across different customer micro-clusters according to demographic and usage profiles.
RESULT
- Client leveraged understanding of churn drivers to re-evaluate churn prevention strategies.
- 250+ telephonic calls (2000+ conversational audio minutes) analyzed on monthly basis, with turnaround SLA of 2 days; 60%+ improvement in conversation analysis time
