The use of transcripts to guarantee successful product sales is an increasingly prominent tool in quality analyses, the improvement of regulatory compliance and the provision of proper customer care. However, the volume of analysis may make listening by human agents impractical. Therefore, automated tools are needed to detect the most important conversations and fragments in order to ensure the proper service and advice is provided to customers, and to guarantee the validity and robustness of the operation.
Implementation of a transcription analysis support model to detect potentially problematic calls and segments. The model must be able to analyze tens of thousands of calls per month, use a transcription priority system and guide the controller to the most important segments of the recording.
Foqum’s model captures problematic segments of the call transcripts and provides them to the agent along with a risk score in an interactive dashboard, enabling aggregation and filtering by different warning types.
Claim metrics have shown that the model makes QA agents four times more efficient in detecting and classifying potentially problematic calls and segments.