CASE STUDY - For One of the Biggest Insurance Company in India
Data Analytics using Speech Recognition & NLP
A Business conglomerate with interests in retail and financial services was seeking to derive Customer Insights by real-time Analytics on customer calls.
The various parameters shall be derived using Azure Cognitive Services + NLP as well as Data Visualization Techniques.
Main Dashboard
Price Range Fit for the Customers who have approved of Sale Agent VISIT.
Formula for Customer
Sentiments:- Avg Sentiment
Score of a Call:-
Less Than 0% – Low
0 to 35% – Medium
More than 35% – High
For a Particular there is count of Positive Sentiment Statements(x), Negative Sentiment Statements(y) & Neutral Sentiment Statements(z).
% Sentiment Score for 1 Call => (x-y)/(x+y+z)
Priority Calls
For Priority Calls, High are the calls where Sale can be Easily Converted. Low are the ones which are difficult to be converted. Medium Calls is the middle category.
Word Cloud
Word Cloud are the words which are most Spoken in the calls.
Value Proposition
First Card Depicts if the Agent has described the Various Parameters of Value Proposition.
Technique Used – ML Based Fuzzy Match.
This Technique checks normal Keyword Phrases.
Customer Emphasis on Particular Features which is depicted
Technique Used – Regular
Expression The AI Model Checks if the Customer has repeated anything from the Dictionary of Keywords. Understanding Score is the Count if the Customer has understood the Brand Value and the Product Features.
Sentiment Score Dashboard
Basis words spoken by customer & agent, AI tool does sentiment analysis for each sentence and thereby, entire conversation.
• Apart from Sentiment, AI tool also captures Brand perception & primary Competition
• These insights helps understand customer behavior better.
• A detailed excel report for each call captures 52 parameters
