About the Client
One of the world’s largest automobile manufacturers with an extensive network of showrooms/service centers and a massive customer base was in need of an automated system for detailed call analysis, transcription, and measurement of customer sentiment.
The automobile industry is changing rapidly, thanks to the penetration of artificial intelligence and machine learning. Autonomous driving, electric powertrains, digital services, and mobile platforms have become critical pivots for this industry, which is on the verge of disruption. Much like technology businesses that have opted for digitalization and automation to become more customer-centric, automotive and mobility players are reinventing themselves for the future. The global two-wheeler market which was USD 110.82 billion in 2021, is projected to double to USD 217.94 billion by 2029, exhibiting a CAGR of 8.7% during the forecast period. Thus, it is crucial for automobile manufacturers to understand and act upon customer needs and sentiments, to ensure a good share of the market.
Understanding customer needs and measuring customer sentiment are essential requirements in any business. On any day, businesses receive multiple calls in different categories such as sales, finance, documentation, and grievance redressal. Leveraging speech and voice analysis, the client required an automated system to broadly classify the calls into sales and service categories. analyze the semantics of the calls to comprehend intent (is the call about product availability, bookings, discounts, exchange, finance, test-ride, insurance, price inquiry, service inquiry, delivery, or CSD?) gauge customer sentiment to improve customer service and derive actionable insights through detailed analytics that can guide company policies and marketing strategies (for example, insights on models and variants in high demand, inventory analytics, correlations between discounts, demand, and sales, etc.).
The client required an automated system to broadly classify the calls into sales and service categories.
We used a blend of vision AI and Deep Learning to solve the customer's challenge. Here is a breakdown of the steps we used:
Step 1: Build an automatic speech recognition (ASR) system
Azure’s automatic speech recognition service, a robust, state-of-the-art system that supports all common Indian languages, was customized and trained on the client’s customer voice data to build a suitable model. This was used to transcribe customer calls through speech-to-text conversion and perform speech and voice analysis on the customer call records, to comprehend the intent and emotion.
Step 2: Integrate a sentiment analyzer
A sentiment analyzer based on voice profiling and analysis was integrated with the ASR to understand customer emotions and gauge their sentiment based on the tone, pitch, stress, tempo, and choice of words. The sentiment analyzer classified calls into positive, negative, and neutral segments. The sentiment analyzer was used to derive insights not only into the customer sentiment but also into the response behavior of the customer care executive.
Step 3: Deploy voice and sentiment analytics dashboard
A voice and sentiment analytics dashboard were deployed to visualize the analytics results through graphs, animations, and tables that are easy and intuitive to use and understand.
- Automatic speech recognition with 95% + accuracy for Hindi and English languages. Can be customized for Tamil, Telugu, Kannada, and other regional languages.
- Precise sentiment analysis with 92% accuracy.
- Analysis of 1 Lakh + call records within a year.
- Solution scaled to all the showrooms across the country.
- Automatic call transcription through a robust speech recognition system
- Effective gauging of customer sentiment and emotion through a sophisticated sentiment analyzer
- Easy visualization of analytics results through an intuitive and user-friendly dashboard