VoiceSense

Power up your customer data with VoiceSense, a powerful voice analytics platform now seamlessly integrated with your CDP. This integration transforms unstructured voice data from call recordings into actionable insights. Enrich customer profiles and make smarter, data-driven marketing and business decisions.

Features

• Seamless Customer Call Data Ingestion: As a user, you can import unstructured call recording data into the CDP for analysis. This is a simple and efficient process, supporting both one-off manual uploads and automated, continuous data streams from various sources (e.g., cloud storage, call centre systems). The system automatically associates these recordings with existing customer profiles, eliminating manual data entry. This ensures that the CDP has up-to-date and comprehensive voice data for enrichment.

• Centralised Model Management & Action Notifications: The Centralised dashboard provides real-time status updates on model performance, training progress, and health. This enables the user to maintain the accuracy and effectiveness of their models without constantly checking multiple interfaces.

• Automated Topic Modelling & Profile Enrichment: Automatically analyse and tag call recordings by training a VoiceSense model to identify predefined topics like 'billing inquiry,' 'product feedback,' or 'technical support.' This process enriches customer profiles in your Customer Data Platform (CDP) by automatically mapping the identified topics back to the corresponding customer profiles, adding a valuable layer of behavioural data.

Performance Analytics & Behavioural Insights

Evaluate model performance and business impact directly from the dashboard. The dashboard provides key metrics such as model accuracy and topic frequency.

Gain valuable insights into customer behaviour trends. For example, you can correlate a surge in "cancellation intent" conversations with a recent product update. This empowers you to make data-driven decisions to improve business outcomes, whether by refining a product, adjusting a service, or optimising a marketing strategy.

Advanced Audience Segmentation & Targeting

Use voice-derived data points—the topics identified by VoiceSense—to create sophisticated, highly targeted audience segments within the CDP. Move beyond traditional demographic or purchase-based segmentation. For example, you can create a segment of customers who discussed a specific feature and target them with a personalised campaign. The system seamlessly integrates these new data points into the CDP's segmentation tools.

Steps to create a voiceSense model

• Navigate to Ramanujan AI > VoiceSense.

• Click + Create New Model. In the Model details tab, you will find the following options to configure your VoiceSense model.

• Model Name: Give your model a unique name.

• Profile Attribute to Update: Select the profile attribute that will be updated with your configured topics. You can select from any of the configured array-type profile attributes.

• Max. Topics to Update: Configure the maximum number of topics that can be updated for a single profile.

• Topic Configuration: You can create, search for and add topics that align with your business goals. This allows your VoiceSense model to track the data that matches your configured topics.

  • Configured topics: These are the topics that will be updated to the customer profile based on their call recording.

  • AI-suggested Topics: You can make use of these AI-suggested topics derived from the call recordings to whitelist topics of business interest which your customers are talking about. You should publish the model for Ramanujan AI to analyse the recordings.

• Save or Publish: Click Save to use it later or Publish the model you configured right away.

On the main page of VoiceSense, you can find both published and saved models. By clicking the three dots in the Actions column, you can review and update models as needed. You can also deactivate a published model by selecting the Deactivate option.

Analytics

• Navigate to the Analytics tab of a specific model

• Use the calendar in the top-right corner to fetch insights for a specific date range.

• Overview: This section shows the following details.

  • Total Minutes Analysed: The total duration of voice recordings analysed by this model.

  • No. of Recordings: The total number of recordings analysed.

  • AI Suggestions Accepted: The total number of topics accepted from AI suggestions.

  • AI Suggestions Dismissed: The total number of topics dismissed from AI suggestions.

• What your users are talking about: This word cloud displays the topics discussed by users, with high-propensity topics appearing larger and more evident. In contrast, other topics are positioned based on their propensity.

• Sentiment: This section provides insights into users' sentiments and emotions regarding any topic analysed from the voice recordings.

  • Topics: A list of topics discussed by users from the configured topics.

  • Total Conversations: Total conversations having discussions around a specific topic

  • Appreciation: The percentage of users who expressed appreciation for the topic.

  • Neutral: The percentage of users who expressed a neutral response.

  • Disappointment: The percentage of users who expressed disappointment.

  • Frustration: The percentage of users who expressed frustration.

  • Escalation: The percentage of users who escalated issues related to the topic.

• Sentiment Over Time: A graph showing sentiment data over time.

• Sentiment Distribution: A pie chart showing the overall distribution of sentiments for topics within the model.