Conversational data helps you make sense of how users are interacting with your assistant. Conversational data is most useful because it shows individual users’ interactions in real time; web analytics, on the other hand, aggregates overall traffic data and is tough to act on while any individual user is browsing a site.
Understand and act on customer and employee priorities, preferences, and expectations and make smarter, user-focused decisions.
Track and gauge key search terms, preferences, satisfaction and more – allowing you to build stronger relationships and rapport with your customers.
Identify and focus on tasks that deliver maximum return on investment. Improve dialog and task execution accuracy, drastically reducing associated support and development costs.
Gain a deeper understanding of your customer’s behavior. The User Journey maps popular user paths, tasks, and fall off points visually, helping to surface patterns and trends that might go unnoticed in traditional timeline based drill downs.
Learning insights allows developers to get a complete understanding of the inference engine and how each user query is analyzed by the platform. No need to call an API – the simulator has a built-in debugger that demystifies how the NLU engine works.
Inference Insights: Developers gain visibility into how the platform calculates confidence scores, dictionary alternatives, and normalization, so they can quickly fine tune and train the accuracy of the assistant.
Sentiment Insights: Our platform programmatically predicts user emotions by extracting keywords and topics from user queries, evaluating them against common user sentiments such as anger, frustration, or fear.
Avaamo’s Continuous Learning technology improves the overall response and accuracy of the Virtual Assistant with every user interaction. Our platform also offers developers an additional “human-in-the-loop” option to further train the Assistant.
Learning via agent interaction: Specific intents as well as user frustration can trigger a tranadwe to live Agent. All live agent interactions are also fed back into the continuous learning engine.
Learning via user interaction: Users can also provide feedback via surveys and feedback forms back into the continuous learning engine.
The Unhandled Query Resolver analyzes all the user queries and provides recommendations on new intents and suggests updates to the training data to improve the overall accuracy of the bot.
Agent Diagnostics: Avaamo’s Agent Diagnostic tool uses unsupervised algorithms to analyzes all the intents and their associated training data providing the user with actions to resolve any known over/under fit in the data models.
Runtime Query Analyzer: All user queries can also be analyzed to provide the developer with additional cues to reduce the “unhandled” queries and any false positives.