Conversational data can reduce chatbot failure rates by up to 35 per cent and reduce chatbot escalation rates by up to 57 per cent while uncovering new ways to handle customer conversations
Dashbot announced the launch of its proprietary Conversational Data Cloud, letting customers build and optimise their chatbots from their businesses’ conversational data. Dashbot’s Conversational Data Cloud™ turns unstructured, noisy, interrelated and often tangled conversational data into immediate action.
Across an ever-increasing number of communication channels (contact centres, support tickets, social media, IVR, live chat, etc.), a business can get up to three million customer messages daily. The pandemic has significantly accelerated this flood of customer communications. In addition, the complexity of human language makes it impossible to predict every way users will speak with bots. As a result, over 50 per cent of chatbot sessions fail. Optimising existing bots can reduce the failure rate by up to 35 per cent and the escalation rate by up to 57 per cent.
Dashbot’s Conversational Data Cloud enables businesses to:
- Centralise conversational data, including chatbot transcripts, Zendesk, email and live agent voice calls.
- Decipher tens of thousands of daily conversations and transcripts.
- Group similar messages and topics to determine areas of failure and opportunities for new use cases, leveraging its proprietary machine learning algorithms.
“We’re expanding beyond reporting and analytics to be able to ingest raw conversational data which can be difficult but also very valuable for our customers. We’re on a mission to decipher the language, one of the most complex types of data ever existed. We listened to our customers that are challenged to make sense of all their conversational data, so we built our Conversational Data Cloud to help businesses automate, analyse and optimise their conversation channels,” said Andrew Hong, CEO, Dashbot.
Dashbot’s Conversational Data Cloud is powered by three core features:
- Transcript Transformer: Ability to search and categorise thousands of daily transcripts
- DashbotML: State-of-the-art machine learning models hyper tuned from over 10 billion conversations. Topic Modeling to visualise flow and conversation loops. Phrase clustering (message grouping) to identify new use cases and unhandled topics.
- Automated Training Data: Export messages as training data to optimise the NLP model.
