A Missed Utterance is a question asked of your chatbot that the bot didn’t have an answer for (indicated through the response “I’m sorry, I don’t know the answer to that question”).
Missed Utterances determine the quality of your end users’ interactions with your chatbot. If your users are having high-quality interactions, then the value you get out of Conversational AI increases.
The graph in the Missed Utterances pane shows the total number of utterances and the number of missed utterances. Graphs can be displayed for the last seven (7) days, the last thirty (30) days, or the last ninety (90) days.
The best practice is to download the raw data, filter to see only missed utterances, and then use those questions asked to either add question variants or add new question/answer pairs to your knowledge base.
To determine the effectiveness of your existing question/answer pairs, calculate the success rate of known utterances, or questions that the chatbot recognized. For example, in the past 7 days you have 955 total utterances, or questions asked. Of those 955, you have had 150 missed utterances. Dividing the 150 missed utterances by the 955 total utterances shows that you have had a 84.3% success rate.
In the chatbot world, anything over 80% known utterances is fantastic. With consistent review of the missed utterances and maintenance of your question/answer pair knowledge base, you can easily get your success rate close to 90% and even larger. A higher success rate means your end users are having more meaningful interactions with the chatbot.