Use response scores to guide follow-up behavior
In addition to generating a follow-up question, Aftercare's enhanced follow-up API will generate scores that can help to guide follow-up behavior. See more details in our API documentation
here.
A score between 0 and 100 that represents how exhaustively the response answered the topic question and question context. A high-completeness response would offer little room for a follow-up question whereas a low-completeness response would invite for one or more follow-ups.
A score between 0 and 100 that represents how pertinent the response is to the topic question and any question context provided. Responses low in relevance are usually non-answer, gibberish, or just completely unrelated answers.
A score between 0 and 100 that represents how well the response agrees with or builds upon previous answers in the conversation. Responses with high consistency maintains the same stance, logic, or narrative as earlier responses, while a response with low consistency may contradict or diverge from what was said before.
A score between 0 and 100 that represents how coherent and logical the response itself is (i.e., gibberish, troll responses, etc.). Responses high in nonsense may indicate low engagement by the respondent.
