Questions are how you handle replies the user types that don't correspond to a button the chatbot has sent. They may be actual questions, such as "how much do tickets cost?", statements indicating the user's intent, such as "book appointment", complaints, requests, updates or other responses.
You can also configure how the chatbot handles responses it can't match.
When your chatbot receives a user response that does not match a button it has sent, it tries to match the response against questions you have defined. Questions can have different context levels, which allow you to tailor your responses to a chatbot user's question in a more natural fashion.
Manage your context levels with:
- Inbound questions, which can match chatbot user questions within a specific conversation, or anywhere in your conversation flow.
- Outbound questions, which only match chatbot user questions sent directly after a specific passage.
You can match user responses with two types of questions:
- Keyword questions, which match words or phrases you define.
- Natural Language Process (NLP) questions, which use machine learning to match training phrases you define. You can create entities to help streamline your training phrases.
The chatbot attempts to match a question by keyword before checking the NLP questions.
Handling unexpected responses
When a user sends a question that your chatbot can't match, it uses its Catch-All passage response, configured in the Fallback settings. You can configure a specific Catch-All to be used in different conversations, and a different Catch-All for very long questions, or unexpected attachments.
A user may also send a question that matches a passage your chatbot recently sent them. Sometimes this is expected: if they're asking for an update on where their package is, receiving a near-identical response is fine. But in other conversations, it would feel unnatural to send them the same passage again. Instead, you may want to guide them to rephase their question or request to talk to a live agent. You can configure Intellimem to handle this for your whole chatbot or within a conversation context in your chatbot's Fallbacks settings.