fair-rose•17mo ago
Analytics for the chatbot
I have made a simple chatbot basically a Q&A from the provided knowledge base nothing much to it. I tested the bot and it was able to always answer from the files I provided in the KB yet when I go to check the analytics, it says the all my queries got registered as a fallback even though the answers are 100% correct. please look at the image below for better understanding.
1 Reply
The situation you're encountering seems to be related to how Voiceflow categorizes and prioritizes responses. In Voiceflow, there are three levels of priority when answering a user's question:
1. Intents: These are designed by you to trigger based on specific questions.
2. Knowledge Base (KB): The KB is a collection of documents that power the Q&A of your assistant.
3. Global No Match: This is a last resort message if the assistant cannot find an answer in your KB.
Even though your chatbot is correctly answering queries using the Knowledge Base, it might be registering them as fallbacks due to the way the system is set up. The fallback mechanism is triggered when the assistant can't find a matching intent or FAQ set. If no match is found, then the KB Fallback will trigger the Retriever and Runtime services.
In your case, it seems like the system might be bypassing the intent matching and going straight to the fallback, which is why your queries are being registered as such. This doesn't necessarily mean that your chatbot is not functioning correctly, but rather it's a reflection of how the system is categorizing the responses.
If you want to change this behavior, you might consider setting up specific intents for the types of queries your chatbot is handling. This way, the system will recognize these queries as intent matches rather than fallbacks. However, this would involve a more complex setup process and might not be necessary if your chatbot is already answering queries correctly.
Remember, the analytics are there to help you understand how your assistant is being used and to identify areas for improvement. If your assistant is functioning as expected and providing correct answers, then the fact that these are being registered as fallbacks might not be a significant issue.