What are knowledge gaps in an AI chatbot?

Short answer

Knowledge gaps are the questions customers ask that an AI chatbot can't answer because the information isn't in its knowledge base. Finding and filling them is the quickest way to make a chatbot answer more questions correctly.

Also called: unanswered questions, missing answers, content gaps.

By Updated

Where knowledge gaps come from

  • Things everyone in the business knows but nobody wrote down, like parking or whether you take walk-ins.
  • New products, prices or rules that aren't on the website yet.
  • Information stuck in files that were never added, like a price list in a PDF.
  • Edge cases: exceptions, combinations and "what if" questions.

Why tracking them matters

Every gap is a conversation your team has to pick up, or a risk of a wrong answer. A list of real unanswered questions tells you exactly what to write, in the order customers need it, instead of guessing which FAQ to update.

For a real estate agent it might be "do you handle rentals?", for a webshop "can I change my delivery address after ordering?". One answer added closes the gap for everyone who asks after that.

Missing answers in SimplyBoost

  • Questions the agent could not answer are collected under Missing answers.
  • Add the answer there and the agent uses it from then on.
  • If the agent couldn't answer twice in a row, the conversation goes to your team, so the customer isn't left stuck.
  • Every plan can learn from every source type, with files up to 50 MB and no limit on the number, so you can also close a gap by uploading the document that holds the answer.

Questions people ask

How do I find knowledge gaps in my chatbot?

Look at the questions it couldn't answer. In SimplyBoost those are collected under Missing answers, so you don't have to read through every conversation.

How often should I check for knowledge gaps?

Weekly at first, while the agent is new and the gaps are many. Once the common questions are covered, checking after a launch, a price change or a new policy is usually enough.

Is a knowledge gap the same as a hallucination?

No, but one can lead to the other. A gap is missing information. A hallucination is a made-up answer, which a model can produce when it has no information to answer from.

More on ai and knowledge