Breadcrumbs

Troubleshooting AI Search “Not Enough Information” Responses

Important: This issue can occur even when relevant articles exist on the site. In many cases, the cause is not missing content, but how the content is indexed, structured, or queried.

Problem

AI Search may return a message such as “Not Enough Information“ or “I don't have enough information to accurately answer this query” even though your site contains relevant published content.

This usually happens when the query is too short, the answer is buried in content that is hard for the AI to match, or the content has not been indexed as expected.

How to troubleshoot it

  1. Check how the query is phrased.
    AI Search works best with natural language questions, not short keyword searches. For example, instead of change password, try How do I change my password?

  2. Confirm the content is published.
    AI Search only uses content that is published to the site. Draft or unpublished pages are not included.

  3. Check whether the site has been updated.
    If your site uses manual updates, recently added or edited content will not be indexed until the site is updated.

  4. Verify that AI Search is enabled.
    Make sure AI Search is turned on in the site settings and that a valid OpenAI API key is connected.

  5. Confirm the correct model is being used.
    AI Search requires gpt-4o-mini. Using a different model can cause errors or unexpected results.

  6. Check whether the content is included in the published page tree.
    Only content that is part of the published site structure can be searched by AI Search.

  7. Review how the content is written.
    AI Search breaks content into smaller chunks, such as paragraphs, list items, and table rows. If the answer is buried inside a long list or dense section, the AI may not match it well.

How to improve results

  • Write important answers in self-contained paragraphs.

  • Use clear headings and descriptive page titles.

  • Repeat key terms naturally in the body content.

  • Use FAQ-style formatting for important definitions or common questions.

  • Encourage users to search with full questions instead of keywords.

Best practice: If a human reader cannot quickly find the answer in a specific paragraph or section, AI Search may also struggle to return it confidently.

Current limitations

  • There are no configuration options to make AI Search more tolerant of short or ambiguous queries.

  • You cannot provide custom instructions or extra context to guide the AI model.

  • Hidden metadata or search-boosting tags are not supported. AI Search relies on visible page titles and body content.

  • There is no automatic fallback to standard search results when AI Search cannot generate an answer.

  • AI Search is not case-sensitive, so capitalization does not affect results.

Example

A short query like blast radius may not return an answer, even if the term appears in your documentation. A more specific query such as What is blast radius in XM CEM and how does it relate to security posture? gives the AI more context and is more likely to return a useful result.

Last updated: