Content gap recommendations provide specific actions to help teammates improve Fin performance. They highlight where Fin couldn’t answer because help content was missing, unclear, duplicated, or contradictory.
Know what to fix and how – Content gap recommendations highlight where Fin struggled and suggest clear, specific content updates.
Skip the manual QA – Content gap recommendations scan unresolved Fin conversations, compare them to human replies, and surface what to fix—no transcript digging needed.
Fix what matters most – Each recommendation is ranked by impact so you can prioritize the fixes that improve the most conversations.
Stay in control – Edit, accept, or reject any recommendations before it goes live—so changes happen on your terms.
Note: To access Recommendations and other AI-driven insights, you’ll need the Pro add-on.
How to access content gap recommendations
To see your content gap recommendations, go to Fin AI Agent > Analyze > Recommendations and filter by Reason is Content gaps.
Note: If you click a direct link to a specific recommendation and get an error such as “That data could not be loaded,” this means the underlying content no longer exists. Pending suggestion reviews will expire automatically after being 4 weeks old to keep suggestions relevant.
How to use content gap recommendations
Content gap recommendations are generated by analyzing:
Failed Fin responses (e.g. escalations or poor-quality replies) and comparing them to successful human replies to similar questions.
Teammate-handled responses to check whether there are gaps in your knowledge base.
Duplicates of the same content in multiple sources.
Contradictions of content in different sources.
For each content gap suggestion, you can:
Review AI-generated create or edit recommendations.
See the exact conversations that triggered the suggestion.
Update public articles or snippets directly to improve future resolution.
Types of content gap suggestions
Action | Goals | Availability |
Add new content |
| Articles Snippets
|
Edit existing content |
| Articles Snippets
|
Review contradictory content |
| Articles Snippets Webpages |
Review duplicate content |
| Articles Snippets Webpages |
Segment content gap recommendations by audience
To make your recommendations more accurate and impactful, you should segment them by audience. This ensures that Fin only analyzes conversations and content relevant to a specific group of customers, preventing confusion from conflicting information (e.g., different data policies for EU vs. US customers).
Click on the settings icon at the top of the Recommendations page to segment content by Fin audiences you’ve set up.
Note:
Segmentation currently only works for content recommendations.
When you save, your current content recommendations will be cleared. New segmented content suggestions will be generated, which may take a few hours.
Tip: For multi-brand workspaces, we recommend adding a brand attribute to your audiences. This helps ensure that suggestions are generated using the correct content for each brand. To learn more about this you can refer to the article, create a branded experience with Fin Identities.
Reviewing content gap recommendations
You can review all content gap recommendations before enabling them for Fin.
Each recommendation includes:
Impact score
A summary explanation
Creation date
Source conversations
Related content
Review actions required
If you open the side drawer in the top right, you’ll find the conversations that directly informed a suggestion.
This makes it easier to:
Understand the exact customer questions behind a recommendation.
Validate whether the fix will meaningfully improve resolution.
Share concrete examples with teammates when planning changes.
Review edits
Review options:
New content: Add new content as a snippet or article
Edits: Scroll through multiple changes including:
Red text (suggested removals)
Green text (suggested additions)
Accepting a recommendation:
Snippets are immediately added to Knowledge and made available to Fin.
Article edits can be saved as draft or published to make them available to Fin.
Tip: You can edit content directly before accepting or rejecting a recommendation.
Create new articles or snippets
Some content gap recommendations might not fit naturally into your existing content. Use the Add button on the recommendation card to add as a new snippet or article, which you can immediately publish to your preferred Help Center and collection.
Move content gap recommendations to another source
If the recommendation is useful but placed in the wrong source, you can move it to different content or convert it into new content.
Search and select the target snippet or article.
The system will try to rewrite the existing content with the recommendation (typically within 20-45 seconds).
If placement isn’t possible, the content will be appended to the end.
If added as new content, the editor opens with the recommendation inserted.
Note: You can’t move a recommendation to content that already has a pending recommendations.
Remove/merge duplicate content
Duplicate content recommendations find pieces of content that contain the same information. Resolving these helps to clean up your content and prevents Fin's context window from being cluttered with redundant information, allowing it to provide better answers.
For example, a recommendation might show you two articles that both contain very similar instructions on how to reset a password.
Fix contradicting content
Our contradicting content tool helps you pinpoint content with information that is at odds with one another. This lets you quickly review and resolve the discrepancies, ensuring your knowledge base is a single source of truth. By fixing these contradictions, you'll help Fin provide clear, accurate, and reliable answers to your customers.
Involvements and resolutions are also shown per content to help you decide how to proceed.
How to act on contradicting recommendations
To resolve a contradiction, you can:
Click Edit to open and update the content.
Click Delete article or Delete snippet to remove the content.
Reject the recommendations to remove the suggestion from your view.
Mark the recommendations as done when you have made the necessary updates.
Note: Recommendations are static as of the time they were generated. Since you may edit your content before reviewing a recommendation, the last updated time is displayed above the content with a tooltip indicating that the shown preview might be outdated.
FAQs
How often are content gap recommendations created?
How often are content gap recommendations created?
Create/edit content recommendations are triggered daily or weekly, based on:
Volume: High number of conversations where a question and answer can be found.
Topic activity: Regular queries (1+ a day) on the same topic for at least 7 days.
Spikes: Rapid increases in related queries over 4 days.
Duplicate/contradictory content recommendations are checked every Sunday. This scans your content and prepares up to 20 new recommendations for you to review on Monday. These may include a mix of potential contradictions (around 15) and duplicates (around 5), depending on what's found in your content.
Why do some recommendations show an error when I open them via a link?
Why do some recommendations show an error when I open them via a link?
Direct links to individual recommendations can become invalid if the underlying recommendations or content has expired or been deleted. This is expected behavior.
To view active recommendations, open them within the Optimize dashboard in Intercom rather than via saved URLs.
What’s filtered out when generating content gap recommendations?
What’s filtered out when generating content gap recommendations?
Conversations without teammate responses
Abandoned conversations
Conversations where a teammate repeated the same answer as Fin
Conversations that mainly focus on a feature request or bug reporting
Existing content in your public articles and snippets
Why would a content gap recommendations be considered high impact if only tied to one conversation?
Why would a content gap recommendations be considered high impact if only tied to one conversation?
A recommendations can be considered high impact even if it’s tied to only one conversation when that single conversation reveals a critical gap, contradiction, or failure in Fin’s ability to resolve customer issues. This is because the impact score is not solely about the number of conversations affected, but also about the potential severity or importance of the issue uncovered.
The system does not fast-track or manually flag individual conversations for recommendations; it relies on patterns and thresholds, but a single conversation can still trigger a high-impact recommendations if it meets certain criteria.
Not all single-conversation recommendations are high impact—only those that reveal significant issues
How do I know if content was generated by AI?
How do I know if content was generated by AI?
You can filter by content Created by Fin in Knowledge to see all AI-generated content.
What’s happening to content from conversations?
What’s happening to content from conversations?
The content from conversations feature has been replaced by the new content gap recommendations feature.
For customers without access to Insights, content from conversations will remain temporarily, but will be deprecated over time.
Who can accept or reject content gap recommendations?
Who can accept or reject content gap recommendations?
Teammates with “Can create and manage content in Knowledge” permission.
Are there any limitations for content gap recommendations?
Are there any limitations for content gap recommendations?
Recommendations are only generated for conversations that have an AI topic assigned.
No option to fast-track or manually flag individual conversations for recommendations.
Low-volume customers (with fewer conversations) may receive fewer or no recommendations.
Why am I seeing older conversations in my content gap recommendations?
Why am I seeing older conversations in my content gap recommendations?
You may notice that some recommendations reference conversations that are several weeks or months old. This is expected behavior and is part of how recommendations are designed to identify meaningful patterns. Recommendations are created for a topic once enough conversations have accumulated to signal a clear knowledge gap or an opportunity for improvement.
For some topics, it can take longer to gather a sufficient volume of conversations to meet this threshold. As a result, a single recommendation can be based on a mix of both recent and older conversations. This ensures that every recommendations is well-informed and addresses a recurring theme, rather than being based on a single, isolated interaction.
Need more help? Get support from our Community Forum
Find answers and get help from Intercom Support and Community Experts