The new Optimize page helps you understand where Fin can’t resolve conversations, and what you can do to improve. It provides detailed insights into the reasons Fin escalates to teammates, along with data-driven suggestions to help you close gaps in content, data access, and automation.
Reasons for why Fin couldn't resolve
The Optimize page details the reasons conversations were sent to teammates, which include:
Missing content: Fin couldn’t respond effectively because help content was missing, incomplete, duplicative, or contradictory.
Missing access to customer data: Fin needed personalized information from an external system that wasn’t available - for example, retrieving an order status or account detail.
Missing ability to take action: Fin needed access to perform an action in another system, such as updating a workflow or cancelling an order.
Investigation needed: Fin encountered a complex or nuanced query that required human review. These edge cases can guide clearer content or targeted automation to prevent future escalations.
These edge cases can guide clearer content or targeted automation to prevent future escalations.
Understanding 'Missing content' insights
The Missing content category in Optimize provides insights into conversations where Fin AI Agent couldn't respond effectively. This typically happens because the necessary help content was missing, incomplete, duplicative, or contradictory. The Optimize page provides data-driven suggestions to help you close these content gaps and improve Fin's resolution rate.
Fin groups content-related issues by topic and provides specific actions you can take. These "create" or "edit" suggestions are triggered weekly.
'Add content' suggestions
You will see an Add content suggestion when Fin identifies a high volume of related customer questions that aren't well-covered by your existing Help Center articles.
This is a direct recommendation to create a new article to address a common query. The suggestion will outline the topic users are asking about.
'Edit article' suggestions
An Edit article suggestion appears when Fin finds an existing article that partially answers a common customer question but may be incomplete or missing key details.
Fin will link to the specific article it recommends editing. This allows you to update the existing content to be more comprehensive, directly addressing the gap Fin identified.
Data and Action Suggestions
For data and action gaps, Optimize provides additional guidance to help you move from manual escalation to automation and a better customer experience.
Each data and action suggestion consists of two parts:
1. Interim escalation guidance
These are ready-to-use guidance suggestions Fin can adhere to when it doesn’t have the required data or access to the external tools to take action. They ensure queries are routed directly to the right team, improving the customer experience immediately while the full automation is being built.
2. Automation setup instructions
These instructions help your team automate such queries fully. Each one includes:
A link to API Details, outlining example endpoints, request and response formats, and typical use cases.
A link to Fin Tasks or data connectors, depending on whether Fin needs to fetch information or take action in an external system.
A complexity score (low, medium, or high) estimating the effort required to implement automation.
Together, these tools make it straightforward to diagnose gaps, apply short-term fixes, and progressively automate more of Fin’s workload.
FAQs
How often are suggestions and insights created?
How often are suggestions and insights created?
Missing content
Create/edit content suggestions are triggered 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 suggestions are checked every Sunday. This scans your content and prepares up to 20 new suggestions 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.
Missing access to customer data and missing ability to take action
These suggestions are triggered every Sunday.
Investigation needed
These insights refresh every time you reload the page.
Need more help? Get support from our Community Forum
Find answers and get help from Intercom Support and Community Experts