Use the Governance Dashboard to identify and resolve content issues that are actively harming user trust and adoption right now.
Addressing High Priority Content Issues
Start here when: You want to identify and resolve issues that are actively harming user trust and adoption right now.
This section covers the three tables that represent your most urgent content risks. Review these before anything else on the dashboard.
1. Content with Recent Feedback to Review
This table surfaces all content that has received a reaction, with a focus on negative feedback.
For each item you can see:
- Content Name and Type
- Reaction and Feedback text
- Username who provided the feedback
- Feedback Date
- Last Edited Date
- Views in the last 90 days
- Freshness (Fresh/Stale) and Popularity (Hot/Cold)
Below this table you will find two KPIs:
- Total Content Feedback — total number of reactions received
- Total Thumbs Down Feedback — total number of negative reactions
What to do:
| Scenario | Action |
|---|---|
| Content has negative feedback and is stale | Update the content immediately, then reach out to the user who left feedback to let them know it's been resolved and ask them to remove their negative reaction |
| Content has negative feedback but is fresh | Review the feedback text carefully — the content may be accurate but unclear or incomplete |
| High views + negative feedback | Treat as urgent — this content is actively misleading a large number of users |
Best Practices:
- Always follow up with the user after updating content — ask them to remove their negative reaction to confirm the issue is resolved. Leaving the reaction in place continues to signal distrust to other users.
- Check the Last Edited Date alongside the feedback before assuming it's unresolved, and read the full feedback text rather than dismissing it — sometimes the issue is subtle.
- Prioritize content with high 90-day views, since negative feedback there has an outsized impact on trust — don't let feedback pile up waiting for a batch review.
2. Sidekick Responses with Negative Feedback
This table shows every instance where a user indicated that an AI Sidekick response was not helpful. For each item you can see:
- Question posed to AI Sidekick
- Date the question was asked
- Username who asked the question
- Source 1, Source 2, Source 3
- Selected Source, Clicked View More
- URL and Source Available
What to do:
| Scenario | Action |
|---|---|
| AI gave an answer but user said it wasn't helpful | Review the source content — it may be inaccurate, incomplete, or poorly written for AI consumption |
| No source was available | This is a content gap — create new content to fill it |
| User clicked View More or selected a source link | The AI's direct answer was insufficient — improve the source content so the answer is more complete |
3. Unsuccessful Searches and Sidekick Chats
This table surfaces every search query and Sidekick question that returned no results. For each item you can see:
- Search Query
- Source (Search or Sidekick)
- Username
- Date
- User Role
- Results Available (will always be No in this table)
Below this table you will find two KPIs:
- No Results Found — total searches and Sidekick queries that returned zero results
- Searches with No Clicks — users found results but didn't click on any of them
What to do:
| Scenario | Action |
|---|---|
| High volume of no results for the same query | A clear content gap — create content to address this topic immediately |
| Searches with no clicks | Users found results but they weren't relevant — review the search query and top results to confirm the content actually answers user intent |
| Single user repeatedly searching the same term | Reach out personally — they may need help finding existing content or it may need to be created |
Best Practices:
- Review the No Results Found list weekly — it is the most direct signal of missing knowledge in your library, rather than waiting for users to ask for content directly.
- Once you create content for a missing search term, notify the user who searched for it. Don't ignore Searches with No Clicks either — it's just as important a signal as no results at all.
- Use the Dashboard Agent to ask "What are the most common searches with no results?" for a fast gap analysis, and prioritize unsuccessful searches by volume and user role rather than treating them all equally.