---
doc_version: "1.0.0"
last_updated: "2026-07-14T15:00:57.718Z"
title: "How to Find and Fix Knowledge Gaps"
description: "Use ChatbotIQ's analytics to discover what questions your bot can't answer, identify missing content, and improve your knowledge base."
keywords: [knowledge gaps, analytics, confidence scores, bot improvement, content gaps, ChatbotIQ]
---

Your bot can only answer questions about content in your knowledge base. When visitors ask about something that's missing, that's a knowledge gap. ChatbotIQ detects these automatically and shows you exactly what content to add.

---

## Open Analytics

1. Go to **Analytics** in the sidebar.
2. Select a **bot** (or view all bots) and a **time period** (7, 30, or 90 days, or a custom range).

---

## Read the summary cards

Four cards at the top give you a quick health check:

| Card | What it tells you |
|------|------------------|
| **Total Queries** | How many questions users have asked in the selected period. |
| **High Confidence** | Questions where the bot found strong matches in your content. These are working well. |
| **Low Confidence** | Questions where matches were weak. The bot answered, but may not have been accurate. |
| **No Results** | Questions where nothing relevant was found at all. These are your biggest gaps. |

> **Goal:** You want High Confidence to be as large as possible, and No Results as small as possible.

---

## Understand the confidence distribution

The bar chart breaks down all queries into four bands:

- **High** - strong matches. Bot is performing well here.
- **Medium** - decent matches but could be better. Consider adding more detailed content.
- **Low** - weak matches. The bot may be giving unreliable answers.
- **No Results** - nothing found. The bot had to say "I don't know" or gave a general response.

---

## Explore knowledge gap clusters

This is the most actionable section. ChatbotIQ groups related unanswered questions into **topic clusters** using AI. For each gap, you see:

- **Topic name** - a summary of what users were asking about.
- **Query count** - how many questions fell into this topic.
- **Average confidence** - how poorly the bot performed on these questions.
- **Decline rate** - how often the bot had to say "I don't know."
- **Gap type** - Content gap (missing from your docs), Product gap (feature doesn't exist), or Out of scope (not relevant to your bot).
- **Sample queries** - actual questions users asked, so you can see exactly what they wanted to know.

---

## Fix a knowledge gap

Once you've identified a gap, here's what to do:

1. **Content gap:** Add the missing information to your knowledge base. Either update your website and re-crawl, or add a [Q&A pair](/reference/source-types/) with the answer directly.
2. **Product gap:** If users are asking about a feature you don't have yet, note it as product feedback. Once you've actioned it, you can mark the gap as resolved.
3. **Out of scope:** If the question isn't relevant to your bot's purpose, consider adjusting your bot's system prompt to explain its scope.

After adding content, refresh your source and check the analytics again in a few days. The gap should shrink as the bot starts answering those questions.

---

## Filter and track gaps

- **Filter by type** - focus on content gaps (the ones you can fix) vs. out-of-scope queries.
- **Filter by period** - choose monthly or quarterly granularity to see trends.
- **Mark as resolved** - once you've addressed a gap, mark it resolved to track your progress.

---

## Related

- [What Are Knowledge Gaps?](/explanation/what-are-knowledge-gaps/) - understand how gap detection works
- [Improve Your Bot's Answers](/how-to/optimize-bot-answers/) - other ways to improve quality
- [Keep Content Up to Date](/how-to/refresh-and-maintain-sources/) - re-crawl after adding content
- [Review Conversations](/how-to/review-conversations/) - see the actual conversations behind the numbers

## Sitemap

Full documentation index: [sitemap.md](/sitemap.md) and [llms.txt](/llms.txt).