Microsoft Clarity Copilot: How to Use AI Summaries for UX Insights

Microsoft added Copilot to Clarity, bringing AI-generated summaries and natural language queries to behavior analytics. Here's what it does well, where it falls short, and whether it's enough for serious UX analysis.

What is Clarity Copilot?

Clarity Copilot is Microsoft's AI assistant built directly into the Clarity dashboard. Powered by the same AI technology behind Microsoft Copilot, it analyzes your session data and provides summaries, answers questions, and highlights patterns — all through a conversational interface.

Think of it as a first pass at understanding your analytics data without watching hours of session recordings or manually interpreting heatmaps.

How to enable Clarity Copilot

Copilot is available to all Clarity users — no additional setup required.

1. Log into your Clarity project at clarity.ms.
2. Look for the Copilot icon in the top navigation bar (it looks like a sparkle/star icon).
3. Click it to open the Copilot panel.
4. Start typing your question or browse the suggested prompts.

Copilot also appears contextually — when you're viewing a specific session recording, heatmap, or dashboard page, it can provide insights relevant to what you're looking at.

What Copilot can do

AI-generated session summaries

Instead of watching a full 5-minute recording, Copilot generates a text summary of what happened in the session: which pages were visited, where the user clicked, what frustration signals occurred, and how the session ended (conversion, bounce, exit).

Example summary:

"The user visited the homepage, navigated to the pricing page via the top menu, scrolled to the bottom of the pricing table, rage-clicked on the 'Enterprise' plan button 3 times (it appeared unresponsive), then left the site."

This is genuinely useful for triaging sessions. Instead of watching every recording, you can scan summaries and only watch the ones that reveal interesting problems.

Natural language queries

You can ask Copilot questions about your data in plain English:

• "What are the top pages with rage clicks this week?"
• "Which traffic source has the highest bounce rate?"
• "Show me sessions where users visited the pricing page but didn't convert."
• "What are the most common frustration signals on mobile?"

Copilot translates these into filters and presents the relevant data. This is faster than manually navigating through Clarity's filtering UI, especially for complex queries.

Trend explanations

When your dashboard shows a spike in rage clicks or a drop in engagement, Copilot can attempt to explain why — pointing to specific pages, device types, or traffic sources that contributed to the change.

Suggested insights

Copilot proactively surfaces patterns it finds interesting:

• "Mobile users on your checkout page have 3x more rage clicks than desktop users."
• "Users from Google Ads spend 40% less time on the pricing page than organic visitors."
• "Your product page scroll depth dropped 15% this week compared to last week."

What questions to ask Copilot

To get the most value from Copilot, focus your questions on specific, actionable areas:

Frustration analysis

• "Which pages had the most dead clicks in the last 7 days?"
• "What elements are users rage-clicking on the homepage?"
• "Are there more frustration signals on mobile or desktop?"

Conversion optimization

• "What's the average scroll depth on my landing page?"
• "How many users click the CTA button on the pricing page?"
• "What percentage of users who visit the product page reach checkout?"

Comparing segments

• "How does engagement differ between new and returning visitors?"
• "Which country has the highest frustration signal rate?"
• "Do Chrome users have more issues than Safari users?"

Limitations of Clarity Copilot

Copilot is a helpful starting point, but it has significant limitations that serious UX teams should understand.

Surface-level analysis

Copilot tells you what happened but rarely digs into why or provides concrete recommendations. It might say "rage clicks increased on the checkout page" but won't tell you "the coupon field validation is too strict, causing users to re-click submit — here's what to fix."

No historical comparison

Copilot works with your current data window. It can't compare this month to last month or track improvement trends over time. For ongoing optimization, you need to build your own comparison workflow or use a tool that provides historical context.

Limited depth on multi-step analysis

Ask "Why is my conversion rate dropping?" and Copilot might point to a page with high exits. But it won't follow the chain: high exits → caused by rage clicks on the checkout button → caused by a JavaScript error on Safari → introduced in last week's deploy. That kind of root cause analysis still requires human investigation or more sophisticated AI tools.

Accuracy concerns

Like all AI assistants, Copilot can generate incorrect or misleading summaries. It may misinterpret the context of a session or present correlations as causations. Always verify Copilot's claims by watching the actual session recordings or checking the raw data.

No automated delivery

Copilot is a pull-based tool — you have to log into Clarity and ask it questions. It doesn't proactively email you when something important happens or send you weekly digests of UX issues.

Copilot vs full AI reporting tools

Clarity Copilot is a chat interface on top of your Clarity data. Full AI reporting tools like ClarityInsights take a fundamentally different approach:

Copilot excels at quick, ad-hoc exploration. When you notice something odd in the dashboard and want a fast answer, Copilot delivers. But for systematic, ongoing UX improvement — the kind that actually moves conversion metrics over time — you need something that proactively analyzes your data, tracks trends, and delivers actionable recommendations without you having to ask the right questions.

Best practices for using Copilot effectively

1. Use it for triage — Scan session summaries to find the most interesting recordings, then watch those manually.
2. Ask specific questions — "What frustrated mobile users on /checkout last week?" beats "What's wrong with my site?"
3. Verify claims — Always cross-check Copilot's observations against the actual data. Watch the recording it references.
4. Combine with filters — Use Copilot to identify patterns, then apply manual filters to validate them with larger data sets.
5. Don't rely on it exclusively — Copilot is one input. Pair it with your own heatmap analysis, session recording reviews, and quantitative analytics data.

The bottom line

Clarity Copilot is a solid free addition that makes Clarity easier to use, especially for non-technical teams who find the filtering interface intimidating. It saves time on session triage and provides a faster way to query your data.

But it's not a replacement for deep UX analysis. If you're serious about using behavior analytics to improve your site, you'll want to combine Copilot with regular manual reviews and consider automated AI reporting for consistent, proactive insights.