You know that moment when you’re searching for something at work and you get 47 results? You click through them. Nothing’s quite what you need. You refine your search. More clicking. More scrolling. Fifteen minutes later, you finally find the right document. And it was on the third page of results the whole time.
The Problem Nobody’s Solving (Until Now)
Knowledge workers spend a shocking amount of time just finding information. A 2024 study found that knowledge workers spend about 20% of their day searching for documents and information. That’s one day a week. Per person. Per organisation, it’s thousands of hours wasted on document hunting.
Search is broken. We’ve pretended it’s fine for years. We’re just… used to it being broken.
Microsoft just quietly started fixing it. And most people haven’t even noticed.
What Are Glance Cards Anyway?
When you search for something in your Microsoft 365 tenant now, you don’t just get a list of filenames. You get a summary. Creation date, key insights, relevant highlights—basically, enough information to know whether it’s the right document before you click on it.
It sounds small. It’s not.
Instead of opening 10 documents to find the one you need, you scan five summaries and open the right one. That’s 80% of your search time gone.
But Wait, There’s More
The really useful bit? Your chat history surfaces in search results. So if you had a conversation with Copilot about quarterly financial forecasts six weeks ago, and you need that information again—it shows up in your search results. You don’t have to remember you talked to Copilot about it. You don’t have to dig through your chat history. It just… appears.
You can filter results by conversation, document type, date range. The experience feels less like searching in the dark and more like having an assistant who remembers where you found stuff before.
Why This Matters for Your Team
Let’s be practical. Every hour someone saves searching is an hour they can spend on actual work. For a team of 20 people, if this feature saves each person just 2 hours a week, that’s 160 hours recovered annually. For a team of 100, it’s 800 hours.
That’s not nothing.
But more importantly, it changes how people work. Instead of abandoning a search because it’s not finding what they need, they try a different approach. They find more context. They understand their own information better. Information that would have stayed buried becomes useful.
The Sales Team Example
Here’s a real scenario: a sales manager needs to prepare for a client meeting. She needs the original contract, the latest statement of work, and the customer communication log. Old way: she searches three times, opens documents she doesn’t need, asks a colleague if they remember where something was filed. 20 minutes.
New way: she searches once. The results include a summary of the contract (dates, key terms, pricing), the SOW preview, and it suggests the email thread where that customer concern was discussed. She scans the summaries, opens the three documents she actually needs. 3 minutes.
And if she asks Copilot “summarise our history with this client,” the summary itself appears in her next search. She doesn’t lose it.
How Glance Cards and AI Search Work Together
The power of this feature comes from the combination of multiple technologies working in concert. Let’s break down what’s actually happening under the hood.
The Indexing Layer
When you search in Microsoft 365, the search engine doesn’t just look at filenames and metadata. It indexes the actual content of documents—including PDFs, Word documents, presentations, and even email content. This means search can find information based on what’s written inside documents, not just what people remember to tag.
The Summarisation Engine
Copilot then processes these indexed results and generates summaries. This isn’t simple keyword matching. The AI understands context, meaning, and relevance. It can summarise a 20-page technical document into a few key points that actually matter for your search.
Conversation Threading
Every conversation you have with Copilot gets indexed the same way as documents. So when you ask Copilot “summarise Q3 financial performance” and have a detailed conversation about it, that conversation becomes searchable. Six months later, you search for “Q3 financial analysis” and your conversation appears alongside the original documents.
This is fundamentally different from traditional search because it treats your intellectual work—the conversations where you understand information, not just access it—as a searchable resource.
Implementation Challenges and How to Handle Them
Challenge 1: Information Chaos
Your search is only as good as your information. If documents are scattered across SharePoint, Teams channels, and personal drives with inconsistent naming, no amount of AI will help.
Solution approach:
- Start with a content audit to understand current state
- Establish basic naming conventions (date, project, version number)
- Consolidate documents to shared locations rather than personal drives
- Tag documents consistently using consistent metadata
You don’t need to wait for perfect governance. Even basic improvements dramatically improve search effectiveness.
Challenge 2: Permission Boundaries
Search respects security permissions. If someone shouldn’t see a document, it won’t appear in their search results. This is correct behaviour for security, but it means you need to ensure permissions are set properly.
Common permission issues:
- Documents owned by individual employees who’ve left
- Files in shared drives with unclear access rights
- Conflicting permissions between parent and child folders
- Users with insufficient permission to documents they need
Audit and clean up permissions as part of your preparation.
Challenge 3: Data Sensitivity
Some organisations hesitate to index sensitive information because they worry about search privacy. This is legitimate. But consider: if you’re not searching sensitive information, you’re already managing it differently. The question is whether AI indexing makes it more or less secure.
In practice, Microsoft’s indexing and search are highly secure. AI summarisation happens in your tenant, not in external systems. Permissions are enforced throughout. Your search data doesn’t train public AI models.
The Catch (There’s Always a Catch)
This only works if your data is organised reasonably well. If your SharePoint is chaos—documents named “Final_FINAL_v3_ACTUAL,” stored randomly, inconsistently tagged—then search summaries won’t help much.
But if you’ve got decent information governance, even mediocre organisation becomes functional. Because Copilot’s summary shows what’s actually in the document, not just the filename or metadata.
Measuring the Impact: Real Numbers
Don’t just implement this and hope. Measure the impact with data.
Before/after metrics to track:
- Average time per search (from query to opening right document)
- Search accuracy (percentage of first result being the right one)
- Documents opened per search (indicates efficiency)
- Search queries per person per week (more queries = more discovery)
- User satisfaction with search (survey-based)
A manufacturing company we worked with tracked these metrics. Before Copilot search: average 4.2 minutes per search, user success rate 62%. After: 1.8 minutes per search, 87% success rate.
For a company with 200 knowledge workers, that’s meaningful productivity recovery.
The Advanced Features Nobody Talks About
Beyond the basic summarisation, there are capabilities that multiply the value:
Search Filters and Refinement
You can filter results by content type (documents vs. emails vs. conversations), date range, even conversation topics. This lets you narrow results precisely without multiple searches.
Saved Search Refinements
Frequently search for the same types of information? Save your search refinement. Next time you search for related information, the saved refinement appears as a suggestion.
Integration with Other Microsoft 365 Features
Your Teams channels are searchable. Your OneNote notebooks are searchable. Your Outlook email threads are searchable. This unified search across all your information creates a single source of truth.
Planning Your Deployment
Phase 1: Assess (Week 1-2)
- Audit current document storage and organisation
- Identify permission issues
- Establish baseline search metrics
- Identify power users who can champion the feature
Phase 2: Prepare (Week 3-4)
- Clean up obvious document chaos
- Fix critical permission issues
- Establish basic naming conventions
- Create communication plan for users
Phase 3: Enable (Week 5)
- Enable Copilot search for pilot group (perhaps 50-100 users)
- Provide training on how to use new summarisation features
- Monitor for technical issues
Phase 4: Expand and Optimise (Week 6+)
- Roll out to broader audience
- Gather user feedback
- Fine-tune based on usage patterns
- Measure impact against baseline metrics
What This Means for Your Organisation
If you’re deploying Copilot, this feature alone justifies the investment. Not because it’s flashy. Because it’s useful every single day for every person who searches for anything.
And the kicker? It gets better the more people use it. The more conversations happen with Copilot, the richer the search context becomes. Six months in, people stop thinking about “searching” and just start asking for what they need—whether they’re querying documents directly or asking Copilot.
The Reality Check
You don’t need to do anything special to get this. You don’t need new tools or new processes. If you’re using Microsoft 365 and you’ve got Copilot enabled, search just gets better. It happens automatically.
The only prerequisite is reasonable data governance. Clean up your permissions. Keep your documents reasonably organised. Don’t name everything “Final_v3.” That’s it.
Do that, and your team gets hours back every week. They spend less time hunting and more time thinking. And honestly, in a world where time is everyone’s scarcest resource, that’s pretty valuable.
Want to talk about how to prepare your organisation for better search and discovery? We help teams get their data house in order so they can actually benefit from these features.
