SharePoint search and Microsoft Search

Search in Microsoft 365 went from "SharePoint Search" to "Microsoft Search" around 2019. The underlying index is shared across SharePoint, OneDrive, Teams, Outlook, and Bing-at-work, and the user-facing surface is the same search box across the apps. Knowing how it works helps both users and admins.

The crawl mechanics and staleness troubleshooting are in SharePoint search and crawl architecture; tenant-level administration is in Microsoft Search administration, and the difference between search and Copilot retrieval is the subject of Copilot grounding vs SharePoint search.

What's indexed

Microsoft Search indexes:

• SharePoint sites and pages — content, metadata, columns.
• Files in SharePoint and OneDrive — including full-text content for supported file types.
• Teams messages and channel content.
• Outlook mailboxes (a user's own mail, primarily).
• People — names, titles, departments, skills, hierarchy from Entra ID.
• Conversations and Loop components.
• Optional third-party connectors — Jira, ServiceNow, GitHub, file shares, and more — via Microsoft Graph connectors.

Each user's search results are personalised and trimmed to their permissions: you only see what you can already access.

Where users search

• The search box at the top of Microsoft 365 apps (SharePoint, Teams, Outlook, Office apps, Bing-at-work).
• The SharePoint search results page for a site or hub.
• The Microsoft Search browser address bar (bing.com signed in with a work account).

Results are unified — files, messages, and people show up in one list, with verticals for files, people, sites, etc.

How admins influence results

Admins control search through the Search & Intelligence admin centre (admin.microsoft.com/search):

• Answers: predefined answers for known questions (e.g. "VPN" surfaces the VPN page).
• Bookmarks, Q&A, and acronyms.
• Locations: known office addresses.
• Promoted results and verticals.

For larger or external sources, Graph connectors bring third-party content into the index.

Improving result quality

A few practical levers:

• Title and description metadata on SharePoint pages matter — they're what shows in results.
• Sensitivity labels and retention ensure old, stale content isn't promoted forever.
• Site hub associations define search scopes: a search from a hub searches across all associated sites.
• Search analytics show what users searched for and which results they clicked — use this to seed bookmarks and answers.

For Copilot tenants, search quality directly affects Copilot quality. Stale, mis-titled, or mis-permissioned content shows up in answers.

Search vs Copilot retrieval — a common confusion

Microsoft Search and Copilot's retrieval both draw on the same underlying index and both respect the same per-user permission trimming, which leads many admins to assume "if it's in search, it's Copilot-ready" — that's not quite right. Search surfaces a ranked list of documents for a user to open and read themselves; Copilot additionally has to extract and synthesise content from what it retrieves, which means a technically-indexed but badly-structured document (no headings, ambiguous filenames, content buried in an image or scanned PDF) can show up perfectly well as a search result while contributing little or nothing useful to a Copilot-generated answer. Improving search ranking and improving Copilot answer quality are related but not identical exercises — fixing titles and descriptions helps both; fixing document structure (real headings, extractable text rather than scanned images) matters specifically for the Copilot half.

Frequently asked questions

Why do two users see different results for the same search term? Results are trimmed to each user's own permissions and further personalised by their own search and interaction history — a manager who regularly opens a specific policy document will tend to see it ranked higher than a colleague who's never opened it, even though both have equal access to it.

Can search results be forced to exclude a specific site entirely? Yes — a site can be excluded from the search index (and, separately, from Copilot's grounding) via the SharePoint admin center or, at greater granularity, through SharePoint Advanced Management's restricted access and archive controls, which remove a site from active search and Copilot retrieval without deleting its content.

How current are search results relative to a document that was just edited? Typically minutes, not hours — the crawler picks up changes on a short cycle for most content types, though very large libraries or unusual file types can lag longer; a document that seems to be missing from search immediately after being saved is more often a permissions or file-type issue than a crawl-delay one.