- Copilot Search is the default in the Microsoft 365 Copilot app for licensed users, potentially superseding Microsoft Search.
- It uses a semantic index, enabling natural language queries, so users do not need to recall specific keywords to locate content.
- Copilot connectors link external systems to Microsoft Graph for an added enterprise search capability.
- Features in Microsoft Search were retired, suggesting Microsoft may no longer be investing in it.
Copilot Search is an AI assistant for Microsoft 365 that allows users to search using natural language without knowing content details. Unlike keyword-based Microsoft Search, it uses a semantic index for more relevant results and supports external data via Copilot connectors for enterprise-level results. Although its value may be difficult to quantify, for organizations planning to deploy M365 Copilot, Copilot Search may help users find content more quickly than Microsoft Search. However, since Microsoft indicated Copilot Search was an alternative to the now-retired Microsoft Search in Bing, Copilot Search may be the future if its licensing model changes.
What Is Copilot Search?
Copilot Search is an AI-driven enterprise search assistant for M365 that helps users locate content using natural language and descriptions, often without having to use specific keywords. Focused on file discovery, it is intended to help users jumpstart their work by either searching for the content they need or browsing recent files. It is available only to licensed M365 Copilot users through the M365 Copilot app and Web site. Unlicensed M365 Copilot users currently still see Microsoft Search in the M365 Copilot app and Web site.
Copilot Search is powered by a semantic index and M365 Copilot’s large language model (LLM), and it can incorporate work and Web content. The semantic index creates relationships among the data in the Microsoft Graph that are used to understand context and intent, enabling the interpretation and personalization of the user’s query. This is why users can describe what they are looking for without knowing specific keywords, although Copilot Search also supports keyword searches.
Although users still sift through search results (improved by the semantic index), most files include a “Who can see this?” option enabling users to manage file permissions for data security. Some searches may also generate a Copilot Answer, which is an AI-generated summary answering the query. Copilot Answer can include Web and work data and usually includes citations. Engaging with Copilot Answer transfers users to Copilot Chat; when returning to Copilot Search, users would have to rerun their search.
Although Copilot Search and Copilot Chat can help users find content, Copilot Search is optimized to quickly find and preview files, while Copilot Chat is better for answers and insights for specific tasks. For example, a user searching for a Q2 sales presentation using Copilot Search can quickly scan, filter, and preview a list of files to find the right one (fig. 1). Using Copilot Chat to search for the Q2 sales presentation generates a summary analysis about Q2 sales but may not include the desired presentation.
Copilot Search also offers AI-generated previews of Word and PowerPoint files on the home and search results pages, allowing users to view content without opening files. Users can filter by date, people, and file type and search across data sources including common files, like Office, and content from M365 Copilot, Teams, e-mail, Viva, and Power BI. External data can be included using Copilot connectors (see below). However, because Copilot Search favors text heavy files, finding authoritative SharePoint sites, like “Human Resources”, may be more difficult than finding the content within it and require defining Bookmarks.
Comparing Copilot Search and Microsoft Search
Microsoft Search is the default search engine for M365 and SharePoint Online, superseding classic SharePoint Search, and it is also used by users without an M365 Copilot license in the M365 Copilot app and Web site. It uses keyword-based queries to return a list of files and links, and it does not leverage a semantic index, use an LLM, ground in Web content, or create generative responses like Copilot Search.
Although Microsoft Search remains active for M365, recent retirements of features like Locations and Q&A suggest its role is diminishing. Microsoft retired its sibling, Microsoft Search in Bing, in Mar. 2025 in favor of enhancing core productivity tools, like M365 Copilot. Microsoft suggested Copilot Search as the alternative for Microsoft Search in Bing, and it is the search engine for licensed users in the M365 Copilot app and Web site, further indicating Microsoft’s direction.
Microsoft Search and Copilot Search use Microsoft Graph for storing content and indexing keywords. They also share Acronyms and Bookmarks, which admins can configure manually or use system recommendations:
- Acronyms provide definitions and descriptions for acronyms and abbreviations that are shown in search results; however, they can take up to a day to publish
- Bookmarks highlight important sites by defining a title, URL, and keywords that trigger the bookmark to be shown in search results. Bookmarks are immediately available.
These similarities may make it easier for administrators to support both search engines.
Copilot Connectors Draw from External Sources
Organizations can connect external systems to Microsoft Graph with Copilot connectors for enterprise searching. This lets M365 Copilot (and Copilot Search) use the data with associated access controls in its responses. Copilot connectors also use semantic indexing to relate and contextualize information. There are over 100 prebuilt connectors, including ServiceNow, Salesforce, Box, and Google, from third-party vendors and Microsoft, and organizations can build their own custom connectors. The data from connectors is also available to Microsoft Search.
Securing Content
Securing content is critical to avoid data leaks for users of Copilot Search. Since M365 Copilot operates in the user’s context, it only returns data to which the user has access. Improperly secured files risk exposing sensitive information to users who should not have access (called oversharing). Conversely, excessive restrictions can reduce the effectiveness of M365 Copilot by blocking qualified users from relevant information.
Tools that can assist with managing content for M365 Copilot include:
SharePoint Restricted Search. Included with M365 Copilot, SharePoint Restricted Search enables administrators to restrict Copilot search to a list of up to 100 SharePoint sites. This can help restrict Copilot Search to sites with properly configured permissions while administrators review and audit other sites.
SharePoint Advanced Management (SAM). A stand-alone suite of services to mitigate the risk of oversharing SharePoint and OneDrive content, SAM is included with M365 Copilot. In addition to reports and policies, SAM’s Restricted Content Discovery feature removes (denies) sites from M365 Copilot and Microsoft Search.
Purview. A data security and compliance suite, Purview offers several capabilities to protect content for M365 Copilot, including sensitivity labels and data loss prevention. Core Purview features are part of M365 E3 and advanced features are included in E5 subscriptions or as a stand-alone Compliance offering.
Licensing
Copilot Search is only available with a Microsoft 365 Copilot license and is not available with pay-as-you-go (PAYG) billing.
Copilot connectors by Microsoft are free, but third-party Copilot connectors may have fees.
SharePoint Advanced Management is included with M365 Copilot licenses; however, SAM requires all users to be licensed for its features, users not licensed for M365 Copilot may require a stand-alone SAM license.
Purview is licensed based on the M365 subscription or add-on pricing.
Directions Recommends
Take advantage of Copilot Search when deploying M365 Copilot. Effective enterprise search remains challenging, but M365 Copilot’s costs may be justified by Copilot Search’s semantic index, potentially producing better results for users. Raise awareness about Copilot Search and provide training to maximize its benefits, since it is only available through the M365 Copilot app and web site.
Focus on securing content before deploying Copilot Search. As with M365 Copilot in general, content security issues, such as oversharing and sensitive information, are exacerbated by Copilot Search. Ensure users are trained and educated, including the “Who can see this?” option in the search results. Apply SharePoint Advanced Management to remediate risky content and Purview for additional protection and compliance.
Monitor Microsoft Search for stagnation or deprecation. Microsoft retired features in Microsoft Search and the Microsoft Search in Bing service to prioritize “core productivity tools” and indicated that Copilot Search is the alternative, so it could be trending towards deprecation. However, Microsoft Search is unlikely to retire until customers’ adoption of M365 Copilot reaches a tipping point for Microsoft, or it changes the licensing model for M365 Copilot. Until then, Microsoft Search probably will not be improved.
Resources
Copilot Search is discussed in the Directions blog “Searching for Answers About M365 Copilot Search.”
A summary of Microsoft 365 Copilot is in the Directions product reference “Microsoft 365 Copilot.”
SharePoint Advanced Management is discussed in the Directions report “Understanding SharePoint Advanced Management.”
Microsoft Search in Bing and other Microsoft Search features are covered in “Guidance for retiring Microsoft Search in Bing for your organization” (Microsoft).