Set up Microsoft Graph connectors in the Microsoft 365 admin center

This article outlines the steps to set up a Microsoft Graph connector in the Microsoft 365 admin center. The setup process is streamlined, with minimal inputs that simplify connection creation. However, you can opt for a custom setup to fine-tune specific settings.

Prerequisites

Before you begin, ensure the following:

• Admin Access: You must have either of the following roles in M365 admin center to configure a Graph connector - Global admin, Search admin or Copilot admin.
• Data Source Credentials: Collect the necessary credentials and permissions for the data source you want to connect.
• Service Account (if applicable): If your data source requires a service account, ensure it has the necessary roles or permissions.

Step 1: Add a Microsoft Graph Connector

Complete the following steps to configure any of the Microsoft Graph connectors (or click here to directly access the Connector Catalog):

1. Open the Microsoft 365 Admin Center:

   • Sign in to the Microsoft 365 admin center.

2. Navigate to Settings:

   • In the left-hand navigation pane, select Settings.
   • Click on Search & intelligence.

3. Add a New Data Source:

   • Go to the Data sources tab.
   • Click on +Add.
   • From the list of available connectors, select the data source you wish to connect (for example, ServiceNow Knowledge, Salesforce, etc.).

Step 2: Enter Basic Connection Details

1. Display Name:

   • To help users recognize the source in Copilot and search results, enter a name to identify the connector.
   • A default name is provided, but you can customize it to suit your organization's needs.

2. Data Source URL:

   • Provide the URL of your data source. For example, if you’re connecting to ServiceNow, the URL might look like https://your-organization-name.service-now.com.

3. Authentication Type:

   • Choose an authentication method to access the data source.

4. Rollout to limited audience

   • Initially, you can deploy the connector to a subset of users for validation in Copilot and other search surfaces. This feature allows you to test the integration before broader rollout.

Step 3: Create the Connection

• Click Create to set up the connection. The connector begins indexing content from your data source using default settings.

Once the connection is created, you can add a description for the connection in the success screen. To help Copilot enhance connection results for users, the description should answer the following questions briefly:

• What kind of content does this connection have?
• How do users refer to this content source in their respective orgs?
• What part of the workflow do users refer to this content in their day-to-day work?
• What are some characteristics of the content?

To learn more, refer to the Enhancing Microsoft Copilot discovery with Graph Connector Content article.

Custom Setup (Optional)

Admins who want more control over the configuration can choose the Custom Setup option. This option provides access to three tabs for detailed settings: Users, Content, and Sync.

Users

1. Access Permissions:

   • Choose whether indexed data is visible to:
     • Everyone in the organization.
     • Only people with access to content in the data source.

2. Map Identities:

   • By default, users are mapped by checking whether their email in the data source matches with a UserPrincipalName or Mail in Microsoft Entra ID.
   • If this default doesn’t work for your organization, provide a custom mapping formula.

Content

1. Manage Properties:

   • Configure properties from the data source, such as making them searchable, queryable, or refinable.
   • Assign semantic labels and aliases to enhance search relevance.

   For more information on Manage Properties, see below.

Sync

Refresh Intervals:

• Configure the frequency of data syncs between the data source and the Graph connector index.

  • Full Crawl: Synchronizes all data at scheduled intervals.
  • Incremental Crawl: Updates only the changed or new data.

• Adjust the default sync settings as required for your organization.

  For more information on Sync settings, see below.

More Resources

Guidelines for 'Manage Properties'

Content property

We recommend that you select a Content Property from the drop-down menu of options, or keep the default if one is present. This property is used for full-text indexing of content, search results page snippet generation, result cluster participation, language detection, HTML/text support, ranking and relevance, and query formulation.

If you select a content property, you have the option of using the system-generated property ResultSnippet when you create your result type. This property serves as a placeholder for the dynamic snippets that are generated from the content property at query time. If you use this property in your result type, snippets are generated in your search results.

Aliases for source properties

You can add aliases to your properties under the "Alias" column. Aliases are friendly names for your properties. They're used in queries and in the creation of filters. They're also used to normalize source properties from multiple connections such that they have the same name. That way you can create a single filter for a vertical with multiple connections. For more information, see Customize the search results page.

Semantic labels for source properties

You can assign semantic labels to your source properties. Labels are well-known tags provided by Microsoft that provide semantic meaning. They allow Microsoft to integrate your connector data into Microsoft 365 experiences such as Copilot, enhanced search, people cards, intelligent discovery, and more.

The following table lists the currently supported labels and their descriptions.

The properties on this page are pre-selected based on your data source, but you can change this selection if there's a different property that is better suited for a particular label.

The label title is the most important label. We strongly recommend that you have a property assigned to this label for your connection to participate in the result cluster experience.

Incorrectly mapping labels causes a deteriorated search experience. It's okay for some labels to not have a property assigned to it.

Search schema attributes

You can set the search schema attributes to control the search functionality of each source property. A search schema helps determine what results display on the search results page and what information end users can view and access.

Search schema attributes include options to Query, Search, Retrieve, and Refine. The following table lists each of the attributes that Microsoft Graph connectors support and explains their functions.

For all connectors except the File share connector, custom types must be set manually. To activate search capabilities for each field, you need a search schema mapped to a list of properties. The connection configuration assistant automatically selects a search schema based on the set of source properties you choose. You can modify this schema by selecting the check boxes for each property and attribute on the search schema page.

Restrictions and recommendations for search schema settings

• The content property is searchable only. After you select it in the dropdown, this property can't be used with the options retrieve or query.

• Significant performance issues occur when search results render with the content property. An example is the Text content field for a ServiceNow knowledge-base article.

• Only properties marked as retrievable render in the search results and can be used to create modern result types (MRTs).

• Only string properties can be marked searchable.

Guidelines for Sync settings

The refresh interval determines how often your data is synced between the data source and Microsoft Search. Each type of data source has a different set of optimal refresh schedules based on how often data is modified and the type of modifications.

There are two types of refresh intervals, Full refresh and Incremental refresh, but incremental refreshes aren't available for some data sources.

With a full refresh, the search engine processes and indexes the items that have changed in the content source, regardless of previous crawls. A full refresh works best for these situations:

• Detecting deletions of data.
• The incremental refresh found errors, and failed.
• ACLs (Access Control Lists) were modified.
• Crawl rules were modified.
• The schema for the connection was updated.

With an Incremental refresh, the search engine can process and index only the items that were created or modified since the last successful crawl. As a result, not all the data in the content source is reindexed. Incremental refreshes work best to detect content, metadata, and other updates.

Incremental refreshes are faster than full refreshes because unchanged items aren't processed. However, if you choose to run incremental refreshes, you still need to run full refreshes periodically to maintain correct data sync between the content source and the search index.

Crawl Scheduling

You can configure full and incremental crawls based on the advanced scheduling options present on the Refresh Settings page. Some connectors do not support incremental crawls and the option to configure incremental crawls will not be available for those connectors. For others, the incremental crawl is an optional crawl and enabled by default. A crawl schedule is selected by default for you based on the connector type. This default setting can be changed during connection creation or edited after a connection is published from the "Edit" flow of a connection. You can choose from these fields:

• Recurrence: You can choose to run the crawls every day, week, 2nd week or 4th week.
• Day(s): This option is enabled when you choose to run crawls only on specific days of the week.
• Run once a day check box lets you choose the start time of the crawl in a day. If not selected, the crawls will repeat in a day by default. You can choose the repeat interval from the dropdown.
• Frequency: Select this option if you want to repeat crawls in a day after certain time intervals. The smallest repeat frequency is 15 minutes, and the largest is 12 hours.
• Starting Time: Select the time when you want the crawl to start.
• Reset: This option will reset the schedule to the connector's default schedule.

Here are certain points to note while configuring the crawl schedule:

• If you leave any of the fields empty, or unselected, Graph connectors will pick the best time to start a crawl. For example if you choose a crawl Recurrence as "Day" and do not select the start time, Graph connectors choose the time based on your last crawl to start the new crawl. If you do not want to specify a start time of the crawl, it is a good practice to let the connector decide when to start the crawl.
• Even if the start time is mentioned, the crawl start may delay by an hour. This delay can be because of reasons such as network load etc.
• If the previous crawl overruns to the time of the next crawl, we do not stop the ongoing crawl and queue the next crawl. After the ongoing crawl is complete, we run the queued crawl only if it is of a different type (full/incremental) than the previous crawl. For example, if an incremental crawl overruns the next full crawl, we do not suspend the incremental crawl and queue the full crawl. After the completion of the incremental crawl, since the queued crawl is of different type (full), we start the full crawl immediately.

These are some of the scenarios:

• Run incremental crawl daily after every 15 minutes

Here selecting the "Run once in a day" checkbox will let you choose the "Start time" to run incremental crawl only once in a day starting at the specified time. However, unselecting it will let you choose the frequency of crawl repetitions in a day. If you want your data to be continuously refreshed, you can choose to run incremental crawls frequently in a day. But if the number of items in data source is large and the crawls tend to be longer, or if frequent updates are not needed for the content, you can choose the incremental crawl to run once in a day.

• Run incremental crawl every week on Wednesday, Saturday and Sunday repeating every 15 minutes and starting at 1:00 AM

• Run full crawl every day at 1:00 AM

• Run full crawl every week on Friday at 8:00 PM