May 3, 2021 5 min read

Microsoft Graph Data Connect: Bulk Access to Microsoft 365 Data

Azure Microsoft Graph Data Microsoft 365 Data Engineering

Microsoft Graph API is great for accessing Microsoft 365 data on a per-user, real-time basis. But what if you need to analyze data across thousands of users? That’s where Microsoft Graph Data Connect comes in.

The Challenge with Graph API at Scale

Using the standard Graph API for analytics has limitations:

Rate limiting kicks in quickly
Pagination through millions of records is slow
Token management at scale is complex
Near-real-time patterns don’t fit batch analytics

Graph Data Connect solves this by delivering bulk extracts of Microsoft 365 data directly to Azure Data Factory.

How It Works

Configure a data pipeline in Azure Data Factory
Request specific datasets (emails, calendar, files metadata)
A Microsoft 365 admin approves the request
Data is delivered to your Azure storage

The data lands in Azure Storage as JSON files, ready for processing with Synapse, Databricks, or any data platform.

Setting Up a Pipeline

First, register an app in Azure AD with the Graph Data Connect permissions:

{
  "name": "graph-data-connect-app",
  "requiredResourceAccess": [
    {
      "resourceAppId": "00000003-0000-0000-c000-000000000000",
      "resourceAccess": [
        {
          "id": "810c84a8-4a9e-49e6-bf7d-12d183f40d01",
          "type": "Role"
        }
      ]
    }
  ]
}

Then create the Azure Data Factory pipeline:

{
  "name": "CopyEmailMetadata",
  "properties": {
    "activities": [
      {
        "name": "CopyFromOffice365",
        "type": "Copy",
        "inputs": [
          {
            "referenceName": "Office365EmailDataset",
            "type": "DatasetReference"
          }
        ],
        "outputs": [
          {
            "referenceName": "AzureBlobOutput",
            "type": "DatasetReference"
          }
        ],
        "typeProperties": {
          "source": {
            "type": "Office365Source",
            "dateFilterColumn": "receivedDateTime",
            "startTime": "2021-04-01T00:00:00Z",
            "endTime": "2021-05-01T00:00:00Z",
            "userScopeFilterUri": "https://graph.microsoft.com/v1.0/groups/{group-id}/members"
          },
          "sink": {
            "type": "BlobSink"
          }
        }
      }
    ]
  }
}

Available Datasets

Graph Data Connect provides access to:

Email

Message headers and bodies
Sent and received metadata
Attachment information (not content)

Calendar

Meeting metadata
Attendee information
Recurring event patterns

People

Organization hierarchy
Contact relationships
Collaboration patterns

Files (OneDrive/SharePoint)

File metadata
Sharing information
Activity signals

Data Processing with Synapse

Once data lands in storage, process it with Synapse:

# Synapse Spark notebook
from pyspark.sql import SparkSession
from pyspark.sql.functions import col, explode, to_date

# Read email metadata from blob storage
emails_df = spark.read.json("abfss://m365data@yourstorage.dfs.core.windows.net/emails/")

# Parse the nested JSON structure
parsed_df = emails_df.select(
    col("id"),
    col("subject"),
    col("receivedDateTime"),
    col("sender.emailAddress.address").alias("senderEmail"),
    col("toRecipients").alias("recipients"),
    col("importance")
)

# Calculate email volume by sender
email_volume = parsed_df.groupBy("senderEmail") \
    .count() \
    .orderBy(col("count").desc())

# Analyze patterns by time
daily_volume = parsed_df \
    .withColumn("date", to_date("receivedDateTime")) \
    .groupBy("date") \
    .count() \
    .orderBy("date")

# Save to Synapse SQL pool for BI
email_volume.write \
    .format("com.databricks.spark.sqldw") \
    .option("url", "jdbc:sqlserver://synapse.sql.azuresynapse.net:1433") \
    .option("dbtable", "analytics.email_volume") \
    .option("tempDir", "abfss://temp@storage.dfs.core.windows.net/tempdata") \
    .mode("overwrite") \
    .save()

Privacy and Compliance

Graph Data Connect includes privacy controls:

Admin Consent Every pipeline requires approval from a Microsoft 365 admin. They see exactly what data is being requested and can approve or deny.

User Scoping You can limit data extraction to specific groups:

{
  "userScopeFilterUri": "https://graph.microsoft.com/v1.0/groups/{group-id}/members"
}

Data Columns You can select only the columns you need, minimizing data exposure:

{
  "outputColumns": [
    {"name": "id"},
    {"name": "subject"},
    {"name": "receivedDateTime"},
    {"name": "importance"}
  ]
}

Pseudonymization Optionally hash user identifiers:

{
  "allowedGroups": ["analytics-team-group-id"],
  "userEmailObfuscation": "hashUserEmail"
}

Use Cases

Workplace Analytics Alternative Build custom analytics dashboards showing collaboration patterns:

Meeting time per team
Email volume trends
Cross-team collaboration

Compliance Monitoring Extract communication metadata for compliance reviews:

External communication patterns
Sensitive content flags
Policy violation detection

Migration Planning Understand usage patterns before migrations:

Active vs inactive mailboxes
Storage consumption
User activity patterns

Cost Considerations

Graph Data Connect charges per record extracted:

Messages: $0.000025 per message
Calendar events: $0.000025 per event
Files: $0.00001 per file

For a 10,000 user organization with 1M messages per month, expect around $25/month for message extraction alone.

Add Azure Data Factory and storage costs on top.

Approval Workflow

The admin approval flow is critical:

Pipeline triggers
Request appears in Microsoft 365 admin center
Admin reviews requested data scope
Admin approves or denies
If approved, data extraction begins

# Check pending requests via PowerShell
Connect-ExchangeOnline
Get-DataPolicyOperationAuditInfo -OperationType ApproveRequest

Incremental Extraction

For ongoing analytics, implement incremental loads:

{
  "typeProperties": {
    "source": {
      "type": "Office365Source",
      "dateFilterColumn": "receivedDateTime",
      "startTime": {
        "value": "@pipeline().parameters.lastRunTime",
        "type": "Expression"
      },
      "endTime": {
        "value": "@utcnow()",
        "type": "Expression"
      }
    }
  }
}

Track the last successful run time and use it for the next extraction.

Getting Started

Request the Graph Data Connect preview access
Create an Azure AD app with appropriate permissions
Set up Azure Data Factory and storage
Work with your Microsoft 365 admin on approval workflows
Start with a small pilot group before scaling

Graph Data Connect bridges the gap between Microsoft 365’s rich data and enterprise analytics platforms. It’s a powerful tool when you need insights across your organization.