Data Protection Issues

How to Configure Custom Sensitive Information Types for SSN in Microsoft 365

Natalie TaylorNatalie Taylor Nov 2, 2026 870 views

Question details

The user needs to optimize a custom sensitive information type to accurately detect Social Security numbers without triggering excessive false positives.

How to Configure Custom Sensitive Information Types for Social Security Numbers
Product
Microsoft Purview / Office 365
Device & OS
not provided
Scenario
Configuring Data Loss Prevention (DLP) rules to detect custom or non-standard formats of U.S. Social Security Numbers (SSN) in documents and emails.
Observed behavior
The built-in detector misses certain SSN formats. Increasing the proximity setting captures more matches but can cause high false positive rates and processing overhead.
Before you start

Ensure you have the necessary administrative privileges in the Microsoft Purview compliance portal to edit or create sensitive information types.

Solution 1Recommended

Optimize Proximity and Confidence Level Settings

Apply Microsoft's recommended limits to your proximity settings and adjust the confidence level to balance detection accuracy with system performance.

While testing might show that increasing the proximity limit beyond the default catches more matches, it also significantly increases false positives and server processing requirements. Microsoft officially recommends keeping the maximum proximity at 300 characters.

1
Access Data Classification

Log in to the Microsoft Purview compliance portal and navigate to 'Data classification', then select the 'Sensitive info types' tab.

2
Edit the Custom SSN Type

Locate and select your custom U.S. SSN sensitive information type, then click 'Edit' to modify its matching parameters.

3
Set Proximity Limit

In the pattern configuration, set the proximity parameter (the maximum distance between the SSN pattern and corroborating evidence) to a maximum of 300 characters.

4
Adjust Confidence Level

Configure the rule's confidence level to an appropriate range, commonly recommended between 55% and 85%, depending on how strict you want the detection to be.

5
Require Corroborating Evidence

Ensure the rule is set to require corroborating evidence (such as keywords like 'SSN' or 'Social Security') to trigger a valid match.

Optimize Proximity and Confidence Level Settings
Understanding Proximity Values: Proximity dictates how close corroborating evidence must be to the primary pattern. A limit of 300 characters is usually sufficient to confirm context without overloading the DLP engine.
Free Microsoft Office alternative

Try WPS Office for Secure Local Document Management

While enterprise-wide DLP rules require Microsoft Purview, WPS Office provides a highly secure, free, and lightweight alternative for everyday document editing. It features seamless format compatibility with Microsoft Office and includes robust built-in document encryption to help you protect sensitive data locally.

  1. 1. Download and Install: Download WPS Office for free from the official website and install it on your device.
  2. 2. Open Your Document: Launch WPS Office and open the document containing sensitive information that you want to secure.
  3. 3. Apply Password Protection: Go to the 'Menu', select 'Document Encryption', and set a strong password to restrict access to the file.
100% format compatibility with Microsoft Word, Excel, and PowerPoint files.Built-in document encryption and password protection to manually secure sensitive SSN data.Lightweight installation with a familiar, easy-to-use tabbed interface.Free to download and use for your daily productivity tasks.
microsoft office alternative - wps office

Frequently Asked Questions

What is the maximum recommended proximity for SSN detection in Microsoft 365?

Official Microsoft documentation recommends a maximum proximity value of 300 characters to effectively balance accurate detection with optimal processing performance.

Why does my DLP rule miss some Social Security numbers?

The built-in SSN detector may miss non-standard formats or numbers without standard separators. Creating a custom sensitive information type allows you to define specific regular expressions to catch these variations.

What confidence level should I set for a custom SSN sensitive information type?

A confidence level between 55% and 85% is commonly recommended. This ensures that only matches with sufficient corroborating evidence trigger alerts, significantly reducing the rate of false positives.