Skip to content
Create account
or
Sign in
The Stripe Docs logo
/
Ask AI
Create account
Sign in
Get started
Payments
Revenue
Platforms and marketplaces
Money management
Developer resources
Overview
About Stripe payments
Upgrade your integration
Payments analytics
Online payments
OverviewFind your use caseManaged Payments
Use Payment Links
Build a checkout page
Build an advanced integration
Build an in-app integration
Payment methods
Add payment methods
Manage payment methods
Faster checkout with Link
Payment interfaces
Payment Links
Checkout
Web Elements
In-app Elements
Payment scenarios
Handle multiple currencies
Custom payment flows
Flexible acquiring
Orchestration
In-person payments
Terminal
Beyond payments
Incorporate your company
Crypto
Financial Connections
Climate
Understand fraud
Radar fraud protection
Manage disputes
Verify identities
    Overview
    Get started
    Verify identity documents
    Handle verification outcomes
    Access verification results
    Review verification results
      Insights
    Verification flows
    Verification checks
    About the APIs
    Verification Sessions
    Go live
    Before going live
    Supported use cases
    Explaining Identity
HomePaymentsVerify identitiesReview verification results

Insights

Learn how to identify risks and understand signals from verification checks.

Stripe Identity’s AI model considers a variety of signals when verifying a user’s identity. It examines a number of factors to produce insights that can give further clarity into Stripe’s decision. These insights are more nuanced than the top-level verification decisions, and you can use them to assist with manual reviews or customer support processes.

See the following definitions:

  • Insights: The name Stripe uses to refer to the collection of all insights.
  • Insight: The specific attribute scored (for example, blur, authenticity). It’s of either type Level or Label.
    • Level: These insights provide a computed level, which is a score that translates to low, elevated, or high. Insights of this type evaluate the potential risk to verification.
    • Label: These insights provide a binary value of being either present or absent. Some of these insights evaluate a potential risk to verification, but others might be neutral, with no inherent risk associated with them.
  • Blocking: This indicates that verifications with a risk level classified as “high” for this insight are rejected as “unverified”.

Document insights

These are the insights produced on document checks:

NameTypeDescriptionBlocking
Activity on Stripe networkLabelIndicates whether the identity presented fails to match a Stripe-known good identity. While it’s common for an identity to be unknown, it does represent a slightly higher risk than a known identity.No
Barcode or machine-readable zoneLevelIndicates if the machine-readable zone or barcode is present and decoded, but was manipulated.No
Document authenticityLevelIndicates whether the document presented might have been digitally manipulated or is otherwise inauthentic.Yes
FraudLevelThis comprehensive score evaluates various signals related to document verification, including signals from Stripe Network data.Yes
Image blurrinessLevelIndicates the severity of the blurriness of the document image presented during the verification process.No
Physical document detectionLevelIndicates whether the document presented is a screenshot of a document.Yes
Risky behavior on the Stripe networkLabelIndicates whether the identity presented matches a Stripe-known identity with a history of high risk activity.No

Selfie insights

These are the insights produced on selfie checks:

NameTypeDescriptionBlocking
Duplicate selfieLabelIndicates whether the selfie presented matches a selfie from a previous identity verification.No
Duplicate selfie with a data mismatchLabelIndicates that the selfie presented matches a selfie from a previous identity verification, but the extracted document data from each doesn’t match.Yes*
Selfie and document matchLevelIndicates whether the selfie isn’t a high confidence match with the face on the document.Yes
Selfie livenessLevelIndicates that the selfie presented might not have been a live person in front of the camera, such as a photograph or screen presentation attack.Yes

* If Synthetic identity protection is enabled on your account, then the “Duplicate selfie with a data mismatch” insight is blocking.

Dashboard usage

The Dashboard page for a submitted VerificationSession contains a panel showing the insights generated for this session.

Was this page helpful?
YesNo
  • Need help? Contact Support.
  • Join our early access program.
  • Check out our changelog.
  • Questions? Contact Sales.
  • LLM? Read llms.txt.
  • Powered by Markdoc