Identity Verification


Identity Verification (IDV) is the primary job type in Vouched. It lets you compare a photo of a user's ID document with a selfie photo to confirm they are both present and who they say they are.

User Confirmation of Photos

You can allow users to confirm the quality of their ID photo and selfie photo before it is submitted. Clicking Confirm, will move the user to the next step. Clicking Retry, will allow the user to retake the photo. The goal of this feature is to reduce the amount of unsuccessful verifications by increasing the quality of photos submitted.


Example of User Confirmation

To enable this feature in the JS Plugin with avant theme, set the userConfirmation property confirmImages to true in your Javascript code.

Confidence Scores, Errors, and Signals

The errors provided in a job are an indicator of the failure of the job. Errors fall under various
categories. For the id and selfie categories, a field called signal is available. These provide further insights into why a specific job may have failed. While errors provide information on the error code leading to the failure, signals provide detailed information on what lead to the error. The signals also help to visualize the potential sources for penalties to a specific category score.

Fraud Checks

Vouched performs multiple fraud checks to help ensure the ID document is real, including:

  • Visual inconsistencies against known ID properties (font, color) as well as textual and data inconsistencies against known ID properties
  • Use file information to check for fraud, including the use of metadata if available.
  • Check against known fraudulent ID properties
  • OCR-based text extraction

False Positives and False Negatives

A false positive occurs when a classification is labeled as true, incorrectly.
A false negative is when a classification is labeled as false, incorrectly.

In terms of conversion, the classification matters. Consider the following scenarios:

Classification: A good user

  • False Positive: A fraudster is labeled as good (positive)
  • False Negative: A legitimate user is labeled as bad (negative)

Classification: A fraudster

  • False Positive: A legitimate user (positive) is labeled as a fraudster
  • False Negative: A fraudster (negative) is labeled as a good person