ID check process

Real ID uses the latest computer vision and A.I. technologies to quickly verify and extract details from ID documents like Driver's Licenses, Passports and Residence Cards.

It can be set up to verify IDs before checkout, after checkout or even before being allowed to view your store. Additionally, you can have it set up to only trigger ID verification on specific high risk triggers like total order value or Shopify Fraud Analysis flags.

You also can require additional documentation like a corresponding live headshot photo, Proof of Address, and eSignature.

The entire process can be completed in as little as a few minutes, and it's entirely automated. But staff can override the automatic results at any time.

How it works

The way a customer is presented with their ID check depends on which flow you choose. But the process is roughly the same for each.

First, the customer opens their unique ID check associated with their account. Then they're guided through capturing the photos required according to your rules.

Computer vision techniques during the process help guide the customer to take the best photo possible for the best results. Then, the customer will be either automatically approved, or placed into review.

tip

These are not the exhaustive list of methods Real ID uses to verify identity, but a general overview of the various technologies and techniques in place.

Autocapturing ID photos and headshots

Real ID uses A.I. during the capturing of ID photos and headshots to help guide the user to take the best possible photo to optimize for their chance of passing ID verification.

This A.I. running on the customer's phone during the ID verification process helps guide them to place their ID within the camera frame at the correct position.

It will help guide customers by giving them feedback to show the ID, move it closer or further way from the camera for focus.

This helps vastly improve photo quality and helps guide customers to take the best possible photo of their ID

A similar A.I. model is used to help guide customers to take a headshot, if required by your rules.

note

The ID autocapture will allow customers to manually capture a photo of their ID after a certain period of time.

This is to help customers in extraordinary circumstances still submit photos of their IDs. Then you'll be able to take the final judgement and approve or reject these submissions manually.

The Real ID dashboard will show you if a submission was manually captured instead of automatically captured.

How is an ID accepted or rejected?

During verification, an ID is accepted when all required fields such as name, address, date of birth, and date of expiration can be read with high confidence and meet your additional rules such as cross checking order details, biometric verification, and age verification.

Real ID uses multiple A.I. models to determine the validity and liveness of the photos as well as models to extract the fields from the ID into a standard format.

But this confidence is customizable, you can lower or raise the threshold of confidence required for an accepted ID.

Additionally, you can manually override the results at any time, even if the system is unable to automatically verify it.

Hard rules

Some rules don't rely on confidence alone, such as detecting expired ID documents. Adjusting the confidence threshold for your ID checks won't affect the expiration date checks for example.

Maximum soft retries

During the ID check process, customers are given 3 attempts per stage to take a high quality, readable and usable photo.

For example, if the ID check requires both the ID photo and a portrait photo to face match against, then customers are given 3 tries to take a good quality photo of their ID as well as 3 photos of their face.

These "soft" retries do not incur extra charges. Customers can use all 3 attempts per stage, and you'll only be charged once for the entire check.

If customers fail all soft retries in a particular stage, then they'll allowed to move onto the next stage.

This allows customers to still continue through the flow in case they have a document that isn't supported like a temporary issued Drivers License, or damaged ID, and gives you a chance to override the failed ID checks.

If you still want to allow customers to try again with a "hard" retry, where customers can try again with a new fresh ID check, please see this article.

Allowed captured methods

By default Real ID allows customers to both capture photos using their device's camera only.

This is the most secure option, since it requires customers to take a live photo of their ID document, as well as take a live selfie photo.

However you can enable manual photo uploading. This allows customers to upload photos from their camera roll if they don't have their ID phyiscally available because it's lost or stolen.

However, manual file uploads bypass the ID and headshot photo Autocapture feature, which may lead to bad quality photos or a higher chance of failing the ID check.

Additionally, disabling the manual file uploads helps increase security because it helps narrow down the possibility of a bad actor using a stolen images.

tip

The Real ID dashboard will show if photos from the ID check were uploaded via a file or captured from the user's camera.

In general, photos captured in real time from the camera are more trustworthy than file uploads.

Face Match

If you enable the headshot requirement, Real ID will also verify the headshot matches the headshot physically printed on the ID document.

You'll be able to see the similiarity confidence of the captured headshot against the headshot on the ID document from 0% to 100% within the ID check details in the dashboard.

If the ID check passes, yet the headshot cannot be confidently matched, then the entire ID check is rejected.

tip

Even if a ID check is rejected to a mismatching photo, you'll be able to override the results in the dashboard.

Text Extraction

Real ID uses modern computer vision techniques to read text fields on the verified ID photo.

Real ID can read text fields from Driver's Licenses, Passports, etc like:

• First Name
• Middle Name
• Last Name
• Document Number
• Address
• Date of Birth
• Date of Expiration
• Endorsements
• Classes

These fields are especially useful for applying a minimum age verification to your products, or rejecting expired ID documents, or verifying that the shipping or payment details match up with the customer's ID.

note

Some fields like the First Name and Last Name fields are required in order for the ID check to pass automatically.

If an ID check fails because one or more required fields are covered, missing or are inlegible, then you'll see that reason within the ID check details in the dashboard.

Completed

When the customer submits their required information, their photos will undergo additional processing with A.I. models running in Real ID's private cloud.

Then Real ID will use the combination verdicts from these models and the results of the enabled additional rules to determine if the ID check passes or fails.

If the customer fails, they're shown feedback that their ID wasn't able to be automatically passed.

If the customer passes, they're shown a call to action to return to the order status page, or return to the cart to checkout, or to notify them they can browse and checkout within your store. Each depends on which ID verification flow you've choosen.

tip

By default Real ID does not cancel or modify the order in any way, even if the ID check failed.

Real ID will update metadata, tags and metafields on the orders with ID checks - regardless if the ID check passed or failed.