Duplicate detection
Updated on 16.01.26
4 minutes to read
Copy link
Overview
Duplicate detection is an automated fraud prevention feature that identifies repeat applicants and potential multi-accounting across all your verification sessions. It works by comparing key data points from each new session against your entire history of verifications.
This check is designed to prevent fraud by catching users who try to sign up multiple times, often for bonus abuse or to circumvent previous restrictions. By analyzing unique identifiers like face embeddings and document IDs alongside SEON’s 900+ unique fraud signals, you can quickly determine not only if a user is a duplicate, but also the intent behind their repeated attempts.
How it works
Duplicate detection is an automated process that runs silently in the background of every identity verification session.
1. Data collection: During any verification session (IDV, Selfie, etc.), key data points are collected and stored, including face embeddings, document numbers and personal information.
2. Automated cross-comparison: For each new session, the system automatically compares the new data points against the historical data of all previous sessions.
3. Duplicate flagging: If a match is found based on the defined comparison methods (e.g., face vector distance, exact document ID match), the session is flagged as a duplicate.
4. Unified results: The duplicate flag is included in the unified verification results, delivered via webhook and clearly visible in SEON, along with a link to the session(s) it duplicates.
Data points compared
SEON uses a combination of biometric and data-based checks to identify duplicate users.
| Data point | Comparison method |
| Face embedding | Compares the biometric vector of the user's face against all previous face embeddings. A match is flagged if the vector distance is below a set threshold. |
| Document ID | Performs an exact string match on the document number (e.g., passport number, driver's license number). |
| Date of birth | Performs an exact match on the date of birth, but only flags a duplicate if it is corroborated by another matching data point (e.g., name or face). |
| Full name | Performs a case-sensitive exact match on the user's full name, but only flags a duplicate if it is corroborated by another matching data point. |
| Email address | Performs a normalized match on the email address (converted to lowercase, with dots and aliases removed). |
Use cases
- Preventing bonus abuse: Stop users from signing up for multiple accounts to take advantage of promotional offers, free trials or sign-up bonuses.
- Blocking serial fraudsters: Identify and block bad actors who repeatedly attempt to create accounts after being previously declined or banned.
- Enforcing single-account policies: Ensure that each user maintains only one account on your platform as per your terms of service.
- Investigating collusion rings: Use duplicate detection signals to uncover networks of fraudulent users who may be sharing documents or identities.
Integration
Duplicate detection results are seamlessly integrated into your workflow.
- Admin panel: Any session with a duplicate will be clearly flagged in SEON, along with a direct link to all matvhing previous sessions.
- Webhooks: The duplicate check result is included in the
idv:session_finishedwebhook payload, allowing you to automate actions such as sending the case for manual review or automatically declining the user.
