ID Score
ID Score is a probabilistic score (0-1) assigned to all users at the time of onboarding that predicts the likelihood that the identity supplied belongs to the person who is creating the account. 1 means we are 100% certain that the user is who they say they are. 0% means with 100% certainty we can say that the user is not who they say they are.
When ID Score is enabled and users have a high enough ID Score, we will only collect Base Document and SSN to onboard a consumer (individual or joint account). For users with low ID Score, we will collect Government Issued ID and Video Auth as well.
At the beginning of the user onboarding process, nearly every end-user submits the following data points:
- Name
- Physical address
- Email address
- Phone number
- IP address
- Date of Birth (DOB)
We then evaluate this submitted data. For example, we will place additional scrutiny (including restrictions) on IP addresses associated with:
- Known web crawlers
- VPN proxies
- Anonymity networks (e.g. Tor)
Sometimes our model will need to account for user-submitted information that may not necessarily be an indicator of fraud. To delve deeper into a location example, a user may submit a phone number with a 415 area code (i.e. associated with the San Francisco Bay Area) while listing a current address in Los Angeles--and our model may be able to disregard the location mismatch if we can find evidence of previous San Francisco residency in their address history and we can independently verify that the user presently resides in Los Angeles. We refer to such 2nd-order logic as “derived features” of ID Score.
Three main factors are considered for determining which vendors are most useful for ID Score:
- Decisioning Tools: Ensuring that we are aligned on how the vendor determines what data to pass on to use and how they make assessments about trustworthiness.
- Coverage: Ensuring that there are no gaps or redundancies in the data.
- Sources: Ensuring that certain groups are not underrepresented or overrepresented.
After receiving the augmented data from other identity verification vendors, we use machine learning models to compare against previously-seen data patterns.
ID Score is built to be conservative with all Synapse clients in mind. Each client has the ability to adjust the score application according to their business profiles and risk preferences.
For each user, ID Score assigns a trust level (
low,med or high) and an id score (0-1) after evaluating the authenticity of user input details.While KYC programs vary platform by platform, our general guidance is the follow the following though process:
trust_level | KYC requirements | Example KYC |
low | Perform EDD by requiring additional documents and maybe even a manual review. | SSN + Govt ID + Video Auth + Manual Review |
med | Ask for additional documentation, but still an automated review. | SSN + Govt ID |
high | Ask for minimum KYC information. | SSN |
Note: A low score does not necessarily mean a user is providing false information or committing fraud, it means that we don’t have enough positive signals to assure their identity.
Step 3: Bases on Trust Level, upload EDD Docs or just standard docs.
[1] Trust Level is located under the documents object of the user document (
trust_level). If you are opening accounts for business or joint account users, please note that each base document has an ID Score value and Trust Level associated with it.After ID Score has been added in your MSA (Master Services Agreement), we will follow the following steps:
- 1.We will analyze your platform’s activity and make KYC Recommendations for all three trust score values.
- 2.Once we agree on a KYC workflow, we will program that for you in your Sandbox environment.
- 3.You will then, Integrate with our API for ID Score.
- 4.Test both EDD and non-EDD flows for ID Score in Sandbox.
- 5.Once testing completes, we will enable the new workflows for you on production.
Last modified 2yr ago