Fraud detection

[drewr]

Hey all! Added an enhancement issue because we at Datum need this functionality pretty soon. But, would love to have a more free-form conversation around the best practices everyone does to flag users as potentially fraudulent.

datum-cloud/enhancements#505

[jacobsmith928]

I think the best way to think about this is in two ways:

• Upfront and ongoing scoring (various integrations should be supported)
• Workflows (e.g. what to do when something changes)

I think that Milo should include an "out of the box" capability that includes three things:

1. Baseline fraud scoring
2. Baseline user enrichment
3. Abuse handling workflow (specific to computing or email providers)

To start, Milo admin users could input their MaxMind license key (or Stripe Radar) as well as a BetterContact or Clay key for enrichment. This would need to pair with a basic fraud and risk workflow that restricts users based upon their score as it changes, and notifies Milo users via Slack or a helpdesk webhook.

A next level up would likely be about integrating with two higher level 3rd party services for more a more robust and durable stance:

• a more robust fraud detection system, such as Sift, Telesign, or Signifyd.
• a KYC (know your customer) service, which might include automatic identity verification