The challenge
A gaming and betting operator decides at sign-up who gets an account, often in many markets at once, and new players arrive with little more than a phone number and an email. This operator had no fraud history to learn from: not a single fraud label. So the question was what a ready-built global fraud model could see on its own: how new players would split into approve, review and reject, which signals push a player into the risky pile, and whether risk is spread evenly or gathers in a few markets.
The solution
Monnai checked every new player in real time with its global fraud model, from the phone, the email, the digital footprint and activity across its network, and gave each one a clear risk score. Every score comes with a band, approve, review or reject, and up to five reasons in plain words: the reviewer’s main tool for the middle band. The type of phone was the strongest signal, then the age of the email. Players with little or no shopping or social presence online scored riskiest.
The impact
9 in 10 players were recommended for automatic approval, so they start playing right away, and about 4 in 100 went to review: a small queue for the team. Risk was far from even. In two European markets, a quarter to a third of sign-ups fell in the reject band, against under 2% in most others, so Monnai suggested a dedicated review queue for each. Fake sign-ups stood out: inboxes anyone can open without registering, and one phone number behind a thousand transactions in a single month.
