The challenge
A merchant platform has two moments to get right: letting a merchant in, and keeping that merchant’s account in the right hands afterwards. Fake and risky merchants have to stay off the platform, and hijacked accounts have to be spotted fast. Both threats are rare, which is what makes them hard to find: almost every merchant is legitimate. The question was whether two scores could pick out the few risky cases and send everyone else down the easy path.
The solution
Monnai gave every merchant two scores from its email, phone and business name: one for onboarding risk, from automatic approval up to manual review, and one for account takeover, from standard monitoring up to an immediate lockdown. Risky merchants showed up on the web: parked websites, domains with little trust, gibberish email addresses. Takeovers left another trail: the merchant’s presence across platforms, its review counts, and phone numbers recently moved to another operator. Results were checked on the platform’s own outcomes, including a later period the model had not seen.
The impact
A cut-off that flags fewer than 1 in 100 merchants caught 88% of the confirmed risky ones before they went live, while 98% of merchants sat in the lowest risk tier and were approved automatically. Most merchants get the fastest path to selling, and the team’s attention goes to a small pile. Takeovers were harder to call, but in the top takeover band, 86% of accounts were confirmed takeovers.
