Given the Fortune 500 company’s friendly return policy, some scammers attempt to exploit it.
The absence of advanced fraud analysis allowed scammers to return empty boxes and claim that the order was not delivered or that the product was damaged. To protect themselves from being blocked by the retailer, they create multiple accounts; both the account and orders commonly share a detail, e.g., phone, email or address.
Using GraphAware Hume, we created a knowledge graph with all the user accounts and order details automatically ingested; we implemented advanced analytics to increase coverage and set up alerts that inform us about emerging risks.
I feel like a caveman who discovered fire.”
— Data Science Lead
The client initially approached GraphAware seeking consultancy and help with the development of an internal application for account association linkages. We introduced our ready-to-use solution Hume instead, considering the amount of labour and costs of developing a completely new graph application. Using Hume enabled the retailer to focus on creating business value and fine-tune graph data science-based algorithms. The soft production was deployed in 3 months, and within 6 months it exceeded the scam detection KPIs by 300%.