It Is Not Real Progress Until It Is Operational

When I think of operationalizing analytics for fraud management, I am reminded of this saying: “Knowledge is knowing that tomato is a fruit; wisdom is knowing not to put it in fruit salad.” Knowledge of the results of a fraud model—even very thorough knowledge—does not guarantee the best results in the production use of the model. Very careful thought must go into deciding how the model will be used in production and what actions need to be taken to achieve the desired customer service and fraud control results. We can have the best data, the best methods to extract information to use in a fraud model, and the best analytical techniques to detect fraud, but if the performance metrics set for the model or the way in which the fraud score is used in production is not aligned with how business objectives are set, the fraud management system is unlikely to be very useful in managing fraud.

For fraud detection, operationalizing fraud analytics is as important as using a data-driven analytical system. Any gap in the collaboration among the operations, business, IT, and analytics teams can have undesired outcomes. This chapter will examine the specific challenges involved in operationalizing fraud solutions and how best to deal with these challenges. The teams each play a key role in this, and no single team's role is any less crucial than the roles of the other teams.

It is important for this partnership among operations, business, ...

Get Bank Fraud: Using Technology to Combat Losses now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.