Written by Yona Ayala
August 13, 2018
Category: Risk Models, Digital Data
Modernize Your Risk Model In Today’s Digital World
You’re on the car dealership’s website, and you’re putting your information into a credit application. After which a bank or a lender runs multiple risk models to make sure they’re extending credit to someone who can pay back the loan with interest and on time. Without timely efficient and appropriate risk models in place, businesses would be vulnerable to fraud (often, running into millions of dollars) and at risk for eventual bankruptcy.
Risk models like these are used every day, on every website you touch. Although retail and financial sectors have a glaring need, these models are behind the scenes on websites or apps, not visible to the naked eye. Examples may include situations where ID verification, profile creation, or even crowdsourcing content review and approval are used. There’s a lot of these we typically touch daily.
In this webinar, JK Venkatesh, our Head of Digital Analytics and Data Science takes us deep into how digital data elements can be used to enhance risk models with little to no investment but a high tangible return on investments (ROI). You will not only gain in-depth knowledge on what goes into a risk model but how your business can benefit from it, as well. Are you ready to cut down on your losses and curb fraud risk? Be sure to watch the whole webinar to get some great insights on today’s modern risk models.
10:47 – The 5 Fundamental Steps of Risk Models
You will be introduced to how risk models are created from the bottom up including critical steps of a proper risk model and how it becomes effective in curbing fraud/risk.
11:54 – Exploration
Every business starts somewhere with risks models, and there are two scenarios in which your business could be in. Each has its own important factors as to how digital data is being used.
14:45 – Assessment
In terms of data, there are many digital data points to tap into. There are several examples JK talks about, some of which are seasonality, geographic location, and customer transaction times.
17:33 – Variable Selection
JK goes one step deeper into how the variables considered in the first step actually get used and explored in the modeling exercise.
20:55 – Finalization
You’ll get to build and test the risk model. JK highlights that you’ll want to have more than one model, just in case things don’t work out the way you wanted it to.
21:25 – Validation & Scoring
Risk models are like cars, in that they need to go through a crash test, right out of the factory. JK highlights that we want the digital data to be complementary to the other data points (including offline) and stresses on the importance of stress testing the model before it gets put into use during production.
30:30 – Case Study #1 (Financial Service)
Money remittance companies are the most exposed when it comes to fraud. In this case study, JK covers the three most important digital elements he used to curb fraudulent transactions that wouldn’t have been otherwise identified with non-digital indicators.
38:15 – Case Study #2 (Non-Financial Service)This case study stresses the fact that risk models are not just for financial sector. In this case study, JK talks about a spa consolidator (think of the Expedia of spas) that used digital risk models to flag suspicious reviews and ensure the validity of the reviews posted.