Predictive Analytics at Work
“Sorry” said the friendly barely-old-enough-to-have-a job young woman behind the counter as she handed my credit card back to me “Your card did not work.”
“No need to apologize” I sheepishly replied. “I blame my bank’s predictive modelers, not you.”
“That’s very sweet, because most people blame me. Do you have another card?”
I had just arrived in Vegas, and stopped at the Walgreens next to my hotel to stock up on Diet Mountain Dew. Since I had just successfully used that same credit card before jumping on the plane to Vegas, I deduced that my bank was doing its best to prevent fraudulent use of my credit card.
And in a sense, it was a success. Sure, I was mildly annoyed, embarrassed, and inconvenienced that my card was not approved. But mathematically speaking, I had to tip my cap. The predictive modelers at the bank had good reason to put up the red flags. They were playing the odds that I would not actually be using the same credit card shortly after having used it half way across the country.
We are not in the fraud-detection business here at B2E, but we do a lot of predictive analytics that benefit from behavioral patterns by large groups of people. We remind our clients often that there is tremendous value in playing those percentages when finding audiences for their marketing campaigns.
Embracing the fact that not every prediction is going to be right goes a long way to reaping the benefits of predictive modeling, because – mathematically speaking – simply putting the odds in your favor enough times can dramatically save costs and drive revenue.
Let us know if you think your company might benefit from doing some predictive analytics for your next campaign.
Peace, Greg
