Why Static STILL Stinks

Solution CTO, IBM

As promised, we’re going to revisit a topic I introduced awhile ago in "Why Static Stinks". Based on what I’m seeing recently, static still stinks, so now is a good time to resurface our discussion. Collectively, we’re just not moving fast enough to fix the glaring issues that static–otherwise known as non-personalized interactions– represent to having healthy customer relationships.

As in our original scenario, let’s imagine if Netflix never updated your movie preferences, or even worse, never asked what you liked in the first. Or imagine if your spam folder never adapted to the changing inbound spam. Or finally, imagine that your Internet radio station presented a fixed playlist that didn’t bother to ask you if you liked what it was playing.

Still crazy, right? Yep, still crazy. So why are so many firms doing it?

Let me give you an example of 1 firm getting it wrong, as done to me. My debit card was compromised recently, and despite the fact that I had just bought gas in my hometown in California, a charge was authorized from movies at a Redbox kiosk in  Baton Rouge, Louisiana – 3000 miles away – around 15 minutes after I bought gas. It’s not likely that it was me renting those movies. But putting that aside, what happened next is the real issue.

I immediately dutifully called my bank, since having a compromised debit card isn’t a great situation. They proceeded to just about ruin my relationship with the bank. Keep in mind the larger situation – here was an 11+ year customer …

  • calling exactly as they instructed, with a clear-cut case of fraud as evidenced by my spending patterns in their transaction record
  • who has never so much as made a late payment on multiple consumer and real estate loans
  • who has never disputed any charges in 11+ years
  • calling to report fraud on a $5 charge.

What was the likelihood I was trying to defraud the bank of that $5? Yep, pretty low. What followed was I received the third degree, clearly being driven from a static script because you could hear the agent reading it. This inquisition didn’t take into account any of this historical context. It took 10 minutes of answering questions, running the bank’s air time and personnel cost mind you, well over that $5.

The impact on my feelings toward the bank? You guessed it–not too positive right now. The glow of competent service received prior to this experience is gone. The liklihood of me responding to the messages of “Tom, we love you as a customer and we’re here for your needs” – just about zero. All of this over a clearly fraudulent charge.

So how could this have gone differently? Well, for starters, the bank could have trapped this transaction for review automatically. Putting that aside, the bank could have easily had a model that said “is this a high risk customer?” or “is this a customer who attempts to skip his responsibilities to us as evidenced by trying to haggle over his responsibilities?” Does the bank have all of this data at hand? Yep. Can this be back-tested – yep.

So why not do it? The most likely causes are inertia and the traditional fully burdened cost of modeling. Big data technologies and agile methodologies remove those barriers, however, so we’re back to why not do it?

So I ask again, why aren’t you doing it?

For more information: