Analyzing big-data can yield unprecedented information about customers. It can also create more problems than you can solve. That's what I witnessed during a recent encounter with my longtime personal bank.
The incident began bright and early Sunday morning at 8:18 a.m., when I received a call from a number and area code I did not recognize. I missed the call, and the caller left no message, so I was left doing detective work to figure out what was going on.
Looking up the area code, I discovered it was from my time zone and state -- it came from Chicago. What's more, when I conducted a search on the phone number, I turned up a user forum with postings by people who were contacted by this number. Most of the posters believed it to be a contractor for my bank (not the bank itself, it appeared). People were complaining about the contractor asking for Social Security numbers over the phone during a supposed fraud investigation.
At this point, I decided to play the waiting game. If it was something important, the bank would contact me again. Besides, the stories on the forum sounded suspiciously like a phone fraud attempt. After all, anyone could claim to be calling on behalf of my bank.
Then on Monday, I was using my bank credit card to swipe into a city parking lot when the reader gave me a strange error. Odd. I switched to my debit card and moved on. I used my credit card later at Starbucks and received no error. It was not until I was shopping on LivingSocial the next day -- and experienced more errors -- that I realized something was amiss.
I went to a local branch of my bank and met with a banking specialist. The bank contacted the fraud department. Lo and behold, the Sunday call at 8:18 a.m. from the shadowy contractor was indeed placed on behalf of my bank. What's more, the bank had put a block on my credit card in suspicion of potential fraud.
It turned out that the bank was very suspicious of a payment I'd made to a certain Website. Brace yourself for the ironic part: This was the site for one of the nation's largest professional engineering organizations. What's more, I had placed an almost identical payment -- association dues and a conference registration fee -- on my debit card almost precisely a year earlier.
The bank rep admitted that the bank could only look back at activity within the last three months, and it never bothered to visit the "suspicious" Website to see what it was before locking my card.
At this point, I did my best to maintain composure. I recognized this as a big-data nightmare. If the bank only had proper algorithms for analyzing its input, it would have tied my debit card transaction from last year to my credit card transaction this year and recognized it as familiar. Alternatively, an algorithm could have recognized the engineering organization and found that I received payments related to that profession on a regular basis.
But my bank does not have such good algorithms. Its glaringly poor ones produce false positives (flagging a yearly payment to a professional society as fraud) and wreak havoc on its business and clients (by locking their cards).
It appears to me that my bank has a big-data problem -- too many data points and not enough sophistication on the algorithm front.
There's a time-honored saying that the first step in fixing a problem is admitting you have one. Well, some companies need to admit they have a big-data problem.
When it's clear you have a lot of data (a big opportunity) and lack the programming competence to create algorithms that can handle that data well, there is only one solution as I see it -- seek outside help.
Failure to seek expert guidance and outside algorithm experience is a mistake that will hurt your clients and hurt your earnings. I suggest you avoid that mistake.
— Jason Mick is senior news editor at the independent tech news site DailyTech.