Friday, August 09, 2013

Was I A Victim of Big Data?

On a recent trip to Phoenix my flight was delayed then cancelled.  Through rerouting and other delays what should have been a 1.5 hour direct flight turned into a 12 hour ordeal.

I wonder was I a victim of Big Data?  Let me explain.

As a Big Data marketeer one of my favorite stories of how Big Data can force changes in the way you do business and cause you to call into question recommendations that are not intuitive is the following hypothetical dilemma.
You are the flight operations manager of an airline.  You have 2 planes about to depart.  It's snowing hard.  The airport calls down to inform you that only 1 of your 2 departures will be granted permission  to depart before the airport will shut down.  One plane has 4 passengers on board, the other is full with 200 passengers.  
What do you do?
You run your new "Flight Operations Optimizer" application.  It's a new "Big Data" application that calculates the down stream 72 hour impact of canceling a flight based on multiple data sets including all passengers impacted, expected weather delays at all downstream destinations, airplane maintenance schedules and crew schedules.
The Flight Operations Optimizer comes back and advises that you let the flight with 4 passengers depart.  That alternative has the least down stream impact to the airline - A better business outcome.
To which you say WHAT! That can't be right.  If I do that I'll have 200 pissed of passengers at the counter to deal with and furthermore I won't make my goal of - most passenger on time departures!
The story points out 2 major elements of Big Data.

  1. If you can analyze enough data often kept in different silos the results can well be counter intuitive.  Leading you to dismiss it and still take the decision that doesn't lead to the best business outcome.
  2. The employee goals you have in place to drive the best corporate outcomes might be driving behaviors that don't actually don't accomplish that.
Now back to my bad travel karma story.

Moments before boarding we were told our flight is now departing from another gate.  Quickly we hurried over to the new gate only to be told our flight was delayed 4 hours.  What happend?

From eavesdropping on the gate agents conversation I learned that the plane for the flight to LA had a mechanical and that our plane was allocated to the LA flight. Our flight was delayed while they looked for another plane.  2 hours later our flight was canceled. 

Could it be that the airline flight operations manager ran the "Flight Operations Optimizer" application the result being that giving our plane to the LA flight had the least down stream impact?

Then after most of the passengers on our flight found alternatives to waiting 4 hours there were very few left on our delayed Phoenix flight.  So the "Flight Operations Optimizer" application advised it be cancelled.

And that is how I suspect I fell victim to Big Data.  Best business outcomes don't always mean best personal outcomes.