Data Danger

I am reading more and more articles that are giddily proclaiming the power of big data.  There was a recent article that highlighted its use in the employee screening and hiring process.  If you set aside the potential legal issues there are more disconcerting elements associated with the power of data.

Romancing the data:

 Don’t get me wrong.  I have worked around a lot of data for a long time and doing so is essential to being successful at managing bad debt or preventing fraud.  There are limits however, to the power of data.   No matter how much data we gather and no matter how long we analyze it there is an asymptotic limit to what it can predict.  Anyone who has spent any time in credit, collections or in the prevention of fraud (or is a parent for that matter) can tell you that predicting human behavior is difficult.  It is tempting to assume that a system designed to manage data can actually solve all of your business issues.  The reality is that data and the systems around it can only present data for you to understand and make decisions on.

Correlation is not Causation:

When I was leading a very large telecom A/R organization we were trying to understand our risk profile.  We had enough customers of our own to do a statistical analysis and we found some interesting and counterintuitive things.   Like most companies at the time, we depended on a FICO score to determine the risk of a potential customer.  It takes into account a number of things but bankruptcy was a major negative factor in the compilation of the score.   It was a very accurate predictor of someone’s likelihood to default on their mortgage; we assumed that if they were going to default on their mortgage, they would stop paying us long before they went bankrupt.  What we found after using a sample of our 10 million customers was that for wireless service, bankruptcy was not as predictive as one would think.  We discovered a large number of entrepreneurs who had declared bankruptcy (some more than once) who were the most reliable customers in the consumer and small business markets.  Turns out, they may let other things go unpaid but not their cell phone.   This allowed us to exploit a market opportunity that none of our competitors would not go near.    We also discovered that there were other nontraditional measures that proved very accurate in predicting risk.  The point is that the correlations are not necessarily causes, and that predictive indicators can and do change.   Which leads to the next danger.

The “Wall-E” effect:

Some times the things that data and systems can do is so powerful that it can make us  passive as if the data or the system itself could make all of the decisions for us.   There is a scene in the movie “Wall-E” where the humans are so dependent on technology that they have become functionally helpless.   Instead of using the technology as a tool to achieve an end, the technology becomes an end  in itself and they are completely incapable of functioning without it.

Data can have the same effect on managers.   They fall into a kind of unhealthy dependence on the data. They cease to understand that data and systems are tools, not ends in themselves.  Data output is the starting point for problem resolution  not the end.

0 Responses to “Data Danger”

  1. Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Follow me on Twitter

Dan Clark

Dan Clark

Principal of Bowline Consulting, process designer/fixer, wireless telecom veteran, addicted pick up soccer player, fly fisher, backpacker, beer brewer, guitar player, choir singer, recovering bag piper

View Full Profile →

%d bloggers like this: