"Our beer choice says WHAT?"
You've heard it a thousand times: Correlation is not causation.
Just because I wrote a love letter to Jessica Alba, and then nine months later she gave birth, it doesn't mean that my writing is some sort of magical celebrity aphrodisiac (although that would help to explain the Helen Mirren pregnancy).
But here's the trouble: We're now living in the age of Big Data. And in the age of Big Data, there's going to be a LOT more correlation -- bucket-loads -- while the amount of causation in the universe is likely to remain the same.
So how do we -- as a species -- armed with our new gigantic data sets, combat the well-documented tendency to fabricate meaning from insignificant, random nonsense?
Well, one way not to do it is with beer.
In February, Charles Duhigg wrote in the New York Times Magazine about how "companies learn your secrets." He was investigating how retailers use their monster data sets to predict significant life changes for their customers, and then market their products accordingly. As it turns out, newly divorced consumers tend to purchase different brands of beer.
To be clear, this is an attempt at predictive modeling, not assigning causation. But the central question is the same -- coincidence or insight?
I guess the cynical viewpoint is that it doesn't matter. If you can use data to segment customer populations effectively, who cares what the underlying psychological factors might be?
In this way, business intelligence is sort of like stereotyping. Sure stereotyping ticks people off, but it's a good way to make some crude, quick generalizations. Only trouble is… it ticks people off!
Just because you CAN market to consumers based on data-driven presumptive generalities doesn't mean that you SHOULD, or that your customers will like it.
Could this be one of those classic instances where corporations get dumber by getting smarter?
I can't answer that definitively right now, because Meredith Vieira is going into labor.