When you spend as long as I do immersed in a world of data you learn to accept many things as essentially constant. You know that they don’t necessarily have to be true, but at the very least generally are. For the most part these are the basic trends:
-People who are more familiar with a product are more likely to buy.
-People are more likely to recommend something to a friend then they are to purchase it themselves.
Every once in a while though, you get some information that defies both your past experiences as well as your logical understanding of what you would expect to happen.
One of these moments happened just this morning. Consumers who attended a mobile tour event were less likely to identify the brand than those who hadn’t. Funnily enough the first reaction wasn’t even to try to figure out why this trend might have occurred, it was to figure out where the mistake in the analysis had occurred.
The conclusion from the data ran so counter to what I expected I assumed human error was more likely than the data’s conclusion. It brought to light the fact that I was making assumptions about my data before analysis, a dangerous prospect for anyone.
But that was just step one.
Once I knew the flaw wasn’t in the data, I still had the issue of figuring out what in the world was causing such an odd discrepancy. Any number of things could have caused it, and there was nothing inherent about the product that made one variable much more likely than another, so I had to take a top down approach.
Maybe most of my event attendees were from a place where the product wasn’t typically sold? Nope. Maybe my event attendees were less likely to buy this type of drink in general? Nope again. Were more of my event attendees of a certain gender, and if so did that same trend appear when comparing men and women? Still not it.
Then the obvious choice presented itself, were my event attendees old enough that they were outside the brands targeting? BINGO! I’d known that average age had be
en trending a bit high, but didn’t take into account how that might affect the data. But we had a brand targeting a younger crowd, so a collection of 40+ year olds not knowing about the product makes sense.
What I had initially treated simply as feedback for the client (your events are hitting slightly above your target age) had real implications for me as well.
So what’s my grand takeaway from this?
I think it’s as simple as “You need to ignore your pre-conceived notions” Staying aware of the fact that your data won’t always fit the way you would expect, even at the most basic levels is important, vital even.
So next time you have something that confuses you, take a moment and think, what am I assuming about this, and might that be wrong?
Photo Source: http://www.flickr.com/photos/suckamc/2643190902/sizes/o/in/photostream/