Bad Data and Spaghetti Trees

A recent Manager Tools Email letter discussed the lessons from a very famous April Fool’s day trick in the UK.

In 1957, the BBC ran a clip of spaghetti being made.  It wasn’t being made in a kitchen, instead the clip showed spaghetti growing on trees.  In the 1950’s, pasta was not an everyday food in Britain. It was an exotic delicacy and purchased in cans.  The joke of it growing on trees was perfectly believable, especially when they used a very respected broadcaster Richard Dimbleby for voice over.

At Manager Tools,  they were looking at data around purchases on their site.  They dumped the data from of the website, and started analyzing.  It was all going well, until someone point out the analysis said 80% of the people who bought only one item had bought it for someone else.  That didn’t seem right.

Like the spaghetti trees, since the data came from a respected source – their database, they didn’t initial question its validity.

A respected source though, doesn’t mean you shouldn’t question what is put in front of you.  “If the data doesn’t make sense, then acting on it makes even less sense!”

“There’s two lessons from this.  First, always have someone else look at your data and analysis if there’s the smallest chance it might not be right.  Second, if someone’s data doesn’t look right, say something.  If you Google “bad data”, you’ll find hundreds of examples of companies spending millions based on data that’s just plain wrong, from New Coke to Enron.   You might feel like a little cog some days, but little cogs drive big wheels.  It was someone like you who started it all, and the story ended on the front page of the Wall Street Journal with a headline that read, “Bad Data ….”

This was a nice and timely reminder.  I’m on a portion of the product program’s deliverables that are dealing with “data analytics” at work.