Tom Webster, writing and speaking

Looking For Money In The Grass

Added on by Tom Webster.

20dollarbillHave you ever found something valuable, just lying on the street? I have a vivid childhood memory of walking to grade school (I would have been about 10) along a shady, tree-lined street where I grew up in northern Maine, and coming across just such a find. I was on a sidewalk adjacent to a grassy median, likely kicking a rock (which explained why I happened to be looking down) when I found it: a twenty dollar bill. Looking back on it, it was a remarkable find - the bill was folded up, and the winds were calm, so it wasn't blowing around - just caught up on a pine cone, or stuck in the long, rarely-mown grass of my town's public spaces. Naturally, I picked it up. For a ten-year-old growing up in the 70's, this was a fortune. I tucked it in my book bag and didn't say a word to anyone; nor did I dare to spend it for at least a week. Rest assured, it did indeed burn a hole in my pocket, and was quickly converted to comic books and candy (I would not have made a very good money launderer.) While the candy didn't last long, my fortuitous find did have a more significant impact - it changed my behavior. To a young boy without a knowledge of statistics, that grassy median was more than just part of the scenic Maine landscape, it was a gold mine.

From that day on, I walked that way to school *every day." When I would get to the grassy median, I would absolutely scour the grass, convinced that another $20.00 awaited me - or perhaps something even better. Surely, this was my lucky patch.

About two weeks later, I struck paydirt: there, lying in the grass in that same median, was a shiny Cylon action figure. No, this was not the smokin' hot Caprica Cylon from the modern Battlestar Galactica series, this was a genuine 1970's toaster-style Cylon. And I was a HUGE Galactica fan. Statistics be damned - you just try and tell THAT 10-year-old boy that there wasn't something magical going on with that patch of ground.

Today, of course, while I'd like to think my childish sense of wonder is intact, my knowledge of statistics is a bit more robust. The $20.00, and even the Cylon action figure, were surely coincidences. In fact, my radical behavior change - walking with my head down, scouring the same patch of ground every day - might have even had an opportunity cost: who knows - the other side of the street might have had a Viper, or maybe even Muffet. But I happened upon a coincidence, and I didn't question it. I never found anything on that patch of grass again.

Whether we know it or not, we face the same temptation that 10-year-old Tom faced every day with analytics - blog analytics, social media data, email stats and more. We mine our tweets and retweets, and discover that noon is the best time for us to post. Or we discover that more of our emails are opened on Thursdays. In short, we look at historical data, and we find a $20.00 bill. We watch webinars telling us that we are more likely to find that twenty bucks on a weekend, or after midnight, or on a boat, or with a goat. We accept the easy answer - the "what." We don't ask the more difficult question: the "why."

Finding those data nuggets is not science any more than my finding that twenty bucks was science. THIS, my friends, is science: you mine your social data and discover that, in the past, more of your tweets were retweeted on Wednesdays than other days of the week. You do not have an answer. You have a question: "hmmm - I wonder if time or day has an impact on the success of my social media efforts?" That you observe a particular day of the week as being more or less fertile is merely coincidence, until you prove it isn't. That means designing a controlled experiment to determine if there even IS a day/time effect on YOUR social media messages, long before you start to speculate on what day or time might be effective.

I'm getting tired of junk science. Of course you will find optimal times to tweet/email/etc. based upon historical data. Random chance assures that. But that doesn't give you a fact. It gives you a hypothesis - and before you start loading up your email blasts for the weekend, you first have to determine if there even IS a correlation between day, time or any other variable and the success metrics you care about. That is science. Not doing that work is nothing more than walking by that patch of grass, day after day, looking for a $20.

Good luck with that.