My friend Jay (owner of REDesignations.com) and I had a great discussion yesterday about my blog post on long tail. He offered a working definition; in a nut: appealing to niches. The nichier the better. I like Jay's nut-shelling of it, and is where I'll be heading in my continued exploration tomorrow about this curious term.
But, in this post I'm addressing the question, "why 'long tail'"? How do you get "long tail" out of niche marketing? Why not call it, oh I don't know, "niche marketing"?
Well, "long tail" is a term used to describe a type of statistical distribution--a bar graph. I guess technically it speaks to a trend line that describes the information in that bar graph. The term itself was popularized in October 2004 when Chris Anderson wrote an essay in Wired magazine about the effects of this concept on current and future business models. And, though the term itself has been widely ascribed to Anderson's popularization of it, there is earlier reference to it in a February 2003 essay titled, "Power Laws, Web Laws and Inequality," written by Clay Shirky.
But why does a long tail graph look the way it does? I mean, what's the concept behind it? Why is it tall on the left and then really tiny on the right?
For an answer that even approaches addressing that question, indulge me a bit in a few visuals.
Visualizing A Long Tail
Imagine a speaker standing at the front of a room full of, say 300 people. She takes an informal poll by asking a show of hands from everybody who is wearing the color yellow; 53 people raise their hand. How many are wearing the color red? One hundred and twenty five hands go up. And then imagine she goes on like this through a list of, say, 25 different colors.
Now, if an assistant were to put all these responses on a bar graph, she might end up with a ragged bar graph that looks like this.
The picture above may not look very long taily at the moment. But, fact is, there's a long tail in it. The trick is you have to rejigger the graph. If you re-order all the bars and their associated data so that the biggest ones (most popular) are on the left and the smallest ones are on the right, then you'll see a decreasing series that looks like this.
And then to make it a little more prominent, let's sort of connect the dots between the high point of one bar graph to that of the next. (We're starting to create something here which statisticians call a trend line.)
Are you starting to see it? The red line is the "long tail" taking shape. It's a line that statisticians and mathematicians like to call a trend line. Right now it has a ragged look to it. But that's only because we're dealing with a finite data set. That is, 25 colors.
But, if we could somehow extend that data set to, oh I don't know, say an infinite number of colors, then the line would take on a smoother look. But since that wouldn't be at all practical, we turn to the mathematical toolbox of statisticians.
Pretending Infinity
Statisticians are great because they have many many math models that help them pretend to have an infinite data set by starting with our list of 25 colors. They can then create smooth trend lines--or curves--to match that pretend set. One such formula is something called a Power Law. Which, if I apply that using my trusty formula functions in my quick and dirty Excel spreadsheet, we end up with a curve that looks like the red line below. (Note: I'm clearly over-simplifying things here. But that's the whole point of this post! That said, it's still worth noting in the back of your head that there are reasons for choosing to apply one formula versus another. Again, another reason why statisticians are as smart as they are.)
Now take the bars away entirely and just focus on the trend line. And you end up with something looks like a... (dare I say it?).
...Do you see the resemblance to a long tail? But, as they say, "wait, there's more." As Anderson and Shirky point out, there are business/marketing insights to be gleaned from all this. I'll try to lay out some of those insights in similar plain english fashion in my next post.
In the meantime, if this helps in any way, let me know. If you have additional insights to help explain this further (in practical, down-to-earth terms please) then please do share. I'll chat more tomorrow.
If you liked this, or any of the articles on this site, please subscribe!
If your organization can use help transforming live training for the web,
find out more here.
