How Much Does The Long Tail Weigh? Part Deux

All this Long Tail talk had me thinking. Was my rather unscientific hypothesis, based on absolutely no empirical data or case study, correct in guessing the weight of the Long Tail at 20% of a total given market? O.K., I confess I was zillowing the Long Tail. I’m in real estate, what did you expect?

To get a better footing, I contacted Professor Michael D. Smith, Information Technology & Marketing professor at Carnegie Mellon University, who, while at MIT wrote, together with Profs. Brynjolfsson & Hu, the oft-cited research paper Consumer Surpluses in The Digital Economy: Estimating the Value of Increased Product Varities at Online Booklsellers (2003). This paper was influential in Chris Anderson’s formulation of the Long Tail. I hoped the professor could give me the answer to my question, to wit, “How much does the Long Tail weigh?” In other words, if the total area under the line in the graph equals 100%, what percentage of a market does the Long Tail cover and should I really be chasing my tail if I can’t get it all in the Amazonian way, i.e carry (or have access to) every darn long tail item.

First, let me say that Mike (we’re on a first name basis) is a great guy, willing to bear my numerous questions and what-ifs with patience and grace. He was also kind enough to share his research papers with me. I could only offer him An Engineer’s Guide to Romance, written for MIT frosh, which I found in the comments section of a post by my blog buddy, Brian Clark of the famed Copyblogger. But I digress.

Here’s what I’ve learned from Mike:

1. The total area in the graph under the line equals 100%. Yes!

2. There is “no precise definition for what is the long tail portion“. Drats!

3. Amazon’s long tails sales (obscure titles not stocked by brick & mortar bookstores) were somewhere between 30-50%. Not enough to answer the question, but close. (personal note: perhaps 40% of sales comes from Affiliates, according to Wikipedia. If so, I’d argue the case of the Long Affiliates as a more valuable marketing strategy)

I still need to know: in the world of book sales, for example, what percentage of total book sales (every seller counted) is attributed to obscure long tail sales? (I’m still wagering only 20%) Is it that I have a bias toward Italian economists like Pareto because of Italy’s World Cup? But I digress again. If the percentage is 20%, you’ll need to be an Amazon to take full advantage of long tail scaled economics (did I just say that?) or have Long Affiliates.

So that’s where I am. Does anyone know how to weigh this monster’s tail?

bhyde

There are lots of populations that are distributed in highly skewed populations, like book sales. Income, wealth, earthquakes, storms, city sizes, number of friends, etc. etc. It's hair splitting fun to point out that some of these are powerlaw and some are other distributions, and it's fun also to debate if the entire population is ruled by the same distirbution or if there is one thing going on way out on the tail of the population (as for example there appears to be with most of them).

The question of how much of the weight, energy, or wealth is actually in the tail is a bit difficult to awnswer for two to four simple reasons.

First, now severly skewed is the distibution; i.e. the income distribution is much more stern in the US than it is in Japan.

Second there is usually a cutoff below which we aren't interested; but you need to know that. In income there is a point were people starve and die. In earthquakes and storms there comes a point were the events are so small that people stop counting. Amazon is unlikely to sell many units of anything for less than the cost of shipping.

There you have to decide where the tail starts. For income that's the question of who exactly will be declared to be lower class? Do you include the middleclass? For books you have to decide where the block busters stop. The weight of a given member of the population is very roughly 1/N; which it get's small fast. So you might be comfortable deciding that the fortune 500 is a good cut off. Though note they are all billionaires.

The final problem is what is zero? Poor people have marginal income that's hard to measure. Do we include pamplet's handed out in exchange for a moment of your time.

Setting aside the difficulties. If you have a distribution that is 1/N (not too stern). We can pick a cut off, say 50. Pick a point where the tail starts, say 500. Make zero be 0. Doing the sums the tail is 56% of the weight. Things get interesting if rather than 5000, you have millions.

You might try doing it for all the propetry accessments in your town. Sort them into decending order and sum up the value of the first 5% v.s. the rest.

jfsellsius

Thank you, bhyde, for your thoughtful comment.

Sellsius real estate marketing blog

How Much Does The Long Tail Weigh? Part Deux