"Chris Andersen outlines how the internet impacts on markets. How endless choice is creating unlimited demand. It’s the new economics of culture and commerce.Everything is changing. Businesses will survive by selling less of more.
Chris Andersen first explored the Long Tail “Theory” in an article in Wired magazine in 2004. The original article had three propositions:
The tail is longer than realised, it’s within reach economically, all niches add up to a significant market.
In the book Chris develops the thesis to suggest the future of internet trading is selling less of more, not in the high volume end of a traditional demand curve but in the endless long tail, transformed by the internet, into a new business opportunity.
The long tail can now be accessed because of changes in technology in relation to production, distribution, marketing and consumer access. The invisible market has turned visible.
The long tail has a mass of niche markets now accessible. There are so many niche products along the tail, collectively they can comprise a market rivalling the hits. p53
He develops a few new rules including the 98% rule - everything (well 98%) on offer, sells at least once. The graph never sells to zero. Expanding variety is flattening the demand curve.
So just what can be made of it all?
“Read this brilliant and timely book if you want to get a look at the future of business” Eric Schmidt CEO Google. “
James Suroweicki author of the “Wisdom of crowds” says The Long Tail is a rare achievement. It takes something seemingly familiar and makes us think about it in a new way. It’s a smart book.
Well it’s a book by a smart author judging by the sales figures. It’s a rare achievement in as much it takes something blindingly obvious and then makes us think about it in a new way by bending the tale.
As Lee Gomes in the Wall Street Journal argues, "it will be a long time before the tail wags the internet".
The reason is quite simple, the internet has a long tail but for business to be successful it needs to be a fat tail.The information in the book identifies a classic Pareto distribution. In 1897 an Italian mathematician Pareto had undertaken a study of income distribution in England in which 20% of the population accounted for 80% of the wealth. When this data was compared to many other countries in Europe, the ratio remained the same.
What Pareto had discovered, was not just a law of distribution as applied to income distribution, but a similar relationship existed in almost any sizeable dataset.
In 1949, George Zipf a Harvard economist analysed the frequency of word usage. The analysis led to the development of Zipf’s law to produce a predictive frequency distribution of words in rank order. Essentially, the frequency with which a word was used was proportional to 1 divided by the rank order in which it occurred.
The second word occurred as ½ as often as the first, the third 1/3 as often and so on.
Data analysed in this way produces a skewed distribution, a lot of occurrence at one end of the sample on the x axis of the graph, with a long tail where not much is happening at the other end.
In most statistical analyses Pareto distribution is rarely 80:20 in most analyses it is 70:30. 90:50 ratios, where 90% of the activity is accounted for by 50% of the parameter analysed, are almost ubiquitous.
The 70: 30 ratio is identifiable in many facets of internet activity. The data included in the Long Tail Book supports the proposition.Rhapsody data (p19) in the book supports Pareto and power law distribution as does data from Amazon, iTunes and Excite (p211)
Google information (p214) supports a classic Pareto long tail phenomenon
Andersen talks of Amazon.com and the proportion of book sales from outside the top 100,000, “it could be 40% according independent research but is more like 25%” according to Andersen.
Both can be right. The original research was undertaken in 1991*. The original research methodology is correct and the data conclusions statistically robust.
In 1991, the top 100,000 titles represented just under 5% of the total 2.3m titles list at Amazon perfectly in line with a Pareto distribution.
By 1994, when The Long Tail was first wagged, the Amazon titles list had grown to over 3.5 million. 100,000 was less than 3% of total titles and would statistically account for around 25% of the sales, still conforming to a normal Pareto distribution.
This is not evidence of the tail flattening as Andersen suggests.The internet phenomenon provides fantastic data sets for empirical analysis which in so many areas confirms a Pareto pattern.
According to Huberman (2001) Patterns on the web follow a universal and lawful behaviour In terms of distribution and performance. (P3)
“Pareto” patterns appear and identify a power law distribution. P3 and Power law distribution is skewed and has a long tail. (P27)
He analyses several activities :
Distribution of pages per site follows a pareto style distribution P3
Distribution of links per site follows a pareto style distribution. P3
Number of clicks per session follows a pareto style distribution. P46
Page down load time follows a pareto style distribution p75
Distribution of visitors per site follows a universal power law. Pareto is omnipresent.
On other sites on the web the same phenomena is observed. The Askjeeeves blog posts data confirming 70% of the search revenue is made from just 30% of the searches.
The 70:30 performance is exactly in line with our own empirical research on keyword search patterns and pay per click performance.
So what can we make of it all?
The long tail is a long tail, it’s a power law continuum. In a perfect market it will stretch to infinity. Everything sells at least once, well at least in part and smaller parts and so on as we move along the curve. It’s a rule.
The 98% rule is a tautology. The x axis, represents the “population” of any sample. If I exist I am on the axis. If I am on the x axis I will sell at least in part. "Je vends donc je suis."
It is true, the internet will expand the ranges on offer and thus the length of the x axis (Andersen p77), compare the offer from Barnes and Noble to Amazon. (p169)
But increasing the offer does not change the shape of the curve. An increased offer will expand the tail along the x axis but is it financially viable?
Statistically it is impossible for the niches to become greater than the hits. 50% of the offer generates 90% of the action. The niche markets along the tail have to share the small crumbs from the left overs.
Andersen admits commercial viability could be tricky. He says, “The further down the long tail, the more the need to keep the day job” (p78) Internet marketing does generate a long tail, the problem is the tail is just not fat enough.
References
Brynjolfsson, Erik M. Smith and J Hu. Consumer surplus in the Digital Economy : Estimating the Value of Increased Product Variety at Online BookSellers MIT 2003 (Based on 2001 data )
Huberman Bernando A The laws of the web : Patterns in the ecology of the web MIT Press 2001
Huberman, B.A., and L.A. Adamic. 1999 Growth dynamics of the world wide web Dynamic Theory of site popularity or Why the top performers just get better.
Gomes L Wall Street Journal It May Be a Long Time Before the Long Tail Is Wagging the Web July 26, 2006; Page B1