Help Stefan Campaign
Generate More Traffic to Your Site:
Understanding the 'Long Tail' Effect
If you've been keeping-up with the latest internet buzzwords, then you will undoubtedly have heard the term 'The Long Tail'. But do you actually know what it means, and are you aware of the implications for your on-line strategies?
The term actually comes from work on natural languages performed by Harvard linguist George Kingsley Zipf and published in 1949 and which was subsequently called 'Zipf's Law'. This law states that: in a corpus of natural language utterances, the frequency of any word is roughly inversely proportional to its rank in the frequency table. So, the most frequent word will occur approximately twice as often as the second most frequent word, which occurs twice as often as the fourth most frequent word, etc.
What you need to know:
Understanding the Mathematics:
The term 'Zipf's Law' has subsequently been used to refer to any of a family of related power law probability distributions. Plotting any behaviour that follows a power law distribution yields a curve such as the one below:
When plotted as a log-log curve then this gives a straight line and this log-log plot is the standard way of testing whether any kind of behaviour has a power law distribution.
Mathematically a power law can be expressed generally as: y = axk where a is the constant of proportionality and k is the exponent of the power law (ie how rapidly the curve approaches 1) and both are constants. Because a power law becomes a straight line when expressed logarithmically this can also be expressed as: log(y) = klog(x) + log(a) which is essentially the same as the classic equation for a straight line (y = mx + c).
But when it comes to matters of human interacton, matters of finance and income (especially if there's a degree of choice involved) then these behaviour are far more likely to display a power law behaviour. Indeed, many natural phenomena demonstrate power law behaviour.
To understand how these work, let's take a very simple example (the example of Zipf's law in fact). Where, if the first value is 1000 then the second value is 1/2 x 1000 = 500, the fifth value is 1/5 x 1000 = 200, the tenth value is 1/5 x 1000 = 100, the 50th value is 1/50 x 1000 = 20 and the nth value in the series is 1/n x 1000. In this series it takes 1000 iterations before the curve reaches 1. In such a curve the first entry accounts for 14% of the total, the second entry accounts for 7% of the total (together the first and second entries account for 21% of the total. Indeed, the first 10 values account for 41% of the total. Yet, this means that after the tenth value each individual entry is less than 9% of the maximum, taken together they represent 59% of all the possible values! This is shown graphically by colouring the Zipf curve below:
This is the so-called 'Long Tail' where, taken together rare events can actually sum together to account for a larger percentage of the total number occurrences than the star players. The phrase The Long Tail was first coined by Chris Anderson in a 2004 article in Wired Magazine. The article itself is fascinating and well-worth reading in its entirety.
The upshot of the article is that whilst normal retailers are constrained by shelf-space and need to shift, say 100000 copies of a CD to cover retail overheads (and only 1% of all CDs have this kind of turnover) virtual retailers, with smaller overheads and with no shelfspace to fill can effectively market each and every CD ever pressed. But why would they?
Economics of the 'Long Tail'
This is where the 'Long Tail' comes in and where the economics of this 'long tail' is quite different from those of normal retailers. After all, most of us are aware of the 80:20 rule (also known as Pareto's principle) often touted my record companies, TV producers and movie studios, where only 20% of a given studio's output will be hits (for major CD labels the numbers fall to 10%). However, all these examples are about hits not about sales per se. If the services were purely digital and there were no storage costs then you could make every single item available. Would this be worthwhile though? Surely hits are hits for a reason...
Well, yes they are. Mostly because the high-street retailers and record producers and movie studios skew the marketplace with what's been called 'the economics of scarcity'. By making only the hits visible they're creating the expectation of hits. This applies to books, tv shows, films, websites, blogs, and movies. Essentially anywhere there's human choice a power law distribution will soon come into play where most of the sales go to a few 'star performers'. This is unavoidable and entirely natural.
However, if people are given large-scale choice then something rather surprising happens. On-line stores such as iTunes (even Amazon) have shown that in the economics of choice rather than only 20% of inventory being successful the answer is more like 99%. OK, 30% of your sales may still go to the top 5 titles. Yet someone will buy just about anything you offer them. Given sufficient people coming to an on-line store then people are diverse enough for their interests to diverge significantly. As a result, there's a market for just about any kind of content and as long as something makes a sale then it can be considered 'successful'. It's just a new way to look at the economics of supply and demand.
The truth is that if you have a broad range of inventory (going into the Long Tail) of the popularity curve (a good example of this is a provider such as Rhapsody [a subscription-based music provider]) then even the most unpopular track will still have some customers. If the customers are given sufficient choice then the truth is that sales of these 'unpopular' tracks will actually outweigh the total sales of the top 80%. For any retailer who can use this model then they can raise their profits four-fold!
The Long Tail and the Small Internet Markerter
That's great for the big retailers, but what does the 'Long Tail' imply for us little guys? First of all, no matter how 'niche-oriented' your website is you will get traffic. However, your actual ranking may well lie very far down the tail of website rankings. Much of the work in terms of website optimization is intended to shift your website up the overall curve. Still, are you looking at the right things here?
For each property that behaves in a power series manner, you can take a sub-property which will also behave in a power series manner. Because the distribution of visitors to all the websites on the internet is a power series, then if your website is focusing on 'widgets' if you plot a curve of all the widget-associated websites and their traffic this will also be a power curve. Rather than focusing on trying to improve your overall web ranking you should be looking at your ranking in comparison with those sites in the same niche or market as yourself.
Just as you can subset sites into groups as abstractions from the entire internet you can also use power laws to look at sites themselves (and this can be very informative). If you plot all your web pages against the number of visitors they receive (say for a four month period) you will notice that what you obtain is a very good power law curve. Some of your pages are going to be very popular whilst others have very few visitors. You're not going to be able to flatten the curve much (though you can possbily skew it upwards somewat. However you will learn which pages are the most popular on your site and you can use this information to tweak the placement of AdSense ads, to improve the navigation from these pages to other sites and to ajust any other ads or offers on these popular pages. You may even discern why these pages are particularly popular and apply that information to the remainder of your site.
There's also quite a lot of discussion going on at the moment about pages' keyword optimization and the 'long tail'. If you're a small site then you're unlikely to gain first position in the search engines for your keywords (unless you're in a small niche). You may achieve this, but the power law states that you're unlikely to do so (and this statistical law makes many SEO firms seem like all they're really doing is touting snake oil). Some out there say that this doesn't matter as the 'long tail' will guarantee that all sites get some traffic.
However, this view is rather erroneous. What the 'long tail' view of keyword optimization means is that sites unlikely to gain the top spots for their keywords should concentrate their efforts on longer keywords (two or three word keywords) and more of them. The secret is understanding the 'long tail'. You can only succeed with this approach by covering as much of the tail as possible. Thus to compete using a 'long tail' approach with those who are first in your sector for a given keyword you need to cover as many of the other possible keywords as possible (the long tail). If you can do this then your overall traffic can actually be higher that a site positioned first for any given keyword.
The 'Long Tail' approach (or, more accurately the 'Economics of Abundance') also has implications for other areas of web businesses as well: especially affiliate system's. You've all seen them, affiliate programs selling you the secrets of making millions on line sold for $97 up to $600 in some cases. Often a number of eBooks or applications are thrown in to bump-up the price. But customers are becoming cannier. They know that distribution costs for these products are essentially nil, production costs are low and the only other question is the cost of the software to produce eBooks, audio commentaries or applications the cost of the time involved in creating them and residual costs from things like credit card processors or on-line payment processors such as PayPal.
The truth is $100 is too steep, so is $60 ($50 may be a more reasonable price). If what you're selling is just an eBook then $25 maximum is more the mark (though $15 would be more sensible). Whatever the information in these products the economies of the internet make higher prices unjustifiable. Indeed, these high prices come directly from the old 'Economics of Scarcity' model where income is generated from abut 15% sales. A move to an internet-based mode of sales (the 'economics of plenty') should mean that lower prices will attract more visitors to buy and in the end (as you're selling to the long tail) you'll actually make higher profits. If you'e starting your own store selling your own products then think long and hard not of what price point will get you maximum profit, but at what price point you'd actually buy your product (regardless of whether your might think it's worth more). Indeed, this is where even the big studios and the record companies are failing now as they're using on-line sales to bolster sales through physical stores (and pricing downloaded tracks and movies the same as if they were on physical DVDs or CDs, which price-wise is a complete nonsense). Because of lower physical storage and distribution costs on-line downloads should always be cheaper than buying physical information). Again it's an attempt at applying the economics of scarcity to a situation that follows the economics of scarcity.
Just as the internet has grown as the home of all kinds of obscure information it's also growing as the place to source all kinds of obscure products that are simply not economical to carry in a standard store (for example this week I bought saffron crocus bulbs, dried waterlily flowers and Australian acacia seeds). There is also the phenomenon that's being called Web 2.0 that effectively uses human interactions to create associations on the net. Amazon are already doing this with their recommendations which uses a customer's previous purchases to inform them of things that they might like. The way that people buy things are also used by Amazon to group similar items together. This can actually link a book that's not selling well with a bestseller and sometimes the poorer-selling item, due to reccomedations and reviews can out-sell the bestseller. Without this kind of association many kinds of items (books especially) would not be sold on Amazon.
Web 2.0 associations also occurs in Blogs where one blog may pick-up a story from another and if it's a good story then this may eventually be picked-up by one of the blogs high-up in the curve. This can drive far more traffic to the initial blog than it would normally have received.
Essentially understanding how the web works is of vital importance if you want to start making any real income on the web. But perhaps more importantly is understanding human behaviour and how this applies to the internet. The application of power law distributions to the internet is an important tool in our understanding of the web and it has a number of implications to how we both build websites and interact with others on the net.
Further Reading
Prices subject to change.
Prices subject to change.
1 2
|
If you enjoyed this page and would like to get more tips, tricks and offers to help you make the most of your most of your web presence please sign up for my Weekly e-mail newsletter. Please note that your details will never be sold and shared with others. You are signing-u for my e-mail only. |