Working with our design partners to deliver advanced web-based solutions

Home page

Blog

 

 
6 June 2005:The long tail

In this issue

Featured article
This month, the subject is a hot one in ecommerce circles - the long tail.
Business intelligence
Information, statistics, surveys from the web.
Technical tip
This month's tip is about web logs - and what you can learn from them.
News from the web
This month in the internet world.



The long tail

‘The long tail’ is a phrase you may hear form time to time.  It was sparked by an article in wired magazine late last year. 
http://www.wired.com/wired/archive/12.10/tail.html

The word ‘Tail’ comes from statistics – don’t worry about it.  It is nothing to do with animals.  It is used by statisticians to describe the outlying parts of a graph where the numbers – well - tail off.  For example spare parts for a Ford Fiesta are pretty heavily sold, but spares for your much cooler 1972 Chevy Impala are a totally different matter.  They would be considered ‘in the tail’.

The thing about the tail is that most high street shops don’t sell anything from it.  Try getting a workshop manual for your Chevy in your local branch of W.H.Smiths and you will see what I mean.  So because these items are hard to get hold of they don’t sell very well. Eventually they get remaindered or trashed and that is that. 

But the web changes the business model fundamentally.  A web operator is not restricted by the cost of floor space.  Maybe your stock is in a warehouse in the back of beyond – maybe you don’t carry stock at all because you are using a supplier who does ‘drop shipping’.   So you can afford to carry a much wider range of products.  Our hardware store www.handleanything.com carries a far bigger range than any local DIY outlet.   

Instead of bashing your head against a brick wall in W.H.Smiths, you type ‘workshop manual 1972 Chevy Impala’ into Google and bingo.  There is a bookshop in the sponsored links section with 40,000 car manuals on their list.

Now some maths.

If you have 10000 products in your tail and the probability of making a sale of any of them is (say) 0.1% (one tenth of 1 percent) then you will make 0.1 x 10000 / 100 sales = 10 sales.

If you have 50 products you sell all the time and the probability of making a sale is 20% you will make 20 x 50 / 100 = 10 sales.

In other words you can make just as much money out of your no-hopers in the tail as in your fast moving items. Possibly you will make more because the margins on specialised items will be higher. 

This new business model is powered by:
- infinite shelf space
- search tools to help people find their way around the much larger inventory
- more information such as readers recommendations

This is unlocking a whole new market which was unsatisfied before because the products were too hard to find.    It would have been hard to find a 1972 Impala workshop manual in 1972 – let alone 30 years later.  But someone somewhere on the web is probably selling it.

The same effect is hitting music, games, software, movies, radio stations and so on. 

So if your client is moving from a traditional market into e-commerce ask them if they should be thinking about long-tail items.  If they don’t know what you are talking about then educate them.  

Business intelligence

I enjoyed a webinar on lead generation in technology companies by Sirius decisions this month. If you are helping your client’s customer acquisition activies you might find the numbers here interesting.

There is an article about how to construct a page title or Google advert.  He gives a nice little formula for construction the phrase. 

Technical tips - Check your site web logs

If your hosting company provides web log analysis this is something you should look at regularly.  Every time anyone loads a page from your web site the server logs the transaction.  Here are some of the things you should look out for:

  • Hits – the statistic most people quote and also the least useful.  A hit is the download of a single file from the server.  If your web page had ten images on it, then when someone looks at the page there will be 11 hits (one for the page itself plus one for each file.    If your client wants to see better results just split all the images in half and double the number of hits.
  • Page views or page impressions.  Much more meaningful, as each page impression is someone looking at a page. 
  • A Visitor session is a single user visiting the site on one occasion.   Normally a gap of say 20 minutes without looking at a page completes a session.
  • Unique visitors: the number of different visitors looking at the site. 

You should also get things like the average session length, and number of page impressions per session.  These will tell you if visitors are studying the site or looking at one page and going away.

Your analysis will also tell you what search engines people are visiting from and what search terms they used, what operating system and what browser they are using. 

What the log analysis will never tell you is the identity of the visitor, although with many big corporations you will may be able to identify the company from their internet address.

News from the web

Advertisers are increasing their on-line spend at the expense of traditional media.  Forrester’s report also forecasts a total spend of $26billion in 2010 …  hmmm that is as much as cable or radio.  Do we believe the forecast – nahhh.  But the figures for 2005 are probably good.  You may prefer to believe an alternative forecast that on-line spend rates will slow.

Internet connectivity in India has been problematic in the past, and only 2 in 10,000 people have broadband.  Dishnet is planning to create a country-wide wi-fi network.

You may remember the distressing story about a girl who was trapped in her bedroom while her parents were being attacked.  She couldn’t dial emergency services because the VOIP service didn’t support the 911 emergency number.  Now the state of Connecticut is suing Vonage about the poor quality of the 911 service on offer.

The FCC in the US has set a four month deadline for VOIP providers to create a good service.

Google acquired the Urchin web analytics company in March (sorry guys I missed it).  We use Urchin for web log analysis and it is pretty good.  They have an on-line analysis service which they just reduced in price from $450 per month to $199.  Google is also enhancing it to cover their own advertising in the analysis.  This may make Urchin a must-have for Google advertisers.  We shall certainly be looking at it.


Firefox continues to grow and has now reached 7% of browser share.

Yahoo has launched a music downloading service in the US and will soon roll out a European version.

Ask a computer a question and it will always come up with the same answer – right?  Not if you have different search engines.  It seems only 3% of search results are shared by the major search engines. 

The New York Times will be charging for its web site content soon – but thankfully not for most current news stories. 

Google is allowing users to customise their home page.

I listen to Internet radio and it works.  It is growing but I have to say despite rather than because of developments in Windows Media Player which seem to always direct you to paid-for services and make radio stations hard to find.