« Open Peer Review | Main | Meanwhile, Casey Kasem is resting comfortably in retirement... »

June 19, 2006


Chris Gilbey

This is not quite a long tail comment, but I think it has some strong relevance...
Apparently - much to my surprise - older demographic consumers are much more interested in buying music on CD's than buying downloads of music. I was really surprised about this - but the guy that told me is involved in recording and selling music of live concerts. And there is high demand from the people who go to see the ancient rockers that he gets to record for CD's rather than downloads to iPods...

I would have thought we had already reached the tipping point. Apparently not so.... Any thoughts on this?

Chris Anderson


That doesn't suprise me, since older people are tyically the slowest to embrace new technology. So given a choice between a download and a CD, they'll often pick the CD. But if they *aren't* given the choice of a CD, which is increasingly the case for non-hit music in traditional retail stores, they are turning to downloads. Witness the suprising strength of classical on iTunes (I'm told it's a double-digit percentage of sales), which reflects the shrinking shelf space given to classical music in most stores. Downloads are now the easiest way to get classical, and so older demographic consumers are turning to that, regardless of their technology preferences.

Jakob Nielsen

Based purely on this slide, I think you are underestimating the size of the tail (by overestimating the size of the head).

The slide says, "Wal-Mart defined as top 50,000 tracks."

In practice, however, Wal-Mart (or any other store) would not be carrying the top 50,000 items, even if they had 50,000 items in stock. Two reasons:

  • it's impossible to predict exactly what will be the best-sellers, and
  • it's impossible to manage inventory such that one never sells out.

I know that Wal-Mart has extremely talented people running their computers, trying to find out what are the best-selling items and trying to avoid running out of inventory. But they can't do a perfect job.

Thus, in practice, when customers arrive at a store trying to buy a top-50,000 item, they will sometime find that it's not available, either because it's not stocked or because it's sold out. Such customers wil tend to leave the store.

On the other hand, a download site obviously won't sell out, and with 1.3M items in the database it's virtually guaranted that all the top-50,000 items are for sale. (Plus, of course, many more "tail" items.)

Chris Anderson


That's a very perceptive point, and you're right that I should have a better methodology for defining head=Wal-Mart inventory. The 50,000 figure I use here is indeed an exaggeration, but I use it because it leads to the most conservative estimate of the size of the tail (it's basically 4,200 unique titles--that's the total in an average Wal-Mart--divided by an average of 12 tracks per album)



Will Page

Hi Chris,

How deep does the data go? I’m particularly interested in extending the analysis to see where we are with the un-bundling of the album to a track-by-track purchasing model. That is, whilst the tail is lengthening and fattening, is there a displacement effect which we’re ignoring? Is there a causation argument that we’ve been overlooking? For the sake of debate, here goes.

Let’s say I spend $10 a week on music, well – five years ago that entire $10 would have been soaked up by purchasing the new hit U2 record. Now, I may spend $2 on the best two tracks from the new hit album the change my consumption with the other $8 going further down the tail. We may well see a constant demand for music in real cash terms – I’m prepared to shell out $10 a week - but the purchasing pattern has changed.

This made me question the 50,000 top tracks used here. Whilst I appreciate it’s a good proxy (4,200 albums x 12 tracks), is it possible to consider long tail effects when there is a clear option available to the consumer of an album or track purchase? My hunch is that we’re looking at a purely on-line hit album (as in how many of the 4,200 were sold in full over the net) versus tail type tracks type analysis, but that we don’t have enough history to establish trends. Still, I’d love to work on some counterfactual thinking with you to see if we can sketch it.

It may well come to the same intuitive conclusion as your work, but I’d like to know how big a part un-bundling is playing on the tail, to recap – my proposition is that it’s the fact that I can consume my optimal amount of U2 and now have a spare $8 which leads me to look at other artists which, in turn, is driving the tail. The conventional tail arguments, (filters, aggregators, unlimited space etc), help facilitate that behaviour – not the other way round.

PS: Obviously, this is all dependent on the customer choosing to pay for music in the first place. Relax that assumption (as most of the digital music consuming population seems to be doing) and you’ve got a whole different analytical puzzle on your hands.

Will Page

Chris Anderson


No doubt there's more we can do to fine tune the analysis, but let's not make the best the enemy of the good. 25,000 or 50,000, we're still looking at a tiny fraction of the inventory available online. My tail calculations may be off by a few percent because of the factors you describe, but I doubt it's more than that.

The best statistical test would be to get a list of Wal-Mart's music sales in a given month and compare that with an online retailer. But unless you have a better in at Wal-Mart than I do (or Soundscan, which doesn't include Wal-Mart either), we'll have to make do with a few informed estimates.

Will Page


Thanks for your reply; I still think my point might still have something we should explore in terms of displacement and causation. Let’s say you spend a constant share of your disposable income on music and run inflation / deflation effects into the analysis. Disposable income has risen, the price of a CD has fallen - add in my previous points about un-bundling and you have consumers with a lot more money with a lot more opportunity to experiment on a lot more different providers of music …as opposed to saving up all your pocket money as a kid so you could buy a hit album like Guns n Roses Appetite for Destruction (...as I did, and was the first kid in Edinburgh to do so and long before they became famous I hasten to add - evidence if need be that I was born to be an A&R man). My revised proposition, (purely for the sake of debate, again), is that this financial 'slack' drives people to experiment further down the tail, and the tail grows and with it the long tail effect kicks in.


Camera Bags

We are working on support for many popular open standards. Our company has historically supported open standards (rtsp, smil, Ogg on Linux), and is continuing to do so (RSS, XML, more coming...).

We have discussed supporting XSPF, and are interested in doing so. However, it is currently prioritized behind several other open standards initiatives we are driving towards.

Pertaining to the "Free Rhapsody" comment above; Rhapsody has a free service which allows for 25 song plays per month plus access to 25 free radio stations. For 9.95 a user gets unlimited song plays and access to all of our radio stations (including custom stations).

The comments to this entry are closed.


The Long Tail by Chris Anderson

Notes and sources for the book

FREE was available in all digital forms--ebook, web book, and audiobook--for free shortly after the hardcover was published on July 7th. The ebook and web book were free for a limited time and limited to certain geographic regions as determined by each national publisher; the unabridged MP3 audiobook (get zip file here) will remain free forever, available in all regions.

Order the hardcover now!