Skip to content

Shortening of the long tails

Chris Anderson popularized the idea that internet will fundamentally change the way media works: while the mass media and retail create a small number of big hits, the internet will “flatten” the field in the sense that there will be a large number of smaller hits. There are particular reasons for short-tailed world:

  • limited shelf space – smaller choice, the ubiquity of the “popular”
  • mass media broadcasts – coverage of the universally (“lowest common denominator”) popular at the expense of the niche interests

The delivery model of internet retail obliterates the shelf space restrictions. The distributed media, such as blogs, can cater to more specific interests. The consequence is that people can find something closer to their needs, but at the same time the amount of investment into those works is smaller.

Ilya Grigorik used the Netflix data (I have written about his analysis already) to examine the shifts in the taste across years on Netflix. He finds that with years from 2000 to 2005, the head has grown bigger, but the tail thinner:

Longtail comparison

The blockbusters have increased, while the tail has become thinner, especially among the hits. On the first sight, this would appear to indicate that the long-tailed vision is wrong. I disagree: the reasons lie in marketing. There is a distinct difference between early adopters and late adopters. Early adopters are more experimental in their tastes, which is the fundamental reason why they would go bother with novelty services. Moreover, they cannot find their favorite niche movie in the local Blockbuster, driving them towards mail distribution.

As the early adopters spread the word, as Netflix polishes up their service and reduces the cost, more and more of the late adopters join the ranks of Netflix customers, but these late adopters 1) do not have the itch for the rare 2) are not experimental 3) are still largely governed by the mass media. My guess is that the thinning of the tail is temporary.


  1. Chris Anderson says:


    Thanks for the interesting post, but I'm afraid you've fallen victim to the "percentage mistake". By using deciles for your horizontal axis, you haven't taken into consideration the fact that Netflix's inventory grew from less than 20,000 titles to more than 60,000 titles over that period. Graph it again with absolute numbers, rather than percentages, on the x axis and I think you'll see the predicted behavior.

    Many others have made this mistake, and I explain more about it here

  2. Koray says:

    So it wasn't just about limited shelf space. It's about the energy required to discover something by oneself. Netflix doesn't yet seem to have reduced that threshold enough.

  3. derek says:

    Grigorik misses the point of long tail marketing: it's not "the long tail will get fatter and the head thinner", it's "modern technology makes stocking and marketing the long tail feasible".

    Whether or not the long tail is getting thinner or fatter is an irrelevant question, the relevant question is whether the tail will become so anaemically thin that the sales volume needed (to find a niche beside the blockbuster vendor) will require the long-tail vendor to handle an unmanageably long tail.

    Since Grigorik destroys the information on how long the tail is getting, this is not a question his analysis is ever going to answer. I would say that the ever-increasing length of Netflix's tail from year to year is evidence that they are far from reaching the technical limits of their ability to make a living stocking and selling minority-interest titles.

  4. Aleks says:

    Chris, thanks for the link. It's fascinating how the discussion is essentially about the scale of the horizontal axis. Since Chris defined the curve, I guess his definition should be the one that is used. I also disagree with the tone of the WSJ attack on his work.

    However, if we just keep adding products, the growth of the tail is somewhat tautological. A more interesting definition would, like Koray reminded us, include the cost of retrieval.

    Some additional thoughts I have not included in my post: 1) The recommendation engines that Netflix has been using might also have an effect on the distribution of sales, perhaps by recommending less popular products based on the history of (mostly) popular purcahses. 2) The movies Netflix started with were all relatively reputable across years, but the new movies added might be close to an exhaustive sample of whatever gets made in a particular year.

  5. Andy says:

    The growth of the tail is not tautological if there is no demand for those products. From what I understand, the thesis of The Long Tail is that technology makes the cost of carrying low-demand products much cheaper, so products that simply wouldn't get sold before now get sold — basically, that there is a lot of demand in the aggregate for those extra 900,000 products but not for any one individually.

  6. Ted Dunning says:

    There lots and lots of problems with this data that make it hard to draw many conclusions.

    The biggest problem that hasn't been mentioned (directly) is that when Netflix started, they could ONLY appeal to the long-tail consumers because other channels had the head covered and because media owners didn't like their business model. Over time, the have been able to attack the head more effectively, but, even now, they have a natural preference for the tail because that is where they have serious advantages.

    In the usage data that I see here are Veoh, we have very good fit to a power law distribution of popularity except for a hundred or so of the most popular titles.

    To my mind, the real issue of long-tail feasibility is what kind of business constraint you have. A floor space constraint limits you to a portion of the X-axis (and most choose the left hand part for obvious reasons).

    A cost of stocking constraint is what Netflix has (the cost of distribution is proportional to usage and thus only scales the graph). This means that they can attack any part of the curve above a threshold. As their threshold moves lower, they will be able to hit more of the total universe of content, but they will not be able to address the truly long tail as long as a film distributor has to have a human who has to talk to a human at Netflix. Driving this threshold cost down will have dramatic impacts on their business if the exponent of the popularity curve really is -1 or less. The curves I see have a slope very close to -1, but we really can't say where that will go in the future.