drawing the short end of the long tail

Over at Harvard Business Review, there’s a very interesting article by Anita Elberse dealing with the “long tail” phenomenon (Teppo had a post that mentioned the Chris Anderson book a while back) in markets for cultural products.  I haven’t read the Anderson book, but from Elberse’s description it sounds like one of those “everything is different in the digital age” proclamations that is high on promise but skimpy on hard data (which means that it probably does well on the lecture circuit).

In essence, the Anderson argument as I understand it from Elberse’s exposition, is that the age of “traditional marketing” for cultural goods, in which failure is the norm and you hope for the “big hit” that will pay for all of the failures (i.e. Hirsch 1972), is over.  With the decline of the brick and mortar retailer and the emergence of online retailing, customers now have at their disposal a virtually unlimited array of choices, which means that even the most obscure niche taste can be satisfied.  Under these conditions, cultural industries can make money by supporting more niche products and not betting everything on the blockbuster bank.

The argument has been popularized as the “long tail” phenomenon because of Anderson’s suggestion that under these conditions the curve that plots observed popularity against the total number of products offered flattens, with a larger variety of products experiencing some level of success (and with the concentration ratio declining).   Notice that this implies that a more rational strategy for a given cultural producer is to expand the range of products offered and to transfer resources previously devoted to the promotion and marketing of a very few candidate blockbusters toward the promotion and marketing of a wider range of items with possible niche appeal.

Elberse argues to the contrary.  Based on her research on video and music sales (I found a preprint of the one of the papers here), she argues that the emergence of online retailing for cultural goods, does not necessarily imply the end of a “superstar” market and the emergence of a long-tailed niche regime.  She finds that while it is true that that there has been a secular trend toward a flatter tail during that last 5-7 years, there has also been an increase in the number of products that experience no-sales (outright failures), as well as a secular trend toward greater concentration towards the fat-end of the distribution.

Drawing on the work of sociologist William McPhee (from Berelson, Lazarsfeld and McPhee), she notes that the data don’t seem to support the long-tail argument because that argument makes the wrong assumptions about the way that the market is segmented and about the actual dynamics of popularity for cultural goods.  Instead as argued by McPhee, it seems that the cultural goods market is partitioned along the following binary: “heavy users” and “light users.”  Popular products do well because they tend to attract a disproportionate share of light users.  They do really well, because they also tend to attract a disproportionate share of heavy users.  Obscure products on the other hand, are only likely to attract heavy users and not light users.

The data (taken from Quickflix and Soundscan) seem to support this hypothesis.  Persons who are towards the bottom end of purchases per week tend to disproportionately select titles from the top popularity decile. Furthermore, persons who select titles from the lowest popularity decile tend to be disproportionately likely to belong to the top group of purchasers.  However, (and I think this is the key to the long-tail’s argument empirical failure), while smaller in absolute size than that of light users, the bulk of the downloads of heavy users is still composed of products from the top popularity decile.

Just in case I just confused you here’s Elberse’s findings in (somewhat) plain English:  non-discerning, occassional culture consumers mainly consume popular culture, discerning, frequent culture consumers consume both obscure and popular culture.  For discerning, frequent culture consumers there is no trade-off between niche and popular culture.

Elberse argues that there’s another reason why a “niche” advantage fails to materialize:  both heavy and light users are tougher (in terms of their ratings) on obscure goods than they are on popular goods.  This does not appear to be an expert-being-tougher-on-niche-goods effect, since heavy consumers go as easy on popular goods as do light consumers.

Given this pattern of audience segmentation, it is clear that a “wired” culture distribution regime, will not result in any of the consequences that the long-tail gurus predict, but will clearly preserve the superstar advantages of blockbusters in the digital distribution era, while producing a slight advantage for some niche items.  This is exactly what the data show.  Furthermore, Hirsch’s dictum that failure is the norm in culture industry systems will probably become even more of an “iron law” than ever before (even after controlling for the increasing number of products offered).

As a side-note, it is also clear that Anderson got carried away with the wrong audience segmentation model.  He imagined a world increasingly composed of niche cognoscenti, literati with very delimited and select tastes who would benefit from a diversified cultural availability regime.  In Peterson’s (1992) words, he imagined the expert culture-consuming audience as a “mosaic” of univores, each with delimited tastes for very obscure stuff (and possibly with a distaste for the stuff that was popular and which the “herd” tended to follow). It is clear from what we know on sociological resarch on audience segmentation that Mr. Anderson was off to the wrong start.  Heavy culture consumers are not niche “highbrows” but are “omnivores”; the fact that they do have strong, expert tastes for niche cultures does not imply that they stay away from popular culture (tautologously defined as “that culture which is popular”).  There are a bunch of reasons for this, but one powerful one is hitting the theaters today in the form of the latest Pixar flick.

Written by Omar

June 27, 2008 at 4:16 pm

5 Responses

Subscribe to comments with RSS.

  1. “…high on promise but skimpy on hard data…”

    Ironic given that the same author has also just published an empiricism++ manifesto/rant on how models and theory are dead: “The End of Theory: The Data Deluge Makes the Scientific Method Obsolete

    Here’s Andrew on the piece.



    June 27, 2008 at 5:54 pm

  2. Nice post. My wife is gonna get a kick out of these papers.



    June 27, 2008 at 6:01 pm

  3. “My wife is gonna get a kick out of these papers.”

    Mike: How come?



    June 27, 2008 at 6:09 pm

  4. I sent this post to my partner, who will no doubt tell me that hipsters are the paradigmatic niche cognoscenti (or they try to be). He will also say that hipsters are just yuppies without money, which may explain consumption trends, and “ironic” consumption.

    And yes, Mike, how come?


    belle lettre

    June 27, 2008 at 9:28 pm

  5. […] post on the long tail, here, makes it into the comments pages of HBR, here (copied below the […]


Comments are closed.

%d bloggers like this: