Lee Gomes responds to Chris Anderson

Late yesterday, in an email exchange, I asked Lee Gomes what he thought of Chris Anderson’s response to his Wall Street Journal column on the Long Tail. (Which I discussed here.) In particular, I wanted to know whether he thought (as I do) that Anderson made a valid point in explaining why summing up sales in percentage terms might obfuscate the Long Tail effect. Gomes sent me a long and thoughtful reply, which he said I was free to post. In hopes of furthering the debate about the extent of the Long Tail phenomenon, I will do that:

Hi Nick:

Yes indeed, Chris has a very valid point in measuring head and tail the way he does. But so do I. I had that paragraph in my column to tell readers that saying that Amazon gets 25% of its revenue from its tail (everything past the 100,000 best-sellers) is just another way of saying that 2.7% of its titles represent 75% of its sales. (It has 3.7 million books.) I made it clear in the column that this was MY method, and that Chris did not approve of it. (By the way, it is simply NOT true that his economist told me my approach was wrong, as he says on his blog; the exact opposite occurred. I also spoke to the head of the statistics department at a UC college who said my approach was the more “natural” one.) Readers, though, are free to pick the method they like. Please note, though, that I at least clearly laid out the two approaches; nowhere in his book does Chris let on how small, as a percent of sales, his tail really is.

While I am at it, I’d like to correct an extremely serious misrepresentation Chris made at the end of his blog posting, to the effect that Anita Elberse of Harvard “urged” me not to characterize her work the way I did. This is manifestly false. Chris is either misremembering or deliberately conflating two separate issues. Prof. Elberse did indeed in an email remind me that the data she had for Netflix was under NDA, and I could thus not report it. But the comment had nothing to do with what Chris says it does. Let Prof. Elberse herself describe whether I got it right; below is the full text of an email she sent me after the story ran:

“I just read your article, and just wanted to thank you for being so careful in quoting me. I wish all journalists stayed this close to what was actually said! :-)

“You did beat me ‘to the market’ with your article, but I hope our academic article (which should be ready in a few weeks) will further clarify the long tail phenomenon (or lack thereof).”

She also posted a comment on Chris’ blog implying much the same thing, and correcting Chris on another matter; I urge you to read it.

More broadly, when you attend Mainstream Media Community College, you are taught that if you say something in your lede, you need to back it up. My lede was that “I don’t think things are changing as much as he does,” and I think I backed it up four different times.

1) While Chris seems to have repealed the “98 Percent Rule” in his interviews with me, he didn’t do as much in the book. This is how he begins the book, and any reader, after hearing the “Rule” described as “nearly universal,” would, if nothing else, assume that it was true at all the examples the book describes. Chris defended the fact that it’s not by noting to me that his book wasn’t titled “The 98 Percent Rule;” does this mean that any sentence without “Long Tail” in can’t be assumed to be accurate? He also complains in his blog comments that I didn’t mention the 95% play rates at Netflix. But I wasn’t trying to show the “Rule” was NEVER true; he is the one who said it was “universal.”

2) The book’s showstopper notion about the possibility of the tail possibly being greater than the head is a fine idea; it is just not happening anywhere in the world, despite the impression the book leaves. (Incidentally, I described the book’s treatment of this possibility in my column in the same future conditional tense as Chris described it in the book; I never said he said it was happening now, as Chris implies on his blog.) And nowhere in the book does Chris say that Netflix and Amazon won’t reach Long Tail nirvana for at least a decade – not by his definition of head and tail, and certainly not by my alternative one. I can’t believe readers wouldn’t want to have been told all this; they weren’t.

3) The notion that “hits matter less,” which is explicit in the book and implicit in the Long Tail idea, was rebutted just about everywhere I looked. The 90/10 rule was pretty constant. And iTunes looks like Billboard, not some paradise of niches.

4) Other economists are reaching very different conclusions from Chris. After I started asking Chris questions about his findings, he gave me the names of three economists who he said would back him up. One of them I ran out of time to call. One of them was Prof. Elberse. The third was MIT’s Erik Brynjolfsson, who told me that there indeed is a shift occurring as things move online, but it’s on the order of an 80/20 distribution moving to a 70/30 one. Had I had more space, I would have mentioned this as well, for it too is less earth-shattering than what the book leads readers to believe will occur.

Nick, I would never say that the Internet isn’t changing a lot of things, including perhaps consumption patterns. But in case I haven’t made myself clear, I don’t think it’s changing things as much as Chris does.


PS Feel free to post any or all of this.

8 thoughts on “Lee Gomes responds to Chris Anderson

  1. Peter Rip


    For me the ‘long tail argument’ isn’t interesting because of the relative distributions of unit volume. It is interesting because of the distribution of profits. “Long tail products” have more distribution friction than the top of the distribution. Matching buyer and seller is a larger piece of the value chain. Therefore the channel can extract higher unit margins from both buyer and seller than they can from the top of the distribution. So cross-sell, a weak form of bundling, is a key.

    Amazon’s real breakthrough (IMO) was not in having a long tale of titles to *stock*. It was recognizing that they could offer a value proposition of completeness by product category and have higher ROE on Zstores (the non-Amazon merchants selling long tail stuff, including used stuff) by capturing the distribution margin with zero marginal PPE &inventory investment. The long tail is more profitable to the channel than the head of the distribution. The channel has more power. But hits attract the shoppers/readers, always will.

    Cross-sell and building a universal brand provides the real method of monetizing the long tail.

    So I don’t care if Gomes or Anderson are right. Volume isn’t the real economic issue. Profit is. I’m looking for companies that can extract value with a ‘long tail’ promise to the customer (completeness), but have a way to capture profit per unit proportional to their search value-added.

  2. Glenn Fleishman

    A missing piece, and the one you’re asking for, is profit, sure. And if you look at Amazon’s long tail, their revenue and profit is much higher on most of those less-sold items. In books, Amazon.com charges full price and a surcharge on somewhere between 1m and 2m titles that aren’t in general distribution. (When I worked at Amazon about 10 years ago, there were about 500k books in general distribution; in-print books were about 1.5m. I believe both numbers are now much higher despite lower overall book industry sales adjusted for inflation.)

    This isn’t true in every industry. Long tail songs might cost the same to a consumer and thus have higher costs to recover across fewer people when you factor in digitization, and even storage. No matter how check storage and backup is, a song that sells 100 songs a year costs the same to store as one that sells 1000 or 1m.

  3. Sid Steward


    I think you’ve made the most important observation about the tail (at least, as I understand you). It recalls to mind comments I’ve read by Bezos regarding the sales of used books next to new books (by third parties no less!) on Amazon. That broadening Amazon’s offerings would boost new book sales, not erode them. That used book buyers are a tail that won’t start wagging the dog.

    In other words, the tail will always be a tail. While this might disappoint some, I believe it’s great news for businesses. Businesses can gain advantages by offering ‘head to tail’ services without fearing that the tail will overtake the head. So it is also good news for tailsters.

  4. Tubby Bartles

    Chris complicated things by writing a complete rebuttal (which may contain other errors) rather than just pointing out Lee’s core, glaring error.

    The 80/20 rule doesn’t become a 70/30 rule, that’s silly and betrays a complete misunderstanding of what the long tail is about.

    Let’s say I carry 10 CD’s in my store. 80% of my sales come from 2 of the CD’s.

    Now I decide to carry 100,000 CD’s. Given the massive selection, now 80% of my sales come from say 200 CD’s – still a small number, but 100X larger than it was before. But because I have 100K in stock, now 80% of my sales come from 0.2% of my inventory. Does that mean there is no long tail effect? That’s the nonsensical implication of Lee’s method. He defends it as “the more standard way of looking at things”, which is true. But that’s the point – the standard way of looking at things no longer makes sense in an infinite shelf space world.

    Saying that the 80/20 rule moves to 70/30 makes no sense – if it’s not 80/0.2, then you’re still discussing a world where shelf space matters.

    The real comparison Lee should have done if he wants to discuss Amazon would be: what % of Amazon’s revenues come from titles not on the shelf at Barnes & Noble? That is the measure of Amazon’s business in the tail.

    The answers range from about 30% (Bezos @ Amazon guesses this off the cuff) to about 60% (Brynjolfsson’s more detailed study at MIT). Either way, it’s a huge percentage, and hardly the level Lee can dismiss. And it’s not just Netflix – it’s Amazon, Ebay, Google, and a host of other companies that exist or succeed only because of this phenomena.

    Whether Chris over-reached or not, Lee’s analysis fails entirely since it’s entirely based on a false assumption.

  5. jjohncm

    If this discussion can be kept in the realm of ‘real’ inventory and ‘virtual’ inventory. Lee Gomes concept fails. Why? Because if you keep iterating by adding more items you get to a point where there will not be enough shelf space to hold items making up 50% of the sales. The .01% of the items making 50% of the sales might number millions.

  6. Hal O'Brien

    “…when you attend Mainstream Media Community College, you are taught that if you say something in your lede, you need to back it up.”

    Actually, when you attend Mainstream Media Community College, you are taught to spell “lead” correctly, and to have faith your copyeditors will pick up any possible confusion. After all… they’re your copyeditors.

    This alone is a big enough mistake as to make me mistrust anything else Gomes has to say for himself. It’s ungrammatical and unprofessional. What a shock he works for the Journal {not}.

    Here’s a hint, Lee: If “lead” is good enough for John Chancellor and Wally Mears, it’s good enough for you.

  7. ted

    Yeah, that’s helpful — slam the WSJ — online since 1996, not enough cred for ya? or maybe the dozens of Pulitzers?

    If you knew anything about the history of newspapers (ya know, ‘journalism’ — not pajama blogging) and industry jargon you’d KNOW that …

    The first paragraph of a news story is the lede. It is often just one sentence, but it can be two.

    * (Lede is often spelled lead. The odd spelling was adopted when newspaper type was set in lead. To keep from confusing the word for the metal with the word for the beginning sentence, the spelling “lede” was used.)

  8. Michael D. Smith

    I am one of the authors of the MIT study (with Erik Brynjolfsson and Jeffrey Hu). I had a couple of quick comments:

    1. Tubby, I appreciate your citing our research on this subject. In our study (available at http://ssrn.com/abstract=400940, see page 18) we found that 39.2% of Amazon’s sales were in books that have ranks higher than 100,000 (the 60% you mentioned is the head not the tail), and 29.3% were in ranks higher than 250,000.

    2. Glenn, I like your focus on profit as the interesting question. You are right that bestsellers are typically sold at a discount and more obscure books at list price. We found similar results in our paper (page 21). If you assume that our numbers are right for the top and bottom 100,000 titles, and further assume that the wholesale price of books is around 47% off list and is constant across titles, then you would get that a little over half of Amazon’s profits (51.9%) are in the long tail.

    3. One final observation: Another way of thinking about this issue is how much consumers benefit from long tail products. This was the main focus of our paper — and we found that in 2000 consumers gained around $1 billion per year in consumer surplus, around 10 times more than they gain from lower prices available for book online. So maybe the real story here is not in the percentages, but in how much you and I benefit from having easy access to these products that would be hard to find in a purely brick-and-mortar world.

Comments are closed.