The Long Tail

Mainstream Media Meltdown II

On the occasion of today's gruesome statistics on the continuing fall of newspapers, here's an updated look at mainstream entertainment and media in decline (April's version is here).

Down:

• Box Office: down by 7% this year (tickets per capita have fallen every year since 2001). 
• Newspapers: circulation, which peaked in 1987, is declining faster than ever and is down another 2.6% so far this year. 
• Music: Sales are down another 5.7% this year; although digital downloads (still just 6% of the business) are climbing nicely. 
• Radio: down 4% this year alone, continuing a multi-decade decline.
• Books: down by 7% in 2004 (but see comments  below for discussion)

Mixed:

• DVDs: sales growth is slowing dramatically, from 29% last year to single digits this year.
• TV: Total viewership is still rising, but as channels proliferate and the audience fragments the rating of the average show continues to decline.
• Magazines: Ad revenues are up a bit although the number of ad pages is flat (they're charging more per page). Circulation is also flat, while newsstand sales are at an all-time low. 
• Videogames: it's the final few months of the current generation of consoles, which tends to the trough of the buying cycle. Sales were down 20% in Sept, but will probably pick up by Christmas with the launch of the Xbox 360.

Up:

• Internet advertising:
  --Banners: Up 10% this year
  --Keywords: Google revenues up 96%

Sizing the book tail

Since my last post on estimating the size of the Long Tail, there's been some more useful discussion of the book tail in particular. Books are one of the toughest industries to analyze, in part because the industry is so archaic and good data is so hard to come by. But it is in sclerotic industries where the Long Tail potential is often largest, so it's worth rolling up the sleeves and digging a bit further into what we know.

The two big statistical questions are:

1) How long is the book tail?
2) How big is it (as a proportion of total book sales), or might it become?

Let's start with the first. Nat Torkington at O'Reilly has an interesting post on the economics of the book business that includes some some stats about the head-heavy nature of the book industry (the comments on that post are worth reading, too; especially Peter K's). Here's the important one:

93% of all ISBN's sold fewer than 1,000 units and accounted for 13% of all sales. (By comparison, in the tech book market 85% of titles make up 10% of sales.)

So does that mean that the book tail is only 13% of the business? Well, we generally define the dividing line between head and tail as the average inventory of the largest offline distributor in a market. In this case, that's a large Barnes and Noble superstore, which carries about 100,000 books. According to Nielsen Bookscan, 1.2m titles sold at least one copy last year, so that 100,000 is about 8% of the total books that are actually in the market. This means that 92% of the books in the market aren't in a B&N superstore, which also happens to correspond to the number of book that sell fewer than 1,000 units (above).

If that were the end of the story, I'd be comfortable sizing the Tail of books at 13% of the business. But it's not. The data above comes from Nielsen Bookscan, which only tracks new book sales. Yet used books represent 8.4% of the overall business. Most of that is still textbooks, but the online side, which is heavily focused on the Long Tail (especially out-of-print titles), is the fastest growing part and now represents more than a quarter of the business. At current growth rates, that will be a half in a few years.

Including used books, my best guess of the business right now is sales of titles not available in a B&N superstore represent about 15%-16% of the total business--the 13% of new sales, plus another 2%-3% for used sales. (That's for the industry as a whole; for online retailers such as Amazon, it's higher--between 20% and 30%--as previously discussed.)

But that's going to rise, as more previously unavailable books become available again. As Tim O'Reilly notes in this fascinating post, the number of out-of-print titles is huge. There are 32m unique books out there, of which only 4% are in print. The picture looks like this:

Until recently, when a book went out of print, it was essentially unavailable. But now, between all those now-online used bookstores, print-on-demand and the book-scanning efforts of Google, Yahoo, Amazon, Microsoft, the Internet Archive and others, the concept of letting a book disappear is itself going away. Out of print is out of style. And those older books that are now becoming available again will make up a growing part of the tail's volume.

(BTW, the robots that do the scanning are very cool. Check out the video here)

So now we have the beginnings of an answer to our two questions:

Q: How long is the book tail?
A: About 31,900,000 titles long.

Q: How big is it (as a proportion of total book sales), or might it become?
A: Today: about 15%. Tomorrow: more.

Becoming a Google Halloween costume: priceless

This totally made my day. Google's AdSense team dressed up as a 300-foot Long Tail for Halloween. And Esther Dyson posted the picture:

Long Tail Camp

Foo camp was great. I hear Bar camp was, too. Now, in the same vein, comes...

Long Tail Camp.

Long Tail Camp is an open, welcoming event for geeks to camp out overnight, get wired on caffienated beverages and think really fast about the long tail, its applications, and implications. Nov 11th, 2005.

Nothing to do with me (indeed, I'm not entirely sure it's serious). But if someone knows more and would like me to update this post with details, email me.

UPDATE: Eran Globen writes with more:

My name's Eran Globen, I'm one of the originators and organizers of Long-Tail camp. First to clear out any misunderstandings, Long-Tail camp is on the cusp between serious and parody. It was first announced on our blog, Supr.c.ilio.us: The Blog, in this post.

Since then we've received some support from the community and it seems that there is indeed some interest to at least have some fun with the idea. We will be having the first meeting of Long-Tail camp in our upcoming launch party at midnite on November 11 (venue undecided yet) and when that meeting is over we intend to release the idea out to the world and let everyone do with it as they will. We already have interest in having Long-Tail camps in SF, NY state and Australia.

The wiki is meant to inspire, help organize and serve as a location to aggregate the experiences from various camps.I think that together, many small camps can make just as much of a difference and be just as much fun (or more) as one big camp.

Thank you for your interest and for creating this great meme.

Eran Globen.
http://supr.c.ilio.us/blog/
http://hellonline.com/blog/

How big could the Long Tail get?

Over at Mediabistro, there's been some interesting discussion on the size of the Long Tail in books that deserves some comment here..

In response to a previous Mediabistro article about the LT, a reader wrote:

Even if half of the books sold on Amazon are long tail and Amazon is just 10% of all book sales, they make up only 5 percent of a book publisher's actual sales, suggesting that the long tail isn't a vast and untapped marketplace, but simply the 5 percent tail. Am I missing something?

It's a fair question, especially since our subsequent analysis suggests that it's closer to a quarter of Amazon's sales. So does that mean that the LT is just 2.5% of the book industry?

Richard Nash, another Mediabistro reader answers some of the question for me:

"Overall the Amazon market share numbers are correct. And they are also correct for Soft Skull in aggregate. But the interesting thing is that in Year Two of a given title's life, Amazon could be responsible for as much as two-thirds of a book's sales. Now those numbers in unit terms could be too small to make it worth a big publisher's time to keep it in print, but for an indie, or a university press, having, say, half your in-print titles moving 50 units a year (in our case that would be for 100 titles) that's about an extra $25K/year...and close to pure margin.

"The appeal of the Long Tail in the world of music is that digital downloads have close to zero marginal cost: They're digital copies! Books of course have a relatively high marginal cost; each additional book could be a dollar, say. But in practice, what we do is make a guess about demand, and print, say, 3,000 books. After about a year, that book has basically had its day in brick and mortar stores and, I would argue, the value of the inventory at that point should be written down to about zero. The books are worthless.

"But in fact they do keep selling! Maybe only 50 units a year, as I said, but it is pure marginal revenue! And, for a publisher that has $1 million in sales, and profit margins of 5 percent, an extra $25K in revenue on which the marginal cost is basically zero (the inventory would otherwise have been pulped), you've increased profitability from $50 to $75K!

"Now that's the short-term benefit from the Long Tail. And there are limits: For 50 units a year, it would be impossible to justify a reprint (unless the book were short enough that you could get $2/book unit cost through short-run digital printing...). The long-term benefit would be when eBooks become viable through advances in reader technology, and the marginal cost of a book truly becomes zero."

Richard's right, but there's more. For certain categories, such as documentaries, a Long Tail retailer such as Netflix can represent a huge proportion of the market. Netflix alone accounted for half of the US business for Capturing the Friedmans, for instance. And the used book market, which is far more Long Tail than new books, now accounts for 8.4% of all book sales.

Meanwhile, in music (or at least for the paid part): digital music sales represent 6% of the business, and the LT represents about 30% of that. It's anyone's guess what's going on in the file-trading world, but some BigChampagne stats we've seen suggest that it's far more LT than the commercial world.

Some industries will be more online than others. I suspect that media and entertainment will all be online in the not-too-distant future, and the head and tail will be in rough balance there. But food and perishable goods will stay largely offline, and the tail will remain tiny in that domain.

Let's stand back and look at the big picture. Jeff Bezos has often said that he thinks that online retail (which is the best, but not the only example of the "infinite shelf space" markets where the LT flourishes) will probably stabilize at 10%-15% of all US retail. Our own calculations, meanwhile, suggest that the LT is likely to account for no more than 50% of that (it's now between a quarter and a third for most such retailers, a percentage that's growing by a few percent each year as inventory expands, filters improve and buyer behavior adapts). For now, let's be conservative and call it a third.

If so, a back-of-the-envelope sizing calculation would put the potential size of the LT at 33% * 10% or 3.3% of all US retail. If that sounds insignificant, recall that annual US retail sales are $3.9 trillion (about 70% of the US economy). That would still make the LT a $125 billion opportunity.

The lesson: A small percentage of a big number is still a big number.