The Long Fail of Books

Andrew Odewahn from our Research group recently went to the Bookscan book summit. One of the presentations was on the state of the overall book market, and had this factoid: 93% of all ISBN’s sold fewer than 1,000 units and accounted for 13% of all sales. It’s tempting to think of that as “93% of all books are failures”, but that’s not the full story. This being the Third Age of Capitalism and all, success and failure aren’t measured by unit sales. They’re measured by greenbacks, baby: wampum, moolah, shekels, Benjamin Franklins. Cold hard cash (or warm soft cash, so long as it’s cash).  Unit sales figures are only part of the equation: cover price and discount to the bookstore are the rest.

Those of you who haven’t dealt with the vagaries of the publishing world before may not know that all books aren’t created equal. Compare Perl Cookbook and Information Visualization: Perception for Design. The former has a cover price of $49.95, but you can buy it for $32.97 on Amazon. The latter has a cover price of $59.95 and you can buy it for $59.95 on Amazon. The difference is because general computer books like Perl Cookbook and textbooks like Information Visualization are sold to booksellers at different discounts. Big distributors like Amazon, Barnes and Noble, Borders, etc. get the general books at around half-price but textbooks at about 80% of the cover price. The rationale is that publishers will make up the bigger discount on bigger volumes. Very rare are the textbooks that sell like Robert Ludlum mindcandy.

So without the discount and cover price information, you can’t figure out whether those sub-1,000 books were really failures. Andrew (remember him, he’s the researcher who went to the conference and heard the 93% titles 13% sales rule) looked at the numbers we have on the tech book market and discovered that for tech books it’s 85% of titles are 10% of sales. The numbers are a little hard to comprehend, so let’s turn it around: overall, bestsellers are 7% of the titles but 87% of sales. In the tech industry, 15% of the titles make 90% of sales.

If you drew the long tail curves for the two markets, you’d find that the overall industry drops off quicker than the tech industry. This probably reflects the broken fiction publishing model: flood the market with different products, publicise poorly, and let the market find the 5% it wants to elevate to bestseller status. In computer publishing (and in O’Reilly in particular) we publicise and market our books. They won’t be found in airport bookstores and supermarket checkout racks and the category systems in bookstores rarely reflect the way technical people think about computer books, so if you want nostalgic gamers to buy your new book, you have to reach out to them.  This was another conclusion from the conference: “quality content for ever-smaller markets”.

Long-tail pricing is a problem for the publisher. A lot of long tail analysis has focused on the retailer: with infinite shelf space, every niche product can become a bestseller. That’s fine for the retailer, but cornholes the publisher (where short print runs are financially murderous), the creator (where low royalties discourage production), and the buyer (because to offset all the pains of short sales, the price is likely to be much higher as with textbooks). You end up with a multi-tiered price structure, so that the 1,000 people who want Perl and the Poetry of Thomas Wordsworth in a Post-Keynesian Economic Dialectic pay as much collectively as the 20,000 who want Crack Hacks.

This is different from what the music industry is trying to do as they pressure Steve Jobs to institute multi-tiered pricing on the iTunes Music Store. They’re attempting to gouge on the bestsellers (“new releases would cost more than 99 cents”) and under-price their back catalog (“oldies as little as 60 cents and recent hits somewhere in between”) to juice demand. This is the delight of virtual goods–with no warehousing and reproduction costs, products can have an infinite shelf life. Classic Chris Anderson economics.  But can you imagine a world where the publishing industry adopted the music industry pricing model? The key to the music business is time: songs are must-have for a brief period, then they become recent hits and ultimately golden oldies. New books would all cost $60 and old books all cost $20. You’d have Windows XP for Dummies (current top tech bestseller) retailing for $80 a copy, and Mythical Man Month retailing for $9 a copy (they’re $21.99 and $34.99 respectively).
This sounds workable until you think about inventory: there are costs to printing and storing huge piles of books, and the economics of production are in opposition to the economics of popularity. The less popular something is, the more it costs in printing and storage. Remainders make old books cheap, as the music industry would do for songs, but by the time a book is remaindered the author and publisher have been cut out of the economics. Every remaindered book represents a publisher taking a loss.

The other difference between publishing and music is competition. The music industry creates stars with extravagant marketing and creates demand for a unique product. If we tried to sell Programming Java for $80, you can bet someone else would sell a similar Java book for $60. In this world, unique titles like Programming Perl (by the author of the language!) and Head First Java (the knowledge goes STRAIGHT INTO YOUR BRANE) would sell for $80 while books without those distinguishing features would sell for less, in lower numbers. You’d get lower rewards as author and publisher for doing a middle-of-the-road book, and consequently there would be fewer books on each topic in the market.

These are the questions the tech publishing industry faces as book sales slowly establish a new (lower) equilibrium after the tech bust of the early 2000s: Do we make blogs our long tail of tech publishing and only do bestsellers? How do we balance inventory and demand? Do we price for balance or for opportunity? It looks like Chris will address these questions in his book, as he already has a very interesting piece on Lego balancing physical goods and electronic sales. That’s good–there are a lot of industries trying to figure out how to keep the Long Tail from becoming the Long Fail and we need all the help we can get.