Quietly I’ve been watching the e-book pricing debate and the feuds between Amazon and many publishers.  I’ve seen quite a few interesting articles on the price of e-books from heavy hitters such as Cory Doctorow (via podcasts) and articles such as this.  There are tons of other articles I’ve read lately and all of them seem to be offering perspectives but very few of them (if any) provided an actual step by step solution that publishers can follow to solve this crazy complex question of HOW TO PRICE AN E-BOOK.

The solution is simple – Dutch Auctions. (wikipedia entry)

For those that are unfamiliar with a dutch auction and don’t feel like reading here’s a short summary

1) The seller sets a quantity of an item to sell, the description of the item, and a time to keep the auction open
2) Bidders get to select a price and quantity they wish to buy at.  For example, they can say they are willing to pay $50 each for 10 units.  Their bid is a contract and should they be a winner, they must pay and will receive the item.  (The second part is key to getting people to be honest in their perceived value).
3) When the auction is over, starting with the highest price, quantities are matched until the quantity is met and that is the price that the top bidders get.  (Ok, this is a mouthful, see the chart below)

EXAMPLE:
Stated Quantity: 80
Bids:

In this case,the quantity level would be met at the price of $50 (10 + 20 + 50), so the 80 people who bid $50+ would be sold the item at $50.

Got it?  Good.  Cause now comes the fun part.

If you have ever taken econ 101 or had to sit into a business meeting with a CFO, you’ve seen this before.  Its the supply/demand curve.  Most of the articles that have talked about e-book sales have mentioned that publishers need to figure out the supply and demand curve for e-books so they can solve the pricing problem.  Well, I’m about to explain how dutch auctions will solve that problem, and how to make the data work.

The Information a Dutch Auction Provides
1) The number of people willing to pay a given price for e-books
2) The distribution of people at a given price
3) How scarcity effects price

To keep this discussion short(er) and focused, I will NOT touch upon hardcover/paperback releases or pricing.

How Does a Dutch Auction Solve This Problem?
If you look at the chart above with bid quantity and price, as you see, your quantity increases as your price decreases.  In actuality, there will be a cutoff where a reduction in price will not yield a significant gain in quantity.  To save on time and screen space, I will use the above table in the rest of these examples.  From the data, we’ve learned that we have a demand of 11,180 from the sample size of people who knew of the auction.  Based on this data, if we know 10% of our addressable market saw the auction, then we have potential sales of 111,800 (assuming that we sell it at $15).  We also see that we have a potential of 11,800 if we sell it at $20.  With this, we can actually plot out that supply/demand curve.  (I will address data quality issues later).  I’ve addressed facts #1 and #2 above.  As for scarcity, based on the quantity defined in the dutch auction, you will have different price distributions (which can help address limited editions)

How Can This Be Done With Statistical Accuracy?
I have a reasonable schooling in mathematics and have held quite a few positions being “the stats guy” so I understand the importance (and irrelevance) of limited/bad data.  Below are the steps to gather enough data so that the information gathered will be statistically relevant and will provide a reasonable cross section of information.

Providing Diversity Within Data:
1) For all of the following lists, they will need to be repeated for the following groups (Suggested min 5 authors per group):
a) Mass market (high recognition) authors such as James Patterson, Janet Evanovich, Stephanie Meyer
b) Niche market (high recognition within a niche) authors such as Neil Gaiman
c) Mid-market (medium recognition) authors – this would be a multi-published author with some popularity
d) Debut authors.  For best results, ignore previously well known figures
e) Celebrities.  It seems their putting out books in greater frequency and you can’t ignore them.

How To Collect The Data:
There are a few rules that need to be followed for any of this to work.  A failure to follow any of these rules will compromise the data and provide less accurate results
1) There needs to be a limited quantity that will be provided during the dutch auction.  This quantity should be less than the total demand (60% or so).  Prior sales can be used as in indicator.
2) This must be the only channel in which the e-book can be purchased.  There must be at least 3-4 months delay before the e-book will be available outside of this dutch auction.
3) No pricing data should be available in the auction description.  Bidders must not be given starting points for comparison with such immediacy.
4) Traffic must be driven to the auction site.  Quantity being sold during the auction should be driven by the expected traffic that will make it to the auction.
5) The auctions should be run during the month or two leading up to the release of the book and should end on the day the book is released (so that winning bidders can receive their e-book on the day of release)
6) Bid quantity must be fixed at 1 unit and bidders only allowed one bid.  This will keep out people trying to fix results or bulk buyers.  The goal is to find out what consumer demand is.

Summary:
1) Publishers should run 20-25 dutch-auctions to gather data on the supply/demand curve of books
2) Publishers should control the supply of those e-books to get an accurate idea of demand distribution
3) After data is collected, a reasonable idea of what specific types of books with different marketing spends/author recognition look like will be gained

I realize this is a blog post, and I’m speaking to a general audience, but I will be available via comments and e-mail (nruffilo@bookswim.com) if you have further questions about this.

-Nick