Dynamic pricing is a buzzphrase making the rounds of publishers at the moment as they begin to get their arms around the opportunities inherent in the agency model. They are aware that Amazon does a lot of pricing automated by algorithms and some of the more creative and tech-minded thinkers are wondering if publishers need to employ technology in a similar way.
The most widespread applications of dynamic, or perhaps we would be clearer were we to say “automated” pricing, are in the airline and hotel industries, pricing seats and rooms. But airline seats and hotel rooms share three characteristics that ebooks don’t have.
One is that there is a physical limit to how many can be sold. Once you’ve sold out the plane or the hotel, you have nothing left to sell.
The second is that the airline or hotel actually pays for the seat or the room whether or not it is sold. There is a relatively small per-use cost: a free Diet Coke for the filled airplane seat and the cost of laundering the sheets in a used hotel room. But the airplane still has to fly and the hotel still has to pay its mortgage and bellmen regardless of the number of seats or rooms that are sold.
And the third is that there’s an effective deadline for each sale. Once morning comes, you can’t sell the hotel room that was vacant last night. And once the plane takes off, the unsold seats will remain forever unsold.
That creates the environment in which dynamic pricing is applied in those businesses. As the deadline for a hotel night or a flight approaches, calculations can tell the hotel or airline whether it makes sense to raise prices (because rooms in that town or flights on that day to that destination are scarce) or lower them (because the consumer has many choices and only lower prices will be competitive and, under the circumstances, anything you can get is incremental profit.)
Dynamic pricing in those businesses is about how to maximize the revenue from a fixed sales opportunity. Ebooks are entirely different. There is no limit to the number that can be sold. There is no deadline beyond which a cost is incurred whether or not a sale is made. And the competitive set — the equivalent of checking the availability of rooms in comparable hotels or available seats on possible substitute flights — is not self-evident.
If dynamic pricing for revenue maximization for hotel rooms or airline seats is a precise science, dynamic pricing for revenue maximization for ebooks is more like alchemy or an art.
That doesn’t mean it isn’t a good idea. But it does mean that it’s a more complicated problem than the hotel or airline industry faces and it probably means that the logic and techniques they’ve used to solve it for themselves won’t apply much to what we need in the ebook business.
The challenge and opportunity is to use data to adjust prices automatically rather than by human scanning of information and manual application of intuition.
For example, our client Dan Lubart of iobyte mapped out one interesting case that calls for further exploration that might best be done by automated pricing decisions. What Dan found was an ebook that had been selling for $12.99 and was on a very gradual, but consistent, “decay” curve. (“Decay” means that sales were going down.) The publisher put a $8.99 price on the book for a few weeks and sales shot up and stayed at a consistenty higher level as long as the promotional price was offered.
But, then, when the publisher went back to “normal” pricing, they went to $14.99, rather than where they’d been at $12.99. What was interesting was that the sales dropped back down to meet the old decay curve and continued to erode along the previous path.
This raised two questions. One is whether the publisher might have been better off to stick with the promotional price longer, since sales were sustaining at a higher rate with that price. But the other is whether the book benefited at all from being $12.99 for a while rather than $14.99. The fact that the decay curve wasn’t affected by the price increase certainly suggests that the higher price might have been beneficial all along.
Because many books are similar but no two books are the same, it would require a substantial number of experiments to yield persuasive data about either of these two points. Only by testing 20 books or 30 books with similar price-and-decay profiles would a trend be both discernible and convincing. The only practical way to find the candidates for testing and to apply the tests would be by generating rules to do the price changes dynamically. And then if the publisher learned there were patterns that repeated, only by applying rules algorithmically could they benefit from that knowledge.
The number of variables is vast. There are different effects at different internet retailers. No doubt the pricing of competitive books has an impact, but even identifying the competitive books isn’t simple, let alone tracking them. (As far as we can see, tracking the pricing of competitors is a bridge or two further than anybody’s been able to go so far although we know that some are thinking about how to do it.)
In fact, the complexity of the ebook market makes using dynamic pricing techniques — creating rules about how prices should be reset based on data in the marketplace — potentially even more valuable than it is for hotels or airlines. It certainly could represent serious competitive advantage for those publishers who do it best and do it soonest.
Smart agents will be watching this carefully. They’re all aware that the ebook royalty they collect is the percentage they work so hard to negotiate times the revenue the publisher receives. Now the revenue is clearly affected by the publisher’s ability to set prices that capture the most possible consumer dollars. It will become clearer over time that changing those prices intelligently is necessary to maximize the revenue which maximizes the royalty. The publishers who figure this out first and best will have a powerful argument that publishing with them puts extra coin in their authors’ pockets.
Nobody is going to be making that argument persuasively in the next few months; we have far too much to learn and we’re still in a world where things keep changing as the switchover from print to digital is still in relatively early stages. But the publishers who get a head start developing this understanding and the tools to implement it will have something very powerful to talk about in a year or two.
At our London conference, one publisher speculated that the ebooks being priced as they are were pulling paperback sales forward when the hardcover book was out. What that would mean is that at the end of Year 1, the hardcover year, the print plus digital sale and margin would look pretty good. But if the theory that pb sales came forward is right, then at the end of the second year — the trade paperback year — there will be disappointment. This is a reasonable theory but it is very hard to know yet whether it is true. You’d need a 2-year old book to be begin to measure it, and a book that came out in hardcover two years ago was in a different world than a book that comes out today. That’s why nobody really knows much for sure yet.