The Shatzkin Files

Dynamic pricing: what it is and what it isn’t

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Share on RedditShare on StumbleUponEmail this to someone

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.

Tweet about this on TwitterShare on LinkedInShare on FacebookShare on Google+Share on RedditShare on StumbleUponEmail this to someone

  Back to blog

  • Chris

    Dan’s the man!

    I would really have liked to see Flynn’s book sit at the $8.99 mark for a bit longer… actually for as long as it took to map a steady decline.

    • I agree with you. I suspect this was a promotion for a specific purpose.
      Also, Dan and I believe that the best way to do price promotion is to
      announce it loud and make it time-limited. I don’t think they did that in
      this case, so they would have had the option to keep the $8.99 price going
      as long as they wanted.


      • Chris

        Yeah, I’ll go with you lads on the ‘loud ticking clock’ thing. A price drop is pointless any other way.

        I wonder if you an upload a cover image to Kindle with a sales burst graphic? I’ve been meaning to try this for a while…

      • If you do try it, let us know how it goes!


  • Jenny R.

    The impact of Amazon’s lists makes it very tough to come up with an algorithm to adjust pricing because position on a commonly searched list plays a huge part in sales volume, but some lists get searched and others don’t, and Amazon (unlike Google) doesn’t share with the public information about the searches that bring people to a book’s page or how commonly a given search is executed.  

    In my experience, once you are at the top of a commonly searched bestseller list even a price different of fifty cents can make a significant difference in sales volume. People who are trying to choose between two books that strike them as fairly similar often use price as the tie breaker.

    But once you aren’t on a bestseller list people who get to your page already have some kind of motivator so the sales decision is based more on whatever brought them to the book page, which again Amazon won’t show you.  Without information about where your buyers come from into the online sales site, it is extremely hard to make intelligent decisions about pricing. 

    • Jenny, what Dan has been looking at is how price affects rank. Rank is only
      a rough proxy for sales, but it is still helpful. What he’s seen is that
      bestseller lists boost rank, and that books can get to the list at one price
      and stay on the list even if the price is raised. That would imply that one
      group “discovers and buys” because the book is 99 cents, and another group
      “discovers and buys” because the book is on the bestseller list.

      There is no doubt that the retailer — Amazon being the largest one — knows
      much more precisely than the publisher can about customer behavior. For one
      thing, we have this theory that the 99 cent buyers are different people than
      the brand-name book buyers, but that’s just a theory for us. Amazon (and B&N
      and Kobo, etc.) actually *know*.


      • Jenny R


        I didn’t realize it wasn’t already known that books can persist on a list after the price is raised once price manipulation gets them that list position. That’s been the strategy that boosted a lot of books onto the Kindle top 100. 

        Re the bargain hungers, there are a whole lot of online sites and e-newsletters that notify subscribers about free books and  $.99 books. They seem to be driving a lot of those books up the charts. 

        What advantage does Amazon get from keeping from publishers where their book’s traffic is coming from? You’d think it would be in their interest to give publishers the data that would help them optimize sales. A “Google Analytics” for Amazon would be worth its weight in gold, and possibly something Amazon could charge for.

      • Jenny R.

        bargain hunters. Sorry!

      • Jenny, the price gaming is “known”, but not “universally known.”

        Amazon protects information about how site traffic behaves partly to keep it
        from their competition (like B&N.) I am not privy to their thinking about
        why any particular piece of information is protected but I think they just
        assume, if in doubt, let’s keep it to ourselves.


  • Pingback: Ebook Pricing: Free, Fee, or Wait and See – Plus a Giveaway~ | Helen Hanson()

  • Stanislav Mamonov

    We are in the process of developing MintRight Analytics for publishers that would help track and optimize pricing decisions.   The initial insight that we have suggests that there are many wrinkles to the optimal pricing question and the big six appear to be evaluating multiple pricing strategies –  agency publishers have imprints that are still on wholesale.

    We shall share the lessons as we go forward.

    • I wasn’t aware that agency publishers had bits of their list still on
      wholesale. If that’s true, it is new information.

      It will take a lot of experimentation and data to get to real knowledge
      here. If you can share useful data, everybody will appreciate it.


  • Pingback: Dynamic Pricing for Ebooks | The Passive Voice()

  • Davidsbooks

    Speaking of high pricing and declining sales; buy a text for $150 and sell it back for $5 –

    • Another topic entirely. But true enough.


      • Davidsbooks

        Didn’t mean to hijack, but the topic interests me as texts seemed to be about the last area of print that showed any signs of life, and that only because it was a captive market. Seems they should just issue every student a laptop (why almost all already have), then just license an e-text for the term at a reasonable rental. (Very few ever look at the book once the class is over.)

      • Students have been notoriously resistant to ebooks for texts. There are a
        number of initiatives aimed at overcoming that resistance and, ultimately,
        professors picking what to adopt and what to require will decide.

        But this really is another subject entirely.


  • Yoav Lorch

    Try to think of it from the reader’s
    point of view. He just wants to buy a book. He’s not sure which, and as you
    say, when still unread many books seem similar. So now he has to throw into the
    mix of decision factors the dynamic pricing curves of each publisher. Should he
    wait a little and possibly buy the book he somewhat wants for somewhat less?
    Should he pounce now on a book he has stopped considering earlier just because
    the price has dropped and may rise again next week ?

    What may be a boon for publishers is a
    complex burden for readers.

    must be a better way.

    • Yoav, I don’t think you have any appreciation for how many times a day the
      book retailers are changing prices *now*. I have no inside information about
      Amazon, but an executive at The Book Depository, a competitor which Amazon
      is in the process of buying, told me they change prices *four million times
      a day.* I am sure that Amazon does pretty much the same, all automated
      responses to conditions detected by their own sales and bots looking at
      other people’s pricing.

      Prices change all the time on the web: for hotel rooms and airline tickets
      and rental cars and everything else. Many sites (including Book Depository)
      have features that will alert you if an item hits any particular price. It
      is my contention, and your note is evidence that I’m right, that most
      consumers don’t notice this. When they shop for something, they’re not
      particularly aware of what the site they’re looking at charged for it
      yesterday, last week, or an hour ago. Price changing is part of our
      landscape. Publishers pricing dynamically would be joining the crowd, not
      presenting an outlier problem.


  • Pingback: Industry News-July 17 | RWA-WF()

  • Georges Lacombe

    One thing that the publishing owner could do is sell books in parts, I could buy only the first chapter. Probably with dynamic pricing each part could have a different price, considering that the person is already engaged in the book. The last chapter of a book can be more expensive. However people will discover that few books are completely read. On the other hand the authors should received their royalties accordingly with how much was sold, and they would start writing to keep readers buying the next chapter…

    wild thoughts

    • Some books “chunk” and some don’t. I would be skeptical that selling
      narrative books like fiction in pieces would be particularly well received.
      For others, I think chunking makes more sense when some of them “stand