The power of advertising?

Warning: this is a self-indulging post. (I can already hear the retort: "you mean all your others weren't...?")

Princeton University Press ran a small ad for my book last Sunday in the New York Times book review. I was curious if it would have any effect on sales, so I ran a little experiment.  I checked the book's sales ranking in amazon.com at periodic intervals starting on Saturday afternoon. I also took down the corresponding information from three amazon sites abroad: amazon.co.uk, amazon.de, and amazon.fr.  The idea was that an ad in the NYT book review should not affect sales in Britain, Germany, or France, as the NYT (especially its Sunday supplements) is not easily available in those markets.  The comparison also allows me to control for day-of-week effects: "serious" books may experience higher sales during weekdays than on the weekend.  In other words, the research design was a simple diff-in-diff.

Here is where things stood, as of this morning:

(Note a couple of things about the chart: I use the log scale for the vertical axis, and a higher number means poorer sales.)

So what did I learn (except for the generally abysmal ranking of the book: the average rank over all four markets is around 20,000 among all English-language books...)? 

First, I was surprised that there was not a more perceptible impact on U.S. sales.  The sales rank at amazon.com dropped from around 15,000 on Sunday midday to around 7,000 this morning. I suppose I exaggerated how many people read the NYT.  (Harvey Mansfield once defined a liberal as someone who wouldn't have anything to do if the NYT did not come out on Sundays.  I guess there are fewer liberals than I thought.) Now, as it turns out, amazon.com ran out of the book midway during the experiment. Can I comfort myself (and my publisher) by thinking that this had something to do with the result? 

Second, I was puzzled by the unexplained and very steep rise of the book in the German site.  On amazon.de the book went from a sales rank of around 27,000 on Saturday evening to a rank below 1,000 this morning.  I have no idea what may have caused this.  I doubt it was the ad in the book review.  So I am clueless.

Third, why is the sales rank in France so stable compared to the three other countries?  What makes the French so consistent in their purchases of English-language books over time?

Bottom line: I was expecting to get some easy results, instead I ended up with lots of questions.    

Comments

[RWB]

I think you make a mistake in using Amazon rankings, which are not very transparent and take into account the sales of other books compared to yours.

You should ask your publisher to provide you with Nielson BookScan sell through numbers, which track actual sales of actual books through most book retailers in the U.S. I'm sure Princeton University Press has access to these numbers either directly or through their distributor.

In my experience, positive reviews lead to almost instantaneous spikes in sell-through, but ads less so. Still, the NYT Book Review is a great platform. Se if you can get several weeks work of BookScan data and see what you discover. Good luck!

[greatzamfir]

I completely agree with the previous poster. Also, what absolute amounts are we talking about? To make a 4-hour resolution meaningful you would have to be selling something like multiple books per hour, or thousands per month through Amazon alone.

Perhaps that's true for America, but I doubt it is the case for European Amazon departments. My suggestion: the two or three changes in your German site ranking correspond to two or three sales of your book, which are then 'smoothed out' by Amazon's algorithm. And during this period you did not sell a book in France.

Keep in mind that with the current dollar value, it is attractive for Europeans to buy English books from Amazon.com. If you're lucky, they even send the book from their European departments, giving you dollar price and euro shipping speed.

[suitable]

Re: Amazon.fr. They don't sell many books in English, and they don't sell your book themselves, they just provide a list of vendors. I suspect they calculate their ranks of books in English only very rarely to have something "significant."

[John V]

What about absolute sale numbers?

It's entirely possible that your total sales would move more than your sales rank would indicate if other books experienced high sales at the same time.

[matt Lister]

On Amazon sales, the sale of just a very few books in a day can change the numbers by quite a lot. So, it's hard to know what they mean. But, a fairly regular ranking of around 15-20K is _very_ respectable for an academic book from an academic press- you'll find very few that do better than that for any period of time. Most are in the hundred thousands or even millions.

[sbmmth]

dif-dif has good properties asymptotically where normality is fine. with such a sample, you have a lot of fat tails and extreme problems. you just get noise.

[Xavier]

I am French and I bought your book on Amazon.com a week ago, so you have to know that French people do not necessarily buy their books on Amazon.com. When I am looking for a book in english, I prefer to go on Amazon.com where there is a larger choice of books.

[Christian]

There was a (in general positive) review of the book in the NZZ last week...that might explain higher sales in German speaking countries resp. via amazon.de...although the comments by the others regarding the sale statistics are probably more important.

[Reginald]

How come there are still no reviews for the book on Amazon.com even though it came out last September? From my limited observations, books with several detailed reviews tend to sell better than books with no reviews (or books that only have a few brief reviews). Maybe you can make some grad students read it and post reviews :)

[mrs juliet]

Do you wish to have your own business/company. Debt free and off capitals
Mrs juliet wee Now Offer Loan at low rate, We are resisted certified lender approved by the government .
more info contact us via e_mail julietloan.offer01@yahoo.com