The uniqueness of our approach is to produce a consolidated ranking across all of the lists from the three publications. Each publisher uses its own sample to build its reports, so while each bases its rankings on sales volume, the difference in sources, methodology and list structure can produce very different results in the published rankings. Our consolidated ranking is intended to level those differences.
We currently track 23 individual lists each week, with a total of 600 entries
(Up from 14 lists with 410 entries when we first started tracking the lists, but
down from the peak of 680 items from 24 lists in 2012 and early 2013 )
Each list has its own unique structure (last updated May, 2013):
Report Date—In order to be consistent, our analysis is based on the week for which sales are reported, through Saturday or Sunday. We use Sunday as the "report date." These are updated on the list websites by the end of the following week, with our update normally online by the following Monday. (The print publication date ranges from a few days after the sales report for USA, 8 days for PW, and 14 days for NYT).
ISBN is the "International Standard Book Number" assigned to individual editions of a title. The ISBN agency for the United States is R. R. Bowker, the publishers of Books in Print. PW shows the ISBN, which sometimes helps us determine which edition to reference on NYT and USA. For example, it has not been unusual for trade paper and/or mass market editions of a title to appear as a movie of a book approaches, replaced at some point by movie tie-in editions in the same format.
Tracking e-books can be difficult because of multiple formats and sources. We currently track e-books using the ISBN for the first Nook edition to appear on bn.com (or the BN ID, usually used for self-published titles in e-book only when there is no ISBN) and the ASIN (Amazon Standard Identification Number) of the first Kindle edition to appear on Amazon.com. Because PW as yet does not track e-books, we now use the weekly list produced by Digitial Book World (digitalbookworld.com).
Being Nr. 1—Our consolidated ranking is based on calculating a rating for each title, which combines the rankings of each edition of a title by its rank on various lists and converts it to a percentile score. The highest rating a single title can achieve is 100. While there can potentially be 23 different Nr. 1 titles across the different lists of NYT and PW, because USA is a single list of 150 titles, only one title can be Nr. 1 by consolidated ranking for a given week. (See the next tab for more details on how the methodology has changed over time).
Last update: February 11, 2013—(paragraph 13, highlighted, dealing with "bonus" editions)
Breaking away from Traditional Patterns—
It was normal in the first years of our list maintenance for titles to follow a very traditional pattern, one that had been in place for decades before we started the database in 2001. Most titles were released first in hardcover ("trade cloth"). If the publisher felt there was enough potential, the title would be reprinted in "trade paper" (same size as trade cloth, with a paper cover) exactly one year later, followed in another year by a smaller "pocket" sized ("mass market") edition. Having multiple editions of the same title on the lists at the same time was rare and when it did occur (as in the resurgence of a hardcover edition when a paper reprint was released), the older edition's rankings were usually inconsequential.
While the traditional pattern was followed by many publishers and authors, things began to change and accelerate through the first decade of the 21st century. Prolific authors like James Patterson started to break the mold, keeping a steady flow of bestsellers on the lists by alternating the release of new hardcover and reprints every four to six months. Not only was the edition cycle breaking out of the strict 12-month pattern, but we began to see some titles jump directly from hardcover to mass market, while others bypassed hardcover altogether, starting as trade paper and then reprinting as mass market. (While this describes most ficiton, since we have been tracking the lists there has always been a proportion of romance titles that have only been published in mass market format).
The lists themselves also began to change. In September 2007 NYT split its Fiction Paperback list into separate Fiction Trade Paper and Fiction Mass Market lists. PW already had separate lists for trade paper and mass market (though mixed between fiction and nonfiction), but they only covered 15 items. In September 2010 PW expanded its four weekly lists to 25-items.
Whether the lists were simply reflecting reality or actually contributing to change, by 2009 we began to see the simultaneous release of trade paper and mass market editions, often as movie or TV tie-ins based on the original hardcover bestseller. The voracious appetite of television has surely helped sustain made-for-television movies and series inspired by books.
E-books—NYT and PW lists are centered on print formats. E-books started to appear on USA in 2010. They were not much of a factor until the last week of December 2010 when many of the top-ranked titles were attributed to e-book editions. As PW reported, this was likely due to a rush to load up Amazon Kindles, Barnes & Noble Nooks, Apple I-pods and I-pads, and other e-book capable devices received as Christmas gifts the previous week. The trend continued into January, adding pressure for us to switch from an edition- to a title-based model for our Consolidated Ranking.
From Editions to Titles—the impact of these changes began to produce problems with our edition-centered Consolidated Ranking as 2010 came to a close. Stieg Larsson's top-ranked Millenium Triology (The Girl with the Dragon Tattoo and The Girl Who Played with Fire) appeared in both trade paper and mass market. Since USA attributes its ranking to only one edition, it began switching between the two, depending on which was dominant that week. There was no problem when it switched for an entire month, but when it split within a month, then both editions fell in their consolidated ranking.
Then came the explosion of top-ranked e-books on USA after Christmas 2010. Using the edition-based Consolidated Ranking, almost all of the top-ranked editions the first part of the month would have lost their positions. So, in January 2011 we converted our analysis to a title-based model. With this approach, each edition was still restricted to a top ranking of 100, but the rating for a title became the sum of all of its editions on the lists during the report period. In its reports for February 6, 2011 sales (in print February 20), New York Times introduced two 35-item e-book lists—which were easily accommodated into the new title-based structure.
From Sum-of-Editions to Title Rating—calculating the rating as the sum of edtiion ratings, where most of our top-25 titles would have ratings above 100, seemed to work well until we ran into a situation not unlike the impact of Stieg Larsson's Millennium Trilogy. This time it was E. L. James' Fifty Shades Trilogy (Fifty Shades of Gray, Fifty Shades Darker, and Fifty Shades Freed). As the trilogy soared to extraordinary sales volume, Fifty Shades of Gray remained stubbornly behind Nicholas Sparks' The Lucky One, whose sales volume was much lower, but whose rating got a boost because the tie-in to the April 2012 movie kept trade paper, mass market, and e-book editions high on the lists.
When PW changed its lists and began showing BookScan volume, it was even more obvious that the sum-of-editions approach needed to be reviewed. Taking a sample of data, we ran an extensive analysis of the calculations used to obtain the rating for each title. Where the previous technique allowed each edition to have a maximum rating of 100, the revised method allows each title a maximum rating of 100, with a single edition contributing no more than 75 points (Nr. 1 on all three lists for that edition).
The new calculation of the rating reduces two types of distortion: first, it prevents a title (like The Lucky One with multiple print editions) from surpassing a title (like Fifty Shades of Gray) that is Nr. 1 on each listing for two or more editions; secondly, it prevents the less likely situation of a title that achieves a "perfect" Nr. 1 on all three lists, but only in one edition, from tying with the title that is Nr. 1 in multiple editions (typically e-book and one print edition).
The result will still not completely align with PW's BookScan numbers because it does not represent all the ISBNs on the three lists (notably e-books), and more importantly because we are consolidating list rankings. Since each list is based on sales volume but uses a different sample, the overall ranking will reflect that volume across the roughly 350 titles that appear on the 23 individual lists each week. Our aim is to produce a consolidated ranking that best reflects those lists.
The edition rating calculation was revisited in the fall of 2012 when we started tracking Digital Book World's e-book list and then Hurricane Sandy shut down that list for two weeks. The calculation was adjusted to compensate for the number of lists available in any given report period. This also takes care of historical reports to better handle PW in years that it did not publish one week in December. Over time, especially with the introduction of e-books in 2011, it has gotten harder for a title to achieve a perfect score of 100. For a time, the title had to be Nr. 1 on five lists: a NYT print list, NYT e-book, PW print, USA, and DBW e-book. However, because DBW represents only 25 titles (where NYT has 70 e-book titles between its two lists), we removed the DBW list from the default number of lists each week. Now, achieving a score of 100 is possible by being Nr. 1 on one NYT print list, NYT e-book, any PW list, and USA.
In February 2013 we dealt with several rare situations where a title had too many editions, which causes an obvious distortion in the resulting consolidated ranking. For example, a title may appear high in paper editions and e-book, but the original hardcover may still appear very low on a hardcover list. This one low ranking will cause the consolidated ranking to be much lower than expected otherwise. To deal with this, we added the ability to designate such an edition as a "bonus," which causes an adjustment in the calculation of the title's rating. The title rating may still be reduced somewhat, but not with the penalty imposed by the normal calculation. The situation has also been noted on PW, which assigns rankings to specific ISBNs, so several titles in January 2012 had two editions in the same list. The second edition can be assigned bonus status. When such an adjustment has been made for either cicumstance, it will noted on reports.
Children's/Youth books—Children's/Youth books have not been covered evenly on the lists, though that improved markedly with the changes in PW in June 2012. Before that, USA was the only list to include chidlren's titles equally in the mix of all titles. When we started the database in 2001 NYT had a mix of three children's and youth lists (Picture, Chapter, Paperback). Then, in response to blockbuster series like Harry Potter, which could dominate the chapter and/or paperback lists, in February 2005 NYT added a Children's Series list. The problem with this list is that it masked individual editions, so titles within a series could not contribute to the ranking of a title, leaving only appearances on USA. This has been alleviated to some extent by the expansion of PW of its monthly Children's lists to weekly 25-item lists for Children's Fiction and Children's Picture Books. In late November 2012, NYT modified the structure of its Children's lists, replacing the Chapter and Paperback lists with Middle Grade and Young Adult to make the lists more responsive to demographic targets within children's/young adult publishing. Those two lists were also expanded from 10 to 15 items.
We have made available a top-25 list that you can view here. It is updated weekly. Otherwise, our full research is currently available through Simba Information's Book Publishing Report (see more below).
During our early work with Tyndale House Publishers, we developed a web-based interface. Among its features was the ability to search by title, author or publisher and then view summary information about a title or see its List History (screenshot). We are planning to update this feature and make it available on a subscription basis. Please contact us if you are interested.
Tyndale House Publishers had a mega-hit on its hands in 2001, the Left Behind series, which was making waves on the bestseller lists. Looking for additional angles to use in publicity, we were asked to study the bestseller lists, which led to the present database.
As the database matured, various types of reports were generated, from regular summaries to detailed analysis performed to answer specific questions. For example, because we started the database a few months before the September 11 terrorist attacks, we were able to track trends in books on religion, Islam, and terrorism over the following months.
By 2006 the Left Behind series was winding down and Tyndale dropped its support of the project. We continued to maintain the database, focusing on the weekly lists, in hopes of securing another outlet.
What developed was a relationship with Simba Information, which began in early 2007 by supplying reports for use in their monthly electronic newsletter Book Publishing Report (BPR). Two standard reports are produced each month: a Scorecard summary of statistics for the previous month, and a Category Review, which looks at trends for a specific category of books, starting with the previous four quarters and then looking for trends over five years. In addition, a number of special reports are produced through the year, looking at a comparison of top publishers, children's books, pricing, and other topics.
©2013 Stuart Johnson & Associates