Why Does NovelRank Underestimate Sales for Popular Books?

Every once in a while, I get an email from an author with Kindle Direct Publishing (KDP) access who wonders why their numbers are different between NovelRank and KDP. First and foremost, let me say that if you have KDP access for your titles, that is the most accurate source of sales data. Secondly, remember that NovelRank estimates sales based off of sales rank, which is also a data point provided by Amazon, a priority system for ranking books based upon their sales. This makes any estimates vulnerable to many variables, and I wanted to provide some additional info for folks.

One final note before I start listing things: I created NovelRank as a resource that allows authors to spend less time reloading webpages, spend more time writing, and get a feel for the effectiveness of marketing campaigns by seeing hourly and daily fluctuations. It’s hard to communicate this throughout the website, and a few folks have taken to disliking NovelRank strictly based on sales estimates. It hurts a little, but I try and focus on those who are benefiting from the service.

Causes for Underestimation:

  • The better the sales rank, the less variation. Since sales estimates are based upon changes in sales rank, this reduced variation is the first source of underestimation.
  • Furthermore, when sales rank is really good (low number), improving estimates usually means adding arbitrary +X amounts to sales estimates. I try and avoid this as it amounts to guessing.
  • The algorithm uses the last 1000 data points to not only identify sales, but also try and throw out false positives (when sales rank changes a small amount not due to sales but due to Amazon adjusting the ‘list’ of books being ranked in that category). This is typically a boon, but in some cases this can miss genuine sales.
  • If Amazon fails to update sales rank for that hour (common occurrence), that data is ‘lost’, because if a sale occurs during that time, by the time a new sales rank is presented, there is less differentiation.
  • Multiple sales on an individual order are represented as 1 sale by sales rank. See this post about Sales Rank Myths for more info.
  • Finally, sales rank is an ever changing secret formula which NovelRank’s algorithm is constantly trying to learn from. It will never be perfect, but I always hope it can keep up.

I hope this clarifies things for new authors. NovelRank exists so you can do more writing, selling, and living, so enjoy!

Similar Posts:

Be Sociable, Share!

    7 thoughts on “Why Does NovelRank Underestimate Sales for Popular Books?

    1. Claudia

      Thank you very much from a German author for this wonderful website. I am using it constantly for one year now.

      Kind regards from Germany


    2. Anthony

      Sales are hard to figure out. My book, “The Diary of the DC Sniper” cracked 80,000 but it was never reported on novel rank. Additionally sales reported is far from remotely accurate.

      Novel rank is an algorithm, it’s not exact. You can jump 200,000 in an hour if your book sells 2 copies.

      It is far from accurate.

    3. Drew Hensley

      I’m sure it is because Baez’s book is selling better in cities with a high concentration of intelligent, open-minded people

    4. Talin

      Hi, thank you for the site. I would like to point out it doesn’t always take a sale to see a drop, this morning at 9am my ranking dropped to roughly 83K. I didn’t sell anything though until this evening. I did sell 5 total for the week up to that point.

      My theory is that was when they reweighted the weekly sales stats. I’ve heard theories of 40% hourly sales, 30% weekly, 20% monthly, 10% yearly to steady out things in the long term. I’ve no clue if that’s true but I’m curious if you noticed an across the board update at that time of day. It could be a rolling update even that affected books out for 1 or 2 weeks more than 1 year.

      Thank you again for the site, hopefully that helps.

    5. Cate

      Hi There-

      I am comparing the sales and pre-sales of two different books. One is Jeff Ashton’s “Imperfect Justice-Prosecuting Casey Anthony” and the other is Jose Baez’ soon to be released “Presumed Guilty-Casey Anthony the Inside Story”. I searched sales data for Months May & June (is it always just current and last month-no cumulative sales?) in US and Canada and UK. What I did not understand is why Jose Baez’ current sales rank is “717” compared to Jeff Ashton’s which was “4340” and when u look at sales Jeff Ashton’s book shows more sales everywhere but is at a lower rank. Can someone explain?

    Comments are closed.