Tag Archives: stats

Amazon Book Sales Statistics for 2012 – Time-Based

These statistics are provided as-is and are based on NovelRank’s sales estimates. Any inaccuracies are universally applied and should not affect the overall trends in the data. However, it is limited by the total books in NovelRank’s database, and thus is automatically biased.

TL;DR Grain of salt.

The very first chart was similar in shape across all domains, thus only one was included. All other charts are also summaries of only 2012 data and is shown for the top 3 Amazon domains tracked on NovelRank. Compared to the previous statistics post, these statistics are much less conclusive. For that reason, I encourage you to draw your own conclusions and post them in the comments below.
Continue reading

Sales Rank Stats Return and Algorithm Updates

Sales Rank Stats

After many months of being restricted to the last 30-60 days, Sales Rank Stats have returned! Head over to a book’s details page (e.g. Example) and you can click the line titled Sales Rank Stats to show the book’s best, worst, average, and standard deviation in each domain that is being tracked. This information is updated the first time you access the page, and then will automatically update itself every 30 days. Unfortunately, real-time updating was making the page load very slowly, thus the 30 day cache was built. This should return page loading to < 2 seconds 99% of the time.

It make take up to a minute to load the page for the first time!

I’ve turned up the power on the server to hopefully speed things up (it’s at max, which is ~$7/day) while everyone (including search engines) forces the statistics to be calculated. You may see some erratic slowness across the site while this occurs.

Sales Estimate Algorithm

After feedback and testing, I’ve improved the new algorithm to use tighter statistical analysis of sales rank changes. It is also now attempting to estimate for sales occurring when a book holds its rank (e.g. 700 last hour, 705 next hour) without improving. This is only for sales rank < 1000, but is also based upon the statistical median. Translation: It’s smart enough to deal with the differences between each Amazon domain as well as the Kindle or print distinction. Not only is it better than the last system I used, but it learns each day how to get better!

Final Note

These changes conclude the upgrades to NovelRank’s core that officially started on November 1st (but have been planned for months). As usual, I’ll keep an eye on things and make adjustments as necessary, but I think you can consider this NovelRank 2.0!

If you run into any problems like weird errors or broken pages, please let me know by sending an email to admin [at] novelrank {dot} com.