From highs to lows: Amazon sales rankings

A lot of authors (including me) find it easy to get hung up on our Amazon ratings. We track them obsessively. It’s pretty addictive. As one New York Times article says, there ought to be a 12 step program. And when we get into the vaulted top 100 for our category, wow!

We must have sold a pile of books, we think. Then we find out our actual book sales and wonder what the heck happened.

Amazon is very secretive about the algorithm it uses, so there’s a lot of speculation on the Internet as to how they arrive at those rankings. Also, sales are weighted, and ranked relative to other author’s sales, making it hard to interpret the data.

I recently discovered a program called NovelRank that is supposed to take some of the guesswork out.  You enter a novel’s ISBN and name and it tracks Amazon sales hour-to-hour, and prepares a daily chart for you. (SalesExpress.com does the same thing, but it doesn’t cover Canada.)

Unfortunately, as I learned this week, NovelRank treats each order as a single book sale, even if the order is placed for multiple books, so it’s not all that accurate.

For example, I ordered five books to replenish the ones I’d sold at my office and at events. That sale was recorded as one sale by NovelRank, not five. Same thing a few days later when I bought two more.

NovelRank’s high/low rankings don’t make much sense either. I’ve been much higher (top 50) and much lower (bottom 45,000) in the rankings than its figure suggests. (It has me at a best rank of 6,262 in Mystery/Thriller and a worst rank of  7,033 when I am in fact ranked at 10,094 as I write this.)

What I did find out, though, was that the purchase of even one book (assuming that not everyone buying from Amazon is placing multiple orders like me) was enough to move The Beggar’s Opera up several thousand places.

One day last week, for example, a single sale moved me from around #10,000  to #98 although I quickly fell back down. Two sales the next day (in separate purchases, according to NovelRank) and bingo, and I hit #34.

This kind of volatility is not uncommon. As that same New York Times article points out, “A book that is the millionth-largest seller on Amazon might surge hundreds of thousands of places with one sale.”

Which means that “Amazon bestseller” doesn’t necessarily mean an author sold a lot of books. A handful of books can make a big difference.  

One an American analyst  who has looked closely at Amazon’s American rankings and done some complicated graphing that’s beyond my math skills says this: 

“The idea behind my reverse-engineering the ranking system was always to give rough idea of how a title was selling, not an exact number. So, don’t read an average rank of 10,000 to mean you sold exactly 60 books that week, or a rank of 100,000 to mean you sold ten and a half copies – Amazon doesn’t sell half copies.

“Read an average rank of 1,000 to mean you have a seriously successful title, an average rank of 10,000 to mean you’re doing pretty good for a book that’s no bestseller, an average rank of 100,000 to mean it’s not going to contribute significantly to your income, and an average rank of 1,000,000 to mean you need to take a break from checking sales ranks.”

No kidding! Now remember, he’s writing about the U.S. market, which is at least ten times larger than Canada. It may mean that an average rank of 1,000 means you’ve sold six books that week. But who knows? Only Amazon.  

(By the way, a few minutes later, I am now at #9,065, so I’ve risen in ranks, but without making a sale according to NovelRank. It’s that crazy algorithm at work. Maybe someone returned another mystery author’s book?)