Saturday, June 25, 2011

Amazon Sales Rankings: What? Why? Huh?

Another in my Sales VoyeurTM series: making the business of self-publishing an ebook that is doing well but isn’t a bestseller (yet!) transparent :o)


So you see a ranking like #8,621 on an Amazon page and want to know what it means in terms of how many copies of that book have been sold. Can you tell by the ranking?

Well, yes and no.

Is the ranking at all helpful for a reader in assessing the popularity of a book?

Well, yes and no.

I'm going to generalize here a bit -- OK, a lot -- as Amazon's algorithms are pretty closely guarded. I've seen a number of different charts and guesses across the web, but none of them seem to match up well with each other's results or my own. So I'm guessing my conclusions -- bearing in mind I've only been at this for 3 months -- come as close as anyone else's.

Note that these observations are for e-books only and only for Amazon US. Amazon UK seems to run completely separate, and it takes fewer sales to climb the ranks.

The problem with using rankings to determine if a book is popular is that you don't know if the author maybe has an ad or other placement that's generating some interest right then. A few days ago, I saw one author's book go from a ranking of 220,000 to 43,000 with the sale of 2 books within 2 hours of a blog post appearing on one of those "cheap Kindle books" sites that featured her book. Her rank dropped to 23,000 when she sold another book within another 2 hours. However, she hadn't sold any books in June prior to the post, and she's only sold one since. Her rank is around 176,000 now. So if you happened to see her book when it ranked at 23,000, you'd think it was selling a whole lot better than it actually is.

Remember, too, that the rankings are relational. Each book is judged against how all the other books out there are selling at a given point in time. So your book could zoom past someone else's that's consistently ranked about the same as yours with the sale of a single copy if the other book didn't sell at all during the same time period. Then next hour, their book sells and yours doesn't and you're back on equal ranking again.

In general, rank to sales breaks down:
  • 30,000 - 80,000 = 1 copy/day
  • 20,000 - 30,000 = 2 copies/day
  • 15,000 - 20,000 = 3 copies/day
  • 10,000 - 15,000 = 4 copies/day
  • 5,000 - 10,000 = 5-10 copies/day
  • 2,000 - 5,000 = 10-15 copies/day
  • 1,000 - 2,000 = 15-20 copies/day
Under 1,000, things start getting extremely competitive, but probably 600-1000/month. And under 100? A lot. (Anyone with more concrete data via actual sales figures over time, please chime in!)

So far, I can vouch for the number of copies sold for rankings between 8,000 and 80,000. But again, this all changes hourly and daily, so it depends on when you happen to be looking at a book's rank. A book could sell 4 copies in the morning and be ranked at, say, 12,000 by noon, then be down to 25,000 by the end of the day if no more were sold. A consistent average in rank across multiple days is your best bet for determining how well the book is doing.

Disclosure: To gain the 8,621 rank seen above on the US site, Spoil of War had been selling consistently for about 10 days (historical sales play a small part in the algorithm, it seems). And I sold 9 copies on Amazon US that day.

So how can you figure out how a book is trending? is a good site to track the fluctuations in sales rank. I've mentioned the site before, but my enthusiasm for it has cooled some because of its limited functionality. It pulls actual rank feeds so those are absolutely valid. Its conversions of those ranks to an estimate of how many actual books have sold is accurate to within about 10% for number of copies sold in a month up until about 35 copies. After that, there's a decay in whatever algorithm it's using and it begins under-reporting by 20-30% and more. Since it pulls true rank feeds from Amazon, though, you can see a pretty accurate graph and tell if a book's sales are fluctuating wildly or consistently going up or down. The flatter the waves, the more consistent the book’s sales. You want the book to flatline -- preferably at a low, low rank.

Note how the waves in the shaded area start to flatten out when a book -- Spoil of War, in this case -- starts selling a few copies a day consistently.

Here's a screen captured last night that shows US, UK and German rankings for Spoil of War. Note that the graphs are accurate, but the estimated numbers are off. For instance, it shows Spoil sold 43 copies in the US in May and 7 copies in the UK. In actuality, Spoil's sales were 49 and 8 respectively. That discrepancy isn't too bad, though.

June sales, however, are way off in the US where NovelRank estimates 79 sales and actual is 105. Since UK sales are low, the 13 estimated is close to the 14 actual. Germany (DE) is consistent only because I haven't sold a book there since last month. Unfortunately, I didn't add Germany to the rankings to watch until after the book sold, so the database has no knowledge of that sale.

Why is NovelRank useful if Amazon delivers the raw data more accurately? For your own book, NovelRank does what Amazon doesn’t and handily aggregates the US, UK, German and Canadian (print versions only for sites into one graph on a single page. That page is public for all to see. That means you can also look at the selling history of OTHER authors’ books and see not just a snapshot in time like you get on the book's product page, but how a book is trending, which can, for example, help you decide if a marketing or advertising avenue is worth your money or not.

Amazon does have an Author Central site, which provides Nielsen Bookscan reports for how well and where books are selling per site, but that feature is limited to print. Author Central also provides a graph of sales ranks, but it looks rather anemic in comparison to NovelRank’s version, and you only get rankings for one sales region.

And here's what Amazon’s report of copies sold, which updates hourly or so -- but you don’t know exactly when -- looks like. This was captured at the same time the NovelRank sales rank above matched the ranking on the Amazon product page (there can often be a time delay of an hour or two). Watching this report is like playing the slots. You never know what you’re going to find when you click the links and refresh. Win or bust? Click. Click. Click. Click. If you have compulsive tendencies, this isn’t a site for you.

US report of month-to-date unit sales. Has the number changed in the last 10 minutes? Maybe I should refresh to find out!

The UK report. The UK database of books is smaller, so even with considerably fewer sales, I've managed better sales rankings on the UK site than the US one. That's why these numbers are all relative.

Only the publisher has access to this report, so even though Amazon provides these numbers hourly-ish, an author or agent may not be aware of the actual numbers until the time the royalty statement comes due. That’s another reason why a site like NovelRank can help a traditionally published author as long as s/he pays attention to it for the rankings alone.

If anyone knows of similar tracking apps for B&N, let me know. For the record, I've only sold 9 copies of Spoil of War through B&N and just 1 copy through Smashwords this month as of yesterday, so if you have any secrets about how to boost sales in those venues, spill them, please.

Tomorrow I’m going to talk about why I think my sales suddenly took a small change for the better at Amazon a couple of weeks ago (when sales started to flatten out in the graphs above), why those sales are probably headed for a downswing in the near future, and why some books aren’t doing as well as their authors had hoped. This includes a scary story regarding a traditionally published book, so will be of interest to all.

In the meantime, I’m going to go play the slots – aka check my sales numbers – again. And you can go read this fabulous review :o)


Wilkins MacQueen said...

Divine One,
Read the review. Oh my Buddha.As soon as I get paypal figured out I am so buying this.

I've tried to get paypal.Really have. My bank people speak no English, I tried on line. No go yet. And I'm dying to read Sector C.

Sarah Laurenson said...

And this is why traditional publishing is still a draw. So much work goes into tracking, and advertising, and pushing at the right time and place, and finding other strategies to try.

More power to you!

Phoenix Sullivan said...

@Mac: Maybe we can work something out ;o) I'll email you later.

@Sarah: Tomorrow's post may be a bit eye-opening on the push, track and advertising side of traditional publishing. I'm not anti that route by any means! I've got a full of Sector C at a trad publisher right now under consideration. But it seems there's some convergence of responsibility these days. I was gobsmacked watching the progress of two novels recently pubbed by Big 6 publishers. It's been really eye-opening for me...

Sarah Laurenson said...

Oh, yeah. A lot less being done by them and more by the author. It's a balancing act and the scales are tipping.

Landra said...

This info is so crazy, and don't I know about the whole 'click click click', 'refresh, refresh, refresh'. It's nerve wrecking, but you can't help to continue clicking the mouse to check the stats.

I really hope the numbers don't downtrend 'crosses fingers'. I'm going to keep promoting the dickens out of this because I love the story so much.