New Ranking Algorithm Seems To Be In Place

Jan Strnad

· #2 · Mar 4, 2013

I don't know about rankings, but I'm jealous as hell of your sales! Big congrats!

Reply

Save

Like

P

Pnjw

· #3 · Mar 4, 2013

I'm going to guess that your old title did better because it has sales history. And it takes more for a new title to get higher in the rankings.

Reply

Save

Like

1

13893

· #4 · Mar 4, 2013

I released a new title this weekend and topped out at 20,000 ranking with 15 copies sold. Somehow, I'm not feeling your pain!

Reply

Save

Like

My other titles were affected too. Books that were sitting at 300, fell to 500 and seem to be moving a similar number of units. It was across the board and occurred a few days ago. It also affected author rank. Anyway, just asking what others were seeing. I thought since we were coming out of peak season that the ranking system would have moved the other direction. Add to that higher dollar book and more copies sold and it has me scratching my head.

Reply

Save

Like

Kwalker

· #6 · Mar 4, 2013

I think it would make sense for things to be changing, considering the affiliate changes, etc. I also noticed I seemed to be floating lower with more sales, but I assumed that was simply that other people were seeing larger increases than I was =)

Reply

Save

Like

J

Jay Allan

· #7 · Mar 4, 2013

My impression is that past sales history does factor in. I suspect an older title and a new one selling exactly the same would end up in different places.

That said, I am sure there is enormous volatility in sales in the top 100, depending on big new releases and external factors (e.g. new Game of Thrones season starts and all five books rocket up in sales).

Reply

Save

Like

PhoenixS

· #8 · Mar 4, 2013

holly w. said:
Okay, I launched a book this weekend and the rank is SO not what it had been in Jan. In a 24 hour period the new title sold 1700 copies at $2.99 (new release). The rank topped out at 63. In Jan when I sold 1200 copies of an older title in a 24 hour period at 99 cents, the rank was 34. $:-\$ Uhm, yeah. That's all I got. Anyone else seeing this? Or is it something to do with new releases?

Holly, are you absolutely sure that you only sold 1200 copies of that January title? We had a title that sold 2292 copies in a 24-hour period on Amazon that only made it to #30. We had a book in the Top 100 the entire month of January and during the time I was watching, 1200 sales wouldn't put a title anywhere close to #34. I'm not disputing your rank, just the number of sales it took to get there.

And price on the bestseller list is not a factor. Never has been.

ETA: To clarify, the title with the 2292 sales over 24-hours in January was not the same title as the one in the Top 100 for the month.

Reply

Save

Like

H.M. Ward

Discussion Starter · #9 · Mar 4, 2013

Phoenix Sullivan said:
Holly, are you absolutely sure that you only sold 1200 copies of that January title? We had a title that sold 2292 copies in a 24-hour period on Amazon that only made it to #30. We had a book in the Top 100 the entire month of January and during the time I was watching, 1200 sales wouldn't put a title anywhere close to #34. I'm not disputing your rank, just the number of sales it took to get there.

And price on the bestseller list is not a factor. Never has been.

Maybe my numbers are off from the lag? It seems to be about 12 hours with rank, but the sales aren't off by that much - maybe an hour or two? It's hard to match them up. If I'm way off of that part, feel free to fill me in. Trying to figure out some of this stuff is like trying to hold onto a greased pig.

So, price has no bearing on rank? I've heard price is weighted in the ranking algorithm over and over again, but haven't seen any proof of that with my own titles. I feel like I took stupid sauce today.

Reply

Save

Like

PhoenixS

· #10 · Mar 4, 2013

holly w. said:
Maybe my numbers are off from the lag? It seems to be about 12 hours with rank, but the sales aren't off by that much - maybe an hour or two? It's hard to match them up. If I'm way off of that part, feel free to fill me in. Trying to figure out some of this stuff is like trying to hold onto a greased pig.

So, price has no bearing on rank? I've heard price is weighted in the ranking algorithm over and over again, but haven't seen any proof of that with my own titles. I feel like I took stupid sauce today.

For the first part, I'm actually saying I don't think 1200 sales in January would have led to a #34 rank. That sounds like it's off by about 1000 sales. The 1700 sales hitting #63 with a lag sounds more like what the sales:rank ratio should be right now.

The price weighting is ONLY in the pop lists, not the bestseller lists. You might well see people erroneously claiming there's a bias in the bestseller lists, but those are folk who don't understand the difference between those lists and the algorithms that drive them and who are playing telephone when they pass along what they *think* they understand.

Reply

Save

Like

valeriec80

· #11 · Mar 4, 2013

Well, not to be overly obvious, but ranking is relative, right?

I mean, that's the standard Amazon answer, that it has to do with how many books are selling that day? I imagine that there's an amount of books that usually sells each day, but there's got to be fluctuations, right? So, maybe the day that you sold 1200 books was just a slow day for everyone and the person in the number one spot sold a lot less than the number one spot usually sells? *grasps at straws*

I did notice that my rank started to fall before my numbers did when Slow Burn started its swan dive. I was still selling about 200 books a day, but at first that got me at #250 in the store and the next few days it fell to the #300s. This led me to conjecture that rank may have a comparison aspect in it, like, something that sold 2 books yesterday and 200 today might be ranked higher than something than consistently sells 250 day in, day out. I can't be sure, of course, but I think they may weight up-and-comers higher to encourage that the lists churn a bit and stay fresh?

Just throwing stuff out here. I know nothing at all.

See less See more

Reply

Save

Like

H.M. Ward

Discussion Starter · #12 · Mar 4, 2013

This thing is totally broken, then. http://kdpcalculator.com/index.php

The number this thing puts out were close to my numbers before March.

Yes, the rank per units sold varies from day to day, just not that much. And every two months or so, it seem like Amazon presses the reset button and all my ranks drop for a few days and then climb back up. If that's true, and predictable, then you can consider that info when planning the release date.

Another something weird, Amazon put down the release date as March 1st, but it wasn't uploaded until the afternoon on the 2nd. What's up with that?

Reply

Save

Like

Rick Gualtieri

· #13 · Mar 4, 2013

holly w. said:
Another something weird, Amazon put down the release date as March 1st, but it wasn't uploaded until the afternoon on the 2nd. What's up with that?

I had that happen too. Updated a book with a new edition and modified the date, which then showed as being a day earlier than I had chosen. Sounds like a calendar glitch in KDP.

Reply

Save

Like

scottnicholson

· #14 · Mar 4, 2013

I think there were some lags this weekend and not all 24 hours are created equal. As well as an obvious pop-list crunching delay that could have affected the entire chart.

I've rarely seen the system work exactly as we "think" it will, even based on rigorous observation, and I have never seen it duplicate the exact same output even with the exact same input. It's a living organism with its twitches and glitches. But I do have a hunch we have just entered a new epoch.

Reply

Save

Like

Terrence OBrien

· #15 · Mar 4, 2013

holly w. said:
Okay, I launched a book this weekend and the rank is SO not what it had been in Jan. In a 24 hour period the new title sold 1700 copies at $2.99 (new release). The rank topped out at 63. In Jan when I sold 1200 copies of an older title in a 24 hour period at 99 cents, the rank was 34. $:-\$ Uhm, yeah. That's all I got. Anyone else seeing this? Or is it something to do with new releases?

Why presume a change in ranking algorithms?

Reply

Save

Like

David Kazzie

· #16 · Mar 4, 2013

Somewhat related, I also think that something happened with the weighting of freebies.

I ended my worst of four KDP Select runs on March 2 (All have been 2 days runs). I ended with less than half as many downloads as my previous worst (December 2012), and didn't make the top 100. But since going back to Paid, I'm selling twice as many copies as I did after the December run.

Reply

Save

Like

M

Monique

· #17 · Mar 4, 2013

Holly, I think you might be misremembering the sales totals and velocity that it took to reach that rank in Jan.

http://www.kboards.com/index.php/topic,138781.0.html

Reply

Save

Like

H.M. Ward

Discussion Starter · #18 · Mar 4, 2013

Monique said:
Holly, I think you might be misremembering the sales totals and velocity that it took to reach that rank in Jan.

http://www.kboards.com/index.php/topic,138781.0.html

Hmmm. So whats the diff b/t time when sales post to kdp and the rank is updated? I think that's what's confusing me.

Reply

Save

Like

Terrence OBrien

· #19 · Mar 5, 2013

There is no set time. Four events occur. A consumer hits the buy button, KDP reports the sale on the dashboard, sales rank updates, and popularity rank updates. The buy button has to happen first. The other three can happen in any order. There is also no set time between the various steps.

We are all connected to different servers. One server may do it one way, while another executes another way. Much depends on how busy the servers are. It's a very complex system, and each task has a priority. Reporting to the KDP dashboard seems to be one of the lowest priorities. Nothing depends on it.