KBoards | a community forum for Kindle Users and Authors

Authors' Forum => Writers' Cafe => Topic started by: Trish McCallan on December 15, 2013, 06:46:30 PM

Title: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 15, 2013, 06:46:30 PM
Hey Everyone,

I’d like to put together an Amazon rank to sales chart with new data.  Sales to rank after Thanksgiving seems a bit inflated, from what I’m seeing on my own title. So for those of you who are recording your book’s rank and sales per day, would you be willing to share your data for November?

I know it takes more sales to achieve a rank than to hold it, so if you could include whether your book’s sales were accelerating, or whether they were holding steady—plateaued—would be helpful.   
If you're uncomfortable posting to this thread, please feel free to send the data to Trishmccallan1@gmail.com

Thanks in advance to anyone who is willing to participate.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Edward W. Robertson on December 15, 2013, 07:45:26 PM
Steady ranks:

400/day = #180
300/day = #235
110/day = #1000
100/day = #1100
Title: Re: Requesting data for an updated sales to rank chart
Post by: Vivi_Anna on December 15, 2013, 07:50:10 PM
I can add...roughly...

600/day = #135
Title: Re: Requesting data for an updated sales to rank chart
Post by: Edward W. Robertson on December 15, 2013, 07:59:45 PM
And 50/day is about #2400.
Title: Re: Requesting data for an updated sales to rank chart
Post by: WG McCabe on December 15, 2013, 08:07:02 PM
1 every other month is 1,262,295

 ;D
Title: Re: Requesting data for an updated sales to rank chart
Post by: Ardin on December 15, 2013, 08:17:28 PM
35 a day is 3000. 20 a day is 7000. 10 a day is 10000. Holding steady.


Sent from my iPod touch using Tapatalk (http://tapatalk.com/m?id=1)
Title: Re: Requesting data for an updated sales to rank chart
Post by: AriadneW on December 15, 2013, 08:41:46 PM
80 a day is around 1500
Title: Re: Requesting data for an updated sales to rank chart
Post by: Charmaine on December 15, 2013, 09:43:46 PM
Here's the equation for the all the data plots :)

y=175625.63126691X^-1.1252673308641



1 day/ rank 175,625
5 day/ rank 28,711
10 day/ rank 13,161
15 day/ rank 8,340
25 day/ rank 4,693
50 day/ rank 2,151
100 day/ rank 986
200 day/rank 452
300 day/ rank 287
400 day/ rank 207
500 day/ rank 161
1000 day/ rank 74
2000 day/ rank 34
3000 day/ rank 21

The higher the rank, the more the graph gets distorted (it's an exponential graph  ::) )
Anyways, here's some shooting off points for your chart
Title: Re: Requesting data for an updated sales to rank chart
Post by: Edward W. Robertson on December 15, 2013, 10:24:05 PM
That formula is very accurate for many ranks, but starts to get inaccurate around #1000 (and #100,000 on the other end). For instance, it only took me 2200 to hit #16 with a climbing rank a few months ago. Still, it describes things pretty well.
Title: Re: Requesting data for an updated sales to rank chart
Post by: LilianaHart on December 16, 2013, 09:06:30 AM
#100 is about 1000 copies per day.

#200 is about 500 copies per day. 

#527 is 150 copies per day.

#450 is 175 copies per day.

#737 is 120 copies per day

#1700 is 57 copies per day

#2800 is 30 copies per day
Title: Re: Requesting data for an updated sales to rank chart
Post by: Hugh Howey on December 16, 2013, 09:30:16 AM
#100 is about 1000 copies per day.

#200 is about 500 copies per day. 

#527 is 150 copies per day.

#450 is 175 copies per day.

#737 is 120 copies per day

#1700 is 57 copies per day

#2800 is 30 copies per day

I call shenanigans!

How in the world would you know what it takes to be #2800?
Title: Re: Requesting data for an updated sales to rank chart
Post by: KeithAllen on December 16, 2013, 09:38:35 AM
This might be a dumb question, but aren't the algorithms price weighted? So if I sell a 1000 copies a day at 0.99 and 1000 copies a day at 3.99 won't my 3.99 title be ranked significantly higher?
Title: Re: Requesting data for an updated sales to rank chart
Post by: Edward W. Robertson on December 16, 2013, 09:45:26 AM
On the popularity lists, price is weighted. On the bestseller lists, it's nothin but volume and velocity.
Title: Re: Requesting data for an updated sales to rank chart
Post by: daringnovelist on December 16, 2013, 09:46:58 AM
This might be a dumb question, but aren't the algorithms price weighted? So if I sell a 1000 copies a day at 0.99 and 1000 copies a day at 3.99 won't my 3.99 title be ranked significantly higher?

I think I accidentally spread that rumor once.  I was talking about how Amazon weights free books when they go back to priced, and about relevancy rankings and conflated the two for a while.  I don't believe there is any evidence that different prices have different weights.

Camille
Title: Re: Requesting data for an updated sales to rank chart
Post by: KeithAllen on December 16, 2013, 10:55:56 AM
On the popularity lists, price is weighted. On the bestseller lists, it's nothin but volume and velocity.

Good to know...Thanks Ed.  Carry on...
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 16, 2013, 11:35:23 AM
Thanks everyone!

Keep those numbers coming. :)
Title: Re: Requesting data for an updated sales to rank chart
Post by: pjchasesports on December 16, 2013, 09:24:30 PM
I've had a book hold strong at 9800 rank with 12 sales a day.  And another of my books hits about a 50,000 rank on 2 sales a day.  Anecdotaly, I've heard one of my friends hanging with a 1300 rank with 75 sales a day.
Title: Re: Requesting data for an updated sales to rank chart
Post by: books_mb on December 17, 2013, 12:02:15 PM
I can add these (all .com store):

2 / day = #50,000
4 / day = #40,000
5 / day = #24,000
7 / day = #18,000

Btw: I already fed all the data from this thread into Origin Pro. The best fit is:

rank = 170,710 / (daily sales)^1.272

with an adjusted R-square 0.978 ... fantastic fit, better than expected. Hope more will contribute, this is gold.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 17, 2013, 03:38:31 PM
I've had a book hold strong at 9800 rank with 12 sales a day.  And another of my books hits about a 50,000 rank on 2 sales a day.  Anecdotaly, I've heard one of my friends hanging with a 1300 rank with 75 sales a day.

Thanks you!

I can add these (all .com store):
2 / day = #50,000
4 / day = #40,000
5 / day = #24,000
7 / day = #18,000
Btw: I already fed all the data from this thread into Origin Pro. The best fit is:
rank = 170,710 / (daily sales)^1.272
with an adjusted R-square 0.978 ... fantastic fit, better than expected. Hope more will contribute, this is gold.

This is great! Thank you so much!
And yes, PLEASE, let's get some more sales to rank data--particularly in 50,000 and up range. 100,000-500,000 could some of you please share what your sales are at this rank? It would be much appreciated. :)
Title: Re: Requesting data for an updated sales to rank chart
Post by: books_mb on December 17, 2013, 04:55:34 PM
And yes, PLEASE, let's get some more sales to rank data--particularly in 50,000 and up range. 100,000-500,000 could some of you please share what your sales are at this rank? It would be much appreciated. :)

I guess you rather share data that shows success ... to break the ice, here's the data from two of my books that are not doing so well:

0.25 / day = #260,000
0.3 / day = #250,000
0.5 / day = #200,000
0.67 / day = #180,000

It's all I got for now, but I'll bookmark the thread and post some new data at the end of December. Note that the provisional best fit fails for ranks > 100,000 (which is no surprise considering the lack of data). In this range the good ol' rank = 100,000 / sales is still the best guess. But I have no idea how reliable that is.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 17, 2013, 06:41:08 PM
I guess you rather share data that shows success ... to break the ice, here's the data from two of my books that are not doing so well:
0.25 / day = #260,000
0.3 / day = #250,000
0.5 / day = #200,000
0.67 / day = #180,000

Oh Perfect, thank you.  :)
Title: Re: Requesting data for an updated sales to rank chart
Post by: PhoenixS on December 17, 2013, 07:32:06 PM
Hey Everyone,
I’d like to put together an Amazon rank to sales chart with new data.  Sales to rank after Thanksgiving seems a bit inflated, from what I’m seeing on my own title. So for those of you who are recording your book’s rank and sales per day, would you be willing to share your data for November?

Theresa Ragan's chart is still a good ballpark, I think, if you just increase the sales by 10% for each rank. The issue with attempting to do any sort of more accurate chart is that, as you note, the numbers are all dependent on store volume. Black Friday to Christmas is one degree of level up. Christmas to New Year's is another degree. New Year's to Valentine's Day is another degree. April, July, August and October are depressed. To be truly useful, there would need to be a chart for each degree of store volume.

As Ed notes, it took 2200 sales to hit #16 in Sept. Yesterday, it took 2800 sales to hit #18. It'll take more than 2800 on Dec 28, and fewer than 2200 on July 28.  

What would be especially useful would be an overlay of sales-to-hit on top of store volume trends. But is anyone here tracking store volume via rank? I did some reverse engineering of rank last year in the period right after Christmas to determine the percent increase in volume at Amazon, but I haven't seen any similar studies from anyone else here. The difference in sales numbers to hit a given rank throughout the year is really more insightful than averages that, very much like a broken analog clock, are only accurate maybe 2 months out of the year. ;)

(Edited sales numbers because I can't multitask.)
Title: Re: Requesting data for an updated sales to rank chart
Post by: VydorScope on December 17, 2013, 07:34:44 PM
This is all very interesting. I can't wait to see the results.
Title: Re: Requesting data for an updated sales to rank chart
Post by: WG McCabe on December 17, 2013, 07:37:01 PM
http://www.youtube.com/watch?v=gZEdDMQZaCU

I kid. This is all fantastic and  I'm glad you are all smarter than me so it can be figured out.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Joe Vasicek on December 17, 2013, 08:41:38 PM
The free ranks are considerably different.  I haven't really measured them, but I've got a story that hovers around #2000 and gets about 70-80 downloads per day on the main amazon site.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 17, 2013, 08:48:15 PM
Theresa Ragan's chart is still a good ballpark, I think, if you just increase the sales by 10% for each rank. The issue with attempting to do any sort of more accurate chart is that, as you note, the numbers are all dependent on store volume. Black Friday to Christmas is one degree of level up. Christmas to New Year's is another degree. New Year's to Valentine's Day is another degree. April, July, August and October are depressed. To be truly useful, there would need to be a chart for each degree of store volume.

As Ed notes, it took 2200 sales to hit #16 in Sept. Yesterday, it took 2800 sales to hit #18. It'll take more than 2800 on Dec 28, and fewer than 2200 on July 28.  

What would be especially useful would be an overlay of sales-to-hit on top of store volume trends. But is anyone here tracking store volume via rank? I did some reverse engineering of rank last year in the period right after Christmas to determine the percent increase in volume at Amazon, but I haven't seen any similar studies from anyone else here. The difference in sales numbers to hit a given rank throughout the year is really more insightful than averages that, very much like a broken analog clock, are only accurate maybe 2 months out of the year. ;)

(Edited sales numbers because I can't multitask.)

I'd checked Theresa Regan's chart before posting, but it didn't look like it had been updated in quite a while, and from my own book and rank/sales friends have shared, it looked like the data had shifted. However, I just went and looked again and she's updated it as of this month. I'll have to compare the numbers coming in here against her new chart. Although, as you said, December does have a much heavier volume--and I'm not sure how many people are posting their pre-Thanksgiving numbers as requested--or their current numbers.

I know your observations about store volume and rank is right on the money- and that rank to sales fluctuate during the year depending on the volume being sold. But I have no idea how account for that.  :-[

You and Ed are so much better at that.  :-*

So how would I go about doing this? "What would be especially useful would be an overlay of sales-to-hit on top of store volume trends. But is anyone here tracking store volume via rank? I did some reverse engineering of rank last year in the period right after Christmas to determine the percent increase in volume at Amazon, but I haven't seen any similar studies from anyone else here. The difference in sales numbers to hit a given rank throughout the year is really more insightful than averages that, very much like a broken analog clock, are only accurate maybe 2 months out of the year. ;)"

What kind of data would need to be collected to do as you suggest above?

Would doing a monthly sales to rank data collection be enough?

 
Title: Re: Requesting data for an updated sales to rank chart
Post by: Kathy Clark, Author on December 17, 2013, 10:33:21 PM
This doesn't seem to take in all the factors.

I just hit #12 with sales for the 24 hours of 505 which doesn't fit the equations.

Doesn't the number of books involved in the ranking have an influence.  I know price makes a difference.  Being#1 with a 99 cent book isn't the same as a $4.99 book.  I have heard that the velocity adjustments made by Amazon factors this in.

It just seems that a simple matrix isn't a good estimator and probably needs supported with multivariate regressions.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 17, 2013, 10:56:37 PM
This doesn't seem to take in all the factors.

I just hit #12 with sales for the 24 hours of 505 which doesn't fit the equations.

Doesn't the number of books involved in the ranking have an influence.  I know price makes a difference.  Being#1 with a 99 cent book isn't the same as a $4.99 book.  I have heard that the velocity adjustments made by Amazon factors this in.

It just seems that a simple matrix isn't a good estimator and probably needs supported with multivariate regressions.

Do you mean you hit #12 in the US Kindle store with 505 sales, because no, that doesn't make sense at all. From everything I've seen and read that wouldn't be nearly enough sales to hit #12 in the kindle store. Or do you mean a different list? A genre best sellers list? The popularity list? A different store?

When it comes to main Kindle store ranking, price doesn't factor in. Ed explains what factors in a several posts back. I think you're thinking of the popularity lists. 
Title: Re: Requesting data for an updated sales to rank chart
Post by: LilianaHart on December 22, 2013, 04:47:26 PM
Do you mean you hit #12 in the US Kindle store with 505 sales, because no, that doesn't make sense at all. From everything I've seen and read that wouldn't be nearly enough sales to hit #12 in the kindle store. Or do you mean a different list? A genre best sellers list? The popularity list? A different store?

When it comes to main Kindle store ranking, price doesn't factor in. Ed explains what factors in a several posts back. I think you're thinking of the popularity lists. 

Right. The last time I hit #12 overall in US Amazon store with a $3.99 book I sold more than 3000 copies in 24 hours, so I'd imagine 505 sales is #12 in a subcategory or non US store.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Trish McCallan on December 22, 2013, 04:58:15 PM
Right. The last time I hit #12 overall in US Amazon store with a $3.99 book I sold more than 3000 copies in 24 hours, so I'd imagine 505 sales is #12 in a subcategory or non US store.

From checking out the poster's books, it looks like she may have meant #12 on the romantic suspense bestsellers list. Her book with the lowest rank was an RS, and quite often books at that spot on the RS list can bring in that volume.

 
Title: Re: Requesting data for an updated sales to rank chart
Post by: PhoenixS on December 22, 2013, 05:37:49 PM
So how would I go about doing this? "What would be especially useful would be an overlay of sales-to-hit on top of store volume trends. But is anyone here tracking store volume via rank? I did some reverse engineering of rank last year in the period right after Christmas to determine the percent increase in volume at Amazon, but I haven't seen any similar studies from anyone else here. The difference in sales numbers to hit a given rank throughout the year is really more insightful than averages that, very much like a broken analog clock, are only accurate maybe 2 months out of the year. ;)"

What kind of data would need to be collected to do as you suggest above?

Would doing a monthly sales to rank data collection be enough?

Sorry, forgot you asked a question! It really depends on what you want to do with the data. What kind of precision are you looking for? It would take several months to benchmark sales volume. Once you have an idea of the averages, then you can predict either sales volume based on rank or rank based on sales volume. But it such precision necessary? I've benchmarked freeloads to rank at different times to help demonstrate that readers weren't tired of freebies but that it was other factors responsible for depressed numbers. But even then I ballparked it using numbers from a handful of days scattered throughout the year since I was just looking for general trends.

I need to go look at Theresa's updated chart. I've been running with the numbers in my head when I advise our authors in advance what ranks they can expect from their sales numbers. I expect Theresa's chart is going to be close enough to work with for most purposes most authors would have.

For numbers to maintain a rank, I use a general 80% of the number to hit. But there's a history factor in there that makes it a bit tricky. The longer a book is at a rank, the fewer sales it needs to maintain. So to maintain a rank for a week after hitting it, still takes more sales than continuing to maintain that rank a month later.  Most of this stuff isn't as noticeable above #10,000, and it's only somewhat noticeable #1000-10,000.

And then other factors play in such as how many books are competing for a Top 100 in the store spot? Is there a KDD with 25 super-popular books on the same day there are 10 super-popular BookBub books and 10 popular authors releasing new books? The Top 50 for that day could all have 5000+ sales and #51 have just 1800. Or #51 could have 4000, which would on the very next day be enough to put it at #10.
Title: Re: Requesting data for an updated sales to rank chart
Post by: Sophrosyne on December 22, 2013, 05:48:26 PM
Here's what my experience has been for sales over a one-day period:

57 sales: went from #18444 to #5719

1119 sales: went from #5719 to #52

211 sales: dropped from #52 to #111

482 sales: dropped from #111 to #240, oddly enough. Not at all sure how that worked.

110 sales so far today: dropped from #240 to #336
Title: Re: Requesting data for an updated sales to rank chart
Post by: Mike McIntyre on January 07, 2014, 05:10:07 PM
I've sold 361 copies of a new book so far today (BookBub ad + Kindle Countdown Deal), and I'm nowhere near the corresponding rankings cited in this thread or on Theresa Ragan's chart. (I'm #1,210.)

This makes me wonder if it's weighted on past performance. Maybe 361 sales won't rank as high when I sold zero copies yesterday.
Title: Re: Requesting data for an updated sales to rank chart
Post by: PhoenixS on January 07, 2014, 06:45:33 PM
I've sold 361 copies of a new book so far today (BookBub ad + Kindle Countdown Deal), and I'm nowhere near the corresponding rankings cited in this thread or on Theresa Ragan's chart. (I'm #1,210.)

This makes me wonder if it's weighted on past performance. Maybe 361 sales won't rank as high when I sold zero copies yesterday.

No. Rank lags by about 4 hours or so. And there's a batch process that seems to happen around 4-6 pm PT during which time there won't be an update, and after which you'll generally note a sharp change. You can look at those 351 sales and determine a future rank in a couple of hours of around probably #420 in early January (would have been about #350 in early November). There are different numbers required to hit and maintain a rank, but most of the matrices you see are for hitting a rank from a standstill just as you're doing. Plus, as you sell more as the evening continues, your rank will improve till it peaks around 1-3 am PT, even though sales will taper off around 9 pm PT.

Which, with folk not knowing exactly when to capture sales and apply toward rank makes using unvetted data not the best choice for building a matrix. ;)
Title: Re: Requesting data for an updated sales to rank chart
Post by: Mike McIntyre on January 07, 2014, 07:05:25 PM
No. Rank lags by about 4 hours or so. And there's a batch process that seems to happen around 4-6 pm PT during which time there won't be an update, and after which you'll generally note a sharp change. You can look at those 351 sales and determine a future rank in a couple of hours of around probably #420 in early January (would have been about #350 in early November). There are different numbers required to hit and maintain a rank, but most of the matrices you see are for hitting a rank from a standstill just as you're doing. Plus, as you sell more as the evening continues, your rank will improve till it peaks around 1-3 am PT, even though sales will taper off around 9 pm PT.

Which, with folk not knowing exactly when to capture sales and apply toward rank makes using unvetted data not the best choice for building a matrix. ;)
Wow, thanks for that analysis, Phoenix!
Title: Re: Requesting data for an updated sales to rank chart
Post by: Sophrosyne on January 07, 2014, 08:05:36 PM
You're at 251 now.  :D
Title: Re: Requesting data for an updated sales to rank chart
Post by: Jay Allan on January 07, 2014, 08:55:00 PM
#70 was around 1,100 a day in December