I love Amazon’s KDP Select almost as much as I love Buster, my 13-year-old Pekeapom, a mix of Pekingese and Pomeranian, the dog you see to the right.

But my rescue pup dirties up the carpet every now and then. And in a way that’s exactly what KDP has been doing lately.

Last month, some alert authors smelled mismatches and oddities in the reporting of their page reads from subscribers to Amazon’s Kindle Unlimited service (KU). “Page reads” is the number of standardized Kindle pages read—this varying amount usually brings U.S. writers between 0.4 and half a cent.

The page read numbers roughly tend to follow a KU author’s retail sales curve (the books people purchase, rather than read through KU), sometimes a day or two delayed.

Longer books normally have more page reads, of course. New releases tend to see a steep jump in page reads that rises to a peak a few weeks or months out as people read through their KU queues. So when long books get released and their KU page reads stay essentially flat even while selling hundreds of copies—well, something is seriously wrong.

As a quick aside, I’ll give you my short bona fides. I make a living as an indie author. I’ve sold over 300K books (sales, not free downloads) and had over a million free downloads, averaging something like 10,000 copies a month if you count KU borrows as sales. I’ve grossed an easy six figures for the past three years, and most of the time I’m firmly in the middle of the Amazon’s science fiction Top 100 authors list. I have several years of data and experience.

Gripes multiplying on KBoards

For extensive discussion of the oddities, KBoards has a long thread with a lot of posts, only some of which are valuable or even germane. As usual, absent hard information from Amazon, there is a lot of speculation, but the gist of it seems to be this: Authors are losing, or not getting credited for, page reads.

Several reputable posters to the KBoards thread claim that they wrote to complain and adjustments were made to their page reads, always upward. Sometimes the changes were relatively insignificant, but at least one poster claims a 30K difference, roughly $140 worth. Now multiply that $140 by the days since the problem was noticed, and you can see how this could add up to serious cash.

What has KDP to say about this? After I complained about my own page reads, which I will detail later, I received the form letter below, the same as others on the KBoards threads posted. I’ve added the bold for emphasis.


I understand you are concerned about lower than expected pages read in your reports. We’ve thoroughly reviewed all of your KU/ KOLL borrows and can confirm that the pages read displayed in your dashboard are accurate.

Emily Taylor
Kindle Direct Publishing

So, in essence, the first layer of customer service denied the existence of a problem, which seems to be normal for KDP. And for Buster, with that cute innocent face.

Now, I’d been noticing my own page reads falling off for the last month, but they really took a downward trend about two weeks ago, even while my retail sales stayed steady. (As I mentioned above, retail sales and page reads curves tend to be roughly similar.) Generally, within plus or minus ten percent on any one day, I tend to see about 500 pages read per retail book sale, usually more, but I am being conservative here.

Thirty percent decrease in page counts in the last two weeks alone

Yet my average page reads has fallen to about 350, which is a 30 percent decrease in the last two weeks alone, with no discernible reason for this happening. In fact, seasonally, page reads and sales tend to rise in September as children go back to school and the adults return from vacations. The weather tends to encourage indoor activities like reading and the holidays provide time for many to catch up on books. Note, if my retail sales had fallen off that much, I’d chalk it up to the vagaries of the market or my lack of effective promoting. But retail sales have stayed very much the same, or risen slightly, despite no particular marketing on my part.

Here’s a screen shot of my sales. Compare for yourself. The red line is retail sales; the blue line is pages read.


Still, this could simply be an extended statistical anomaly. Sure, similar things have been noted by other authors, but we could all simply be experiencing some kind of mass reader burnout or disenchantment with KU and a switch back to buying retail. Why avid readers would suddenly give up their $9.95 per month book fix to pay the same price for one to five books on average, I couldn’t say, but hey, it’s possible. Or could some kind of legitimate change in the algorithms have adjusted everyone’s numbers and how they were reported (although if that were true, you’d expect a quick one-day adjustment, followed by a plateau)?

Note that the blue line finally seems to be plateauing at this new lower level, despite the big red spike on the October 1.

What is that big red spike, you might ask? That was me releasing not one but two box sets of 3000+ pages each (long books are capped at 807 print pages/3000 Kindle page equivalents for KU payment purposes in order to discourage scammers). So, you say, maybe all is well and I’m starting to pick up significant page reads from those box sets? Nope. It’s hard to believe that Apocalyptic Fears III, a box set of ten full-length novels, has sold over 200 copies since being released six days ago, yet has not achieved a mere 3000 pages read. That’s right. By the math of my recent average of about 350 page reads per retail copy sold, you would expect 70,000 pages read, the equivalent of about 23-24 of those books, over the last six days.

Actual count? 2665 as of this writing, which is about 4% or about 1/26 of expectations. Here’s the screen shot of this individual book:


It’s hard to see, but there were actually four (4) pages read on the day it was released, October 1 (the sales on the 30th were pre-orders). So someone opened the book up and…you’re kidding, right? Read four pages? That doesn’t even take you past the front matter to the first story. So again, something wrong.

Then there are over 2000 pages read. So far so good. But then the page reads drop to under 300 the next day, and then under 100 for the next three days, despite over 200 books being sold at retail. Sure it’s possible that only, say, five or ten people in KU got the book while 200+ people bought it retail, but that’s extremely unlikely. Most of the time, the math says KU borrows and retail sales numbers remain within 20 percent of each other.

One other note: my page reads for non-English books have remained steady and apparently unaffected. Only my English-language books seem to be affected, so removing the non-English books from the equation would result in an even more dramatic drop than I’ve outlined here.

My empirical test: First-hand evidence of an accident on the carpet

So how could I really prove something was going wrong? Well, empirical testing is always great, so I opened up my own brand-new KU account and “read” the entire book of Apocalyptic Fears III. I even flipped through it page by page, albeit quickly, in case Amazon had changed their reporting system to not report skipped pages. i.e., vs. simply going to the end using the TOC. I’ve never done this before. I’ve never even had a KU account before, though my wife has, and then only to check the quality of our own books from a reader perspective.

I opened the same book in my Kindle reader and the Cloud reader and “read” through it to make sure KDP had multiple chances to record my pages read. That should have resulted in 3000 pages read. Just to be doubly sure, I did the same with Apocalyptic Fears IV, another 3000-page tome. I also closed and reopened both books, in case there was some sort of report triggers that needed tripping. Then I waited a few hours. Sometimes, KDP delays reporting for several hours. Occasionally, even up to 24 hours. So I waited 24 hours.

Kazam! The page reads showed up. But, subtracting out those page reads, the daily number for 3000-KENPC books which had sold hundreds of copies remained fewer than 100. PER DAY. For probably at least 100 people that have gotten it through KU, if the world has kept turning on its axis. So again, there’s an apparent software glitch.

In some of the recent posts to the KBoards thread, authors have noticed that books opened in the nifty new Page Flip mode don’t seem to register any page reads. In other words, people are reading the books, but authors are not getting credited with more than one (1) page read.

So, I did another experiment. Using my KU account, I borrowed one of my own low-selling books that gets fewer than 100 page reads per day, normally zero. I used my iPhone and the Kindle app, so that the book opened automatically in Page Flip mode and kept it there. I paged through the entire book. Result? One (1) page read exactly, up from zero (0). Yup. One. Just as many others are reporting on KBoards and elsewhere.

Here’s the next email I got from KDP Executive Customer Relations regarding the problem (with bold I’ve added). Note that they were not specifically addressing my experiments, but only responding to my complaints of at least two days ago:

Dear Dave,

My name is Ann Fairday with Kindle Direct Publishing Executive Customer Relations. Jeff Bezos received your message and asked that I reach out to you on his behalf. I apologize for the delayed response to your concerns about KENP [Kindle Edition Normalized Pages].

As you may have seen on our forum post https://kdp.amazon.com/community/ann.jspa?annID=1144, we completed the results of our monthly audit of September pages read data.

As the forum post indicates, total audit adjustments for the month were an increase of roughly 2% of pages read (though the amount will vary from author to author). We are currently updating reports and changes should be visible within the next day.

If you have any additional questions about your account, please contact us at https://kdp.amazon.com/self-publishing/contact-us.

Best regards,

Ann Fairday

You will note that, unlike the first email, they are finally admitting there was some problem. However, when I tried to open that link in the email, it reached the KDP forums but there is nothing there. The post has apparently been deleted. Hmm.

Others have reported that KDP emails to them noted that their “2%” was restored on September 30. I checked my own September 30 page reads and they jumped by almost 10,000 on that day, an increase of about 30 percent day-over-day for me personally, before dropping by about 5000 the next day to the new plateau. This would tend to indicate I was credited with about two percent per day for 15 days, exactly in line with predictions. However, my page reads have been about 30 percent lower than normal for those 15 days, so in reality, I’ve been credited with less than 1/7 of what I’ve lost. But the key takeaway here is that, yes, Virginia, there is a problem. And I don’t believe it’s merely two percent.

Here’s an excerpt of what another KBoards poster reported getting back in a KDP email:

“Thanks for the recent questions from some authors about how Page Flip is being used by customers and its possible impact to pages read. Page Flip is designed to make it easy to explore and navigate in books while automatically saving your place, and that is how customers are using it. We checked for effects on pages read before launching Page Flip, and investigated it again to re-confirm that there is no impact. We do not see any material reading volume happening within this feature, but we will continue to monitor it closely.

“We greatly appreciate the questions and feedback we get from author; please keep it coming. If you have specific questions about your account, please contact us at…”

Clearly, some people at Amazon have not delved deep enough. At least some apps on some devices (such as my Kindle app on my iPhone) do not report page reads.

Interestingly, while my two major new releases of box sets have not seen any significant rise in page reads, just today my overall page reads have started climbing dramatically. As of this writing, they have jumped over 30 percent in less than a day. Coincidentally (?) most of the uptick seems to have come from seven (7) books that I made alterations to and re-uploaded yesterday.

Having noticed that, I’ve just re-uploaded my test cases (the new box sets) and also another box set, to see if re-uploading them has any positive effect on pages read. If the page reads take off tomorrow like gangbusters, at least I may have identified a potential solution without really understanding the problem.  Right now, I feel like I am flailing around with partial information and little recourse except to keep pleading with KDP to find out what’s wrong and fix it.

Who has all the information needed? Amazon does, of course. And they don’t seem to be telling. There’s been reasonable speculation that KDP is underplaying or keeping silent on this issue at the advice of Amazon’s legal department. Some of the authors on the KBoards thread say they’ve personally been told the legal department would contact them, but they have received no callbacks or contact. Here’s hoping that Amazon gets its act together soon.

Fodder NOT for Amazon bashers

For those who point and say, “Aha, I told you so, you should have stayed out of KU, and wide on all vendors,” I get it. I agree in principle. I wanted to stay wide. I tried to stay wide. But, considering my income nearly doubled last year when I went all-in with KU (for my science fiction titles, that is: interestingly, my mystery titles do better wide on all vendors), I still have to go with Select/KU, despite the issues. I’m certain that we’re talking about software glitches—or human errors—rather than any malicious intent at Amazon. What’s more, I would hope that Amazon’s detractors in the business world not depict this situation as corporate iniquity, just “mistakery,” so to speak. Yes, Amazon is the Kong of digital publishing. Every time they twitch they may knock over 18 monkeys. But that doesn’t mean that Kong is bad. Kongs can simply make honest mistakes—I just want Amazon to fix this one ASAP.

As long as my page reads stabilize or recover, I’ll still be a big fan of KDP and Select, just as I’m a big fan of Buster.

Except when he messes up my carpet or keeps me up at night.