Introduction
I'd like to focus on my own experience with Amazon AMS ads over the last 9 months.
For information about Amazon AMS ads in general, look for free introductory courses online, and I recommend some of the for-fee courses by people like Mark Dawson for in-depth guidance.
For context, here are some basics. (If you're already familiar with AMS ads, you can skip this.)
- You can only run ads for your own book and, at this time, only for the US. Other regions are anticipated, e.g., the UK.
- There are “Sponsored Product” ads (which show up at the bottom of product searches and below the “also-boughts”) and “Product Display” ads (which show up near the “Buy” button and on Kindle screensavers) — I'll only be talking about Sponsored Product ads
- Each ad is a “campaign”. You supply up to 1000 keywords or keyword phrases for each campaign, and a maximum price you're willing to bid for the ad. You compete with other advertisers to show your ad prominently in its display area.
- I call a cluster of campaigns for a single product (to use more than 1000 keywords) an “ad farm”.
- You supply a 150-character ad copy, and Amazon supplies the book image from your book listing. There are restrictions on what you can claim in the ad (e.g., “Bestseller”).
- Amazon will suggest some “automatic” keywords of minor usefulness, but I will be talking about the “manual” keywords I supply
- You are charged the bid amount each time someone clicks on your ad, whether or not they buy your book once they look at the book's page. You are not charged for impressions (the display of your ad).
- You set a daily budget for each campaign which caps the maximum spend. Raising the budget for a successful campaign does not necessarily make Amazon display the ad more frequently — it is difficult to really maximize the use of successful campaigns, once identified, aka “Amazon won't spend my money”.
Why Amazon Ads?
This post arose in response to an innocent question on a forum about “Why should I care about Amazon Ads?”
This is why.
I write fantasy, mostly, and among these 24 titles there are two completed fantasy series. Series 1 (abbreviation ANNWN – The Hounds of Annwn) is 4 novels and a story collection, and series 2 (abbreviation CHAIN – The Chained Adept) is 4 novels. Each series also has book bundles – one for books 1-2, and one for the remaining books. Series 1 is in green, above, and series 2 is in yellow/orange. Series 2 is more attractive to new buyers (wizards) than series 1 (wild hunt), but both are in the 4-5 star ranks with favorable reviews, if not as many reviews as I would like.
My books are priced in the middle of the fantasy genre overall (indie & trad), where book 1 is $5.99, book 2-4 is $6.99, and each 2-book bundle is $9.99. This graph shows the results for full-price books (I rarely do discounts). The lower your book price, the harder it is to make money from advertising.
I do very little marketing of the newsletter variety, so this is a nice extreme example of the difference between doing nothing vs running AMS ads. You can see the immediate spikes for new book releases, followed by the long decay as no new books come out.
I run ads for book 1 of each series, and for the 1st book bundle of each series: 4 ad farms in all, each composed of several campaigns. It is harder to make money from advertising for single books which are not part of a series.
Keywords
One of the big challenges is coming up with a lot of keywords for your books. There are different philosophies of where to look.
Authors like you/titles like yours – your opinion
Since you are familiar with your own genre, possibly over decades of reading in it, you have your own list of authors and titles where you're pretty sure a reader of those would also like your book.
Readers like yours – their opinion
The “also-bought” section of your book page, or a schematic of that from a product like Yasiv, shows you what readers who bought your book also buy. That's a list of authors and titles reflecting current buying practices.
Current buying trends by keyword – finding potential readers
You can take your current metadata keywords, the handful of things you think are important about your book, and feed that into Amazon lists of books by detailed genre, looking for popular books that also consider those keywords important. There would naturally be some overlap with their readers who have already discovered you (also-boughts), but mostly not.
Searches made by others – unfiltered chaos
Taking the authors and titles which you think might appeal to readers, you can do searches using products like Google MerchantWords to see what people have typed in when searching. This will turn up every variant of “Book X” or “Author Y”, such as “Book X TV Show” or “Book X volume 7” or “Author Y obituary”.
I've tried all of these approaches. My first ad farms, which I retired in January altogether, went for volume of keywords based on searches made by others. This was founded on the “Authors like you” approach as seeds, and a certain amount of “Readers like yours”. There were 20000 keywords in each ad farm.
Halfway into January, I retired all the existing ads and made a new set (10000 keywords/ad farm) based on successful keywords from the prior campaigns and the “Readers like yours” and “Current buying trends” classes.
One key takeaway… It doesn't matter if I have a perfect list of classic authors whose readers would like my books, if no one is buying those authors currently and seeing my ads. It's much better to find authors who are currently being bought whose readers buy my books now, or might buy my books based on similarities of metadata keywords.
Covers
Most retailers will not let you use 3D images for book bundles, but Amazon is the exception. You will certainly want to display book bundles as 3D images for the instant identification that it represents a multi-book purchase.
Ad Copy
It's hard to cram your message into 150 characters, and you have even less space for a bundle, since you need to make sure the message supports the multi-book nature of the product. Here's my current ad copy for the first book and first bundle of series 2.
Book 1 of series
A strong wizard with unanswered questions and an unbreakable chain around her neck. Who was she before it happened 3 years ago, and why was it done?
Book 1 of the series bundle
A strong wizard with unanswered questions and an unbreakable chain around her neck. Who was she before, and what happened? 1st 2 books in the series.
Tuning
While you want the ads to generate more sales, the primary consideration is to prevent them from generating unnecessary costs. That's caused by clicks (for which you are charged) which do not result in offsetting sales.
This happens for many reasons… Your ad copy might be misleading. Your ad might be running on book pages that are too far removed from your target audience and the idle curiosity of the clicker costs you money. The content of your book page, once they arrive, might be disappointing (full book description, reviews, “look inside” issues).
You want to find those keywords that cause unproductive clicks and suspend them, before they cost you any more money. This means you must monitor the performance of the ad campaigns on at least a weekly basis to make sure that some of those keywords are not harming the efficiency of your campaigns. If you don't do this, you will find your advertising is costing you more money than you are earning from it.
Once you have enough data, you can also fine-tune the productive keywords that do result in sales. After all, if it takes 20 clicks to generate 1 sale, you'll never make money. So make sure that the amount you spend in clicks is reasonable for the amount you earn.
There's also the question about monitoring impressions, assuming you have checked that your book cover seems attractive and sends a good genre signal. Impressions don't cost you anything, but do you really want your ad appearing on book pages for unrelated genres, and so forth, where it has no relevance at all? While you're monitoring your keywords, look at the impressions, too. If you have more than, say, a few thousand impressions for a keyword and no clicks, I would suspend it — why clutter things up? While no one knows exactly how Amazon's algorithms work, it can't be a good thing to have keywords generating meaningless impressions when, perhaps, better keywords might be chosen by the algorithms.
Staleness
AMS ads do go stale and become less effective. As you can see from the graph at the beginning of this article, the ads ran smoothly from April through August, and then began to sag.
The causes are unclear. I suspect they include over-familiarity (the same ideal targets have seen the ad too many times), but no one actually knows. The remedy is to retire the ad and remake the ad farms with some of the surviving keywords (the ones not suspended) and perhaps new ones, with a modified ad copy.
I retired my original ad farm in mid-January, when I brought out the bundles for series 2, and am on track to reach the same heights I reached in April by the end of the month.
Statistics
AMS provides notoriously poor statistics. You can only track the lifetime stats for a campaign, that is, you can't get daily or even monthly cuts — only the cumulative as-of-now numbers. Everyone who is serious about trying to track periodically to support decision making is doing some form of creating their own database of the statistics, since there's no other way to track changes over time.
I use an Excel spreadsheet into which I paste the raw data from AMS every few days, and each month I copy the month-end statistics that I care about into a separate column so that I can do month-over-month comparisons. I also aggregate all the campaigns in an ad farm for one product into a single row. To maintain lifetime stats, I include campaigns that have terminated.
This is my complex and detailed dashboard. Explanations below.

Number of units sold
The AMS system is used for all Amazon products, not just books. The assumption it makes is that each ad stands alone and should be profitable in its own right. In AMS terms, that's an ACOS (Cost of Sales) figure of 70% or less (the royalty you receive for a book) — if your cost for the book ad is below 70%, then you're making money.
At least, that's the theory. But there are lots of problems with the data. To begin with (button 2), AMS not only lags in reporting sales attributable to a click on an ad, by a couple of days, it misses tons of them. I sell only a handful of books/month outside of the AMS system since I currently do almost no marketing. I track my actual sales very closely using the KDP dashboard, which I know to be accurate (that's the basis for Amazon paying me royalties). But you can see that over the lifetime of these ads, I have sold 710 units of the advertised products, and AMS has only reported 581 (an estimate based on dollars/unit price). That's a very large discrepancy.
The lag is also important, for new campaigns. You can see by the dashed red arrow that the brand new book bundle for series 2 has sold 22 units, but AMS has only reported 17. Over time, the lag becomes less important as the numbers get larger, but the gross inaccuracy of the count for button 2 remains.
Series vs Product Advertised

I have very good information on how my series upsell, how many readers read through an entire series. Both Series 1 and Series 2 have very similar curves. On average, a reader who buys book 1 of the series for full price will buy about 1.16 more books in the series (some buy none, some buy them all).
So for every sale I make of the advertised product, I know how much my additional sales will be for that series. As you can see from button 2, I sold 1535 units altogether related to the advertised product sales of 710 units.
I calculate the success or failure of my AMS advertising based on the whole-series numbers, not just on the advertised product.
On button 1, I post the current KDP results for the advertised product (total 77) vs the whole-series units (138). I know my average earnings/unit ($4.44), and I use that to calculate my profit from advertising (button 3) of $978.62. AMS, using just the advertised products and its inaccurate units-sold count would count that as a loss $3256.91. Makes a big difference where the numbers come from and what's included.
My lifetime return from the ads is 16.8% (button 4), but that includes some negative months, as you'll see below. For my most recent whole-months, November and December, my ROI was running at circa 60%.
I use a high maximum bid price for my ads ($0.51) because I want to ensure that I'm on the first page of ads whenever I appear. The average price I pay per click is only $0.33, even so, thus it seems like that's a worthwhile investment.
ROI – Return on Investment
So, what does this look like, month over month?

KDP Units per Month
Here's where you can see the 1.16 additional units on average of series read-through. The actual sales that are triggered by the ad is represented in orange.

AMS Spend vs KDP Estimated Proceeds
As you can see, I stopped checking the tuning of unproductive clicks in September and October, and the unnecessary costs ate my profit for that period.
I ruthlessly pruned the ad farms, and November and December shot up in profitability, even though the ads were generating fewer impressions over all (going stale).
Cost per Sale vs ROI

My overall lifetime ROI for the ads is only 16.8% (button 4 in the dashboard above) and I haven't earned a lot per book, but what a difference looking at it monthly makes.
I calculate the Cost of Sale at 100%, since I'm using actual earnings from KDP rather than the AMS ACOS statistic which is based on retail sale price, not earnings.
November and December are just what you want to see: the cost per sale (advertising) going down, and the return on investment going up. If my lifetime ROI were 60%, as November and December were, that would mean a net earning of $2.23 per book advertised.
Conclusions
Amazon AMS ads have a significant learning curve, and it costs money if you make simple mistakes, so I recommend treating this seriously as a marketing investment and paying for a for-fee course to make sure you don't blow more money on doing this poorly. (You can always use the free online introduction courses to get some feel for what you're getting yourself into.)
You will need to pay for ads now whose related earnings don't show up for 60 days, so it requires upfront capital.
You have to keep an eye on how your ads are doing; monitoring and tuning are a requirement for efficient cost control.
If your books are priced too low, or not part of a series, it is much harder to make a profit using AMS ads.
Because the AMS reports are so ill-suited for decision making, you'll need a good handle on your real sales figures from the KDP Dashboard, including series upsell results.
Most of the courses speak in terms of getting good results for book 1 of a series and treating the additional series units that arise from that as gravy. I'm treating the entire series as a unit (its sale vs the costs of advertising book 1) for my statistics. That means that if you looked at my results for book 1 of series (advertised product only) they wouldn't be nearly so attractive, so be cautious when comparing this case study to the standard courses. In the end, I care about the overall result, which is why I do it that way. But as a consequence, you can see from the updated statistics charts below (bottom left) that the proceeds per book (for each book of the series) are reduced, too. My opinion is that a lot of people see my books at their standard full price in these ads and that is where I'm pegged in their mind for my brand. When they come across me again, outside of AMS ads, they will not be surprised at my full price and, if they had a good experience, should be reasonably likely to buy. The fact that I earn much less for the books sold via the ads for book 1 is less important to me in the long run (as long as it does remain profitable, of course.)
There are limits to how many additional units you can sell even with highly successful ads — it can be hard to make Amazon take your money.
Volume is its own reward — more reviews, more exposure, etc.
Don't forget the psychology of moving more units… I've sold more books using AMS ads in the first 2 weeks of January than in all of 2015, when I only had the one stale series. That may not be a lot, but when you despair over “no action” inbetween releases, it's a great boost.
UPDATED CHARTS FROM MID-JULY 2018
Notice how volatile results can be — the exact same ads are running in May, June, and July of 2018, and they are equally profitable, but the impressions are wildly different. That is not something you can control. So don't panic if the units sold swing up and down as long as the statistics (ROI, etc.) remain good. When impressions go down, units sold goes down, too, but then so does the money you spend on ads. At the end of the day, you want profits more than volume.

The number to focus on is the monthly and lifetime ROI from the chart on the bottom right. Don't get spooked (or elated) too much by volatility.
Unfortunately, it always looks as if the primary job is marketing, not writing. You are doing BOTH – and selling lots of units. And still have to be very careful that your marketing expenditures don’t consume all your profit.
Thank you for sharing your data – and your analysis. Given that it took me 15 years to write the first book, and it’s already been over two years for the second (and the end is not in sight, though I hope it will be this year), volume isn’t going to work for me.
But I’m always fascinated by what indies are accomplishing. Kudos!
Thanks so much for sharing your statistics, Karen! I have only two offerings at the moment–novel and related short story–and have experimented with them with both my own keywords and Amazon’s own choices (secret, of course). I find that Amazon’s picks for my novel yield purchases of my paperback, which has a much better ROI for the advertising. If only I could leverage that to the best effect.
Your many offerings inspire me to do some production on nonfiction titles I have from a couple of decades of online writing instruction. The material is all there, ready to be repackaged and put on sale (and marketed, of course). You’ve got my wheels turning!
This is a very good case study.
You make a good point in that if you plan on making money paying for placement on Amazon, or even Google or Facebook for that matter, you got to have a strong branding / upsell+cross sell strategy in place.
Selling one book to one customer one time is not going to work when other people you are competing against are doing more.
The one thing I would add to this is I would consider even more outside the box upsell monetization methods such as running a fan or early release club offer or anything else crazy like that you can think of to get more out of the top 10% of your customer list that you do now.
[…] AMS ads and has kindly agreed to us to republish excerpts from it. You can read the whole article on her blog, and we really recommend that you do […]
Thanks for sharing this , Karen. You’ve presented a lot of great comparison data. I think that this shows AMS ads are not a get rich quick scheme. 🙂 A lot of people think it is, but the problems with the platform reporting, the inability to scale in any predictable way, and the need for higher priced products all point to it being an ad platform for product maintenance at a higher level than it is without anything.
Your data matches mine. I ran ads with similar results Sept. through mid November with similar results. I stopped them during the holidays because I was afraid the bid cost to compete would be too much. Interesting about you bidding 51 cents but being charged 33. I’ve been bidding 35 cents thinking it was just enough over the average 25 cents to get better placement.
I do have one question for you. Are you in KU? Because you are looking at read through, I wasn’t sure. I’m wide myself, and thus the numbers seem more like what I can expect. I’ve heard that people who are in KU get better return on their AMS ads because of page reads.
Again, thanks for taking the time to post and share this.
Interesting to hear that you had a similar experience, Maggie.
I’m not in KU — completely “wide”. What I meant by “read-thru” was how people read through an entire series of books: if they buy my book 1 in either series, on average they buy about 1.16 more books in the series. That number is actually creeping up a little bit: more people will “read-thru” a series if all the books are out and available than if they started with book 1 while the rest were still being created.
Thanks, Karen! I’m glad to hear that your stats reflect wide distribution. I do understand read through. I’m fortunate in that my read through tends to be two books past the first one. Now that I have a boxset for Books 1-3, and 3 more past that I’ll be interested to see if that makes a difference too.
[…] Was this worth doing? I think so. Here’s the result of doing AMS ads. […]
That was an excellent report. I’m just starting and was becoming enamored of AMS ads until I saw that one off (non-series) do as well. There is a lot of other good info here. I’ll read it more thoroughly. Thank you, Karen.
I’m glad you may find this helpful! My January/February ended up back at the levels of where I started, and my ROI overall is continuing to improve.
It seems like a hell of a way to advertise something. I have published over 250 titles.
In the “old days” I was very successful mail order advertiser of my titles. Some ads pulled 300 to 500 orders. A price was set by the publisher of the magazine you were advertising in. You could haggle o over the price and so ads would work and others would not. But there was one price and you knew the basic circulation of the magazine (publisher often fantasized the number if they were not audited by the ABC (which only the big publishers were, cause you had to pay for the service) of using their accounting services). This is like flying blind. I dont see how a publisher could come out a head. But just getting the clicks is hard enough. I have uped my bid to 70 cents to see if it makes a difference; cause I was getting hardly anything on a non fiction subject that is of general interest (i.e. paranormal, UFOs), fringe science)
I found the ads are very sensitive to exactly what the ad copy is, and the cover. If they aren’t generating clicks, tinker with that, and good luck.
This is great info! I really appreciate you putting it all down in one neat package for those of us starting in the AMS world. I’ve been doing the ads for about a month now. I find it inaccurate at the least (disingenuous at the most) that Amazon gives your conversion rate based on retail sales not taking the royalty into account. It pretty much makes that stat worthless. Thanks again!
Conversion rate is about the number of units that sell, not about the price or royalty.
ACOS is based on the retail price, true, but that’s easy enough to figure for Amazon, since it’s 70% for $2.99-9.99, so the ACOS you want is… less than 70%. (Or 35%, if you’re determined to advertise books priced at that level).
Good luck!
Thanks so much for this, Karen. Really in-depth and useful information I will certainly put into practice. Great post.
Thanks. I’ve added an update at the bottom for the current graphs, from 6 months later.
Thanks for the detaled breakdown. Great info.
One aspect of AMS ads you may not have focused on here is optimum methods of scaling up ads, apart from ‘tuning’ individual ads. As far as I can figure out, the only way to scale up to produce profitable impressions, clicks and sales is to be constantly generating new ads with new keyword lists, eventually having hundreds of ads running at the same time that are regularly being added to.
Would you agree with that?
That’s not how I do it, though it may well be valid and certainly I hear of lots of other people doing it that way.
You’re correct that I haven’t solved the “scale up” part of the equation.
I prefer to retire old ad farms and concentrate on new ones (keeping the keywords that are working and adding more). I do think that ads get stale, possibly because they’ve been seen too many times, and that makes them inefficient, so (if that’s true) why keep inefficient ads running which cost more?
I know that there are people who can get $10000 of ads running — I advertise 4 of my titles at a high-click price ($0.51), but the most I’ve ever been able to spend in a month is $1000, and that only once. More typically, I spend circa $500. So I don’t know how the $10000-spenders achieve that.
I have circa 9000 keywords/ad farm now. In the past, I’ve had 20000. That gives me 36 ads to monitor weekly. If I’d kept every ad I ever had alive, it would be 150 ads to monitor by now.
When I rebuild my ad farms every few months, I find it can take a couple of months before each ad in the farm gets any impressions — it’s very uneven.
Certainly, advertising more books would help, but right now, with only the 2 completed series, the best targets I know are first-in-series and first-in-series-bundle. By that logic, each new series will bring me 2 more ad targets when the series is done — though for a series “in progress”, I will also advertise latest-release to remind my existing readers. (I’ll have a chance to test that out in a few months when I do quick releases on the first few books of a new series.)
How do you monitor hundreds of ads? Do they use the same keywords, or is each keyword only represented once in the ad farm for a title? Do you run ads for series entries which are not “first” or “latest”?
Hello Karen,
One of the problems you have with amazon ads is the marketing stats you get. If you ever advertised with Google Adwords you know what I am mean. The table data you get is accumulated and without downloading the campaign stats every day you are out of luck figuring out how to improve your ad performance and get a better ACoS.
I started to make screenshots to compare two days and get my daily stats. After that, I decided to brush up my programming skills and created some nice charts. This helped me to improve my ads, finally turn a profit, and I am happy to announce that this tool is now available for other authors as well.
https://www.bookadreport.com/
cheers
Thomas Herold
I quite agree — the statistics are notoriously poor.
I did my own (just for me) spreadsheet development which is now so customized I’m unlikely to switch, especially since I track whole-series sales, not just the advertised product. Still, I’ve encountered many people who are looking for your sort of solution. Good luck with it — it looks like a VERY useful tool. Planning to add Facebook ads to the mix?
Hi Karen,
Thank you for the feedback and the extra note. No Facebook ads yet – maybe in version 2 or 3.
cheers
Thomas Herold