Welcome back for Part 4 of the Ultimate Guide to Book Marketing! If you’re just stopping by and want to start from the beginning, you can find the complete series here. Each part stands alone, though, so if you’re just interested in a particular topic, feel free to jump in wherever you see fit.
First, a little refresher on our Ultimate Book Marketing Formula, which forms the backbone of the guide:
We’re about to take our first foray into traffic and promotion (you know, the stuff that usually come to mind when people hear the word marketing). But this part of the guide isn’t about how to direct a steam of prospective readers to our book page—the various promotional and advertising options that you can use to drive readers to Amazon and other retailers will be discussed at length in Part 5. Right now, however, we’re going to dive into how you can amplify all your marketing efforts by leveraging the power of the Amazon algorithms (otherwise colloquially known as “the algos”).
Now, if you’ve been an indie author for a bit, you’ve probably heard a lot about the importance of Amazon’s algos. But why do they matter so much?
Simple: if you know how they function and what they reward, you can put Amazon’s marketing machine to work for you. And Amazon is the most powerful bookselling force on the planet. Even if you spent $10,000/mo on Facebook, you couldn’t match the power of Amazon’s organic visibility. By understanding the algorithms, we can structure our marketing campaigns and launches to maximize the impact of every promo dollar.
Sound good? Then enough preamble. Let’s demystify the algos.
What are the Amazon Algorithms?
Amazon is essentially a giant AI recommendation machine. When authors refer to the algos, they’re really referring to a specific subset of the thousands of algorithms and rules currently running behind the scenes on the site. That is, they’re talking about the book recommendation algorithms.
Unlike other retailers, almost all the merchandising on Amazon is regulated by automated mechanisms. This is one of Amazon’s major retailing innovations. By taking staff members largely out of the merchandising equation, their system can update recommendations in near real-time, churning out millions of up-to-date, highly specific recommendations to Amazon’s many customers. Not only do these occur without human intervention, the system often makes far more accurate guesses than a human can about what a prospective buyer likes based upon Amazon’s massive stores of customer data.
First, it should be said that there is no single algorithm governing book selling. Amazon is a complex organism with a ton of code interacting in complicated ways. Technically, the main beast driving the show is called A9. Diving into the technical details of A9, however, isn’t what this guide is about (after all, this is a guide to selling books, damnit, not an engineering white paper).
Instead, by analyzing patterns, trends, and how Amazon’s site is constructed, we can peek behind the recommendation engine’s curtain.
And then we can put that information to use to sell more books—without an advanced degree in statistics or computer engineering.
So what we know about Amazon is this: the algorithms aim to show buyers what they want to buy when they want to buy it (or perhaps before the customer even knows he wants to buy it). Their engineers are constantly refining and updating the code that governs these recommendations, always trying to find new ways to generate more sales and increase customer satisfaction. The latter part is really the key: Amazon’s core drive is the customer experience. This cannot be overstated, and should not be ignored; it is built into the framework of their cultural DNA. They are obsessed with delighting customers because they know that providing the customer with an incredible experience is the key to repeat business, word of mouth, and dominating the retail landscape.
Thus, they strive to recommend hyper-relevant items not only because are customers more likely to purchase them (e.g. relevant items have higher conversion, in marketing parlance), but because it people love finding exactly what they want without having to do all the legwork of tracking it down.
How does this apply specifically to books, then? Well, Amazon has a number of automatic recommendations triggered by factors we’ll discuss shortly. But first, let’s look at just a few of the places these recommendations appear:
- The Also Boughts
- Merchandising Emails
- The Bestseller Lists
- The Popular Lists (known as the “Pop” lists)
- Other automated merchandising locations on the site
Examples of Amazon’s automated recommendations hard at work:
The Also Boughts
Kindle First Email w/ Automated Sci-Fi/Fantasy Recommendations
The Top 100 Dark Fantasy Bestselllers
There are probably over a dozen of these little merchandising spots located in nooks and crannies all over Amazon; this is merely a taste of how powerful the recommendation engine is.
And if you can convince Amazon that readers are buying and enjoying your book, they’re happy to put this engine behind you as an author.
A Note on the Pop List
You’ve probably heard the Pop Lists (short for Popular Lists) mentioned before. To check them out, you can hit this link (nicholaserik.com/pop) to see the cozy mystery pop list (it’s at the bottom of the page). Getting there on your own is cumbersome and not really necessary, as buyers don’t really browse them (although certain devices do default to them). These lists, however, give us a a little peek into the Amazon recommendations. Because unlike the bestseller charts, which are simply a list of what’s selling best, the popular lists are often much more reflective of what Amazon is pushing through their automated recommendations.
They’re also based more upon revenue, so they will often differ dramatically from the bestseller lists. Thus they’re often dominated by trad-pub books, which have higher prices. Amazon wants to generate revenue. So while the pop list isn’t useful for browsing, they push the books on the pop lists – e.g. their top revenue generators – via their recommendation engine emails and so forth.
This is not perfect. No one knows the exact algorithm that goes into the pop lists. Free downloads used to be counted on an equal basis with sales, then that switched to about 1/10 of a sale around 2013. Now it seems to be much less than even that tiny sliver, although free books are still counted. Without knowing exactly what’s going on under the algorithm’s hood, it’s impossible to say for certain that spot #2 is making more than #3. But spot #2 is almost certainly making more than spot #20.
Anyway, this brief aside is just so you know that the bestseller lists aren’t the only (or perhaps even the primary) driver of recommendations on Amazon’s site.
The Key Things to Know
What do Amazon’s algorithms like to see? Well, to be clear, no one knows for sure. They’re somewhat of a black box, and obviously Amazon’s not opening the doors to show everyone exactly how they function. Nonetheless, as mentioned earlier, we can examine patterns in the data to get a general idea of what they like, including a book’s overall popularity, the buyer’s purchase history, price (e.g. how much a product makes for Amazon), verified reviews (apparently; which goes against the grain of popular belief) and probably hundreds of additional factors.
In practice, most of this stuff doesn’t matter when it comes to generating visibility via Amazon’s recommendations, also-boughts, popularity lists (now referred to as “Featured” in Amazon’s search options), or bestseller charts; ergo, variables like reviews, which are allegedly factored into the algorithm, don’t mean jack when it comes to rank.
So let’s talk about what really matters.
Three things matter more than anything else for tripping the algo wires:
- Sales volume & velocity (this is the main factor) (key for pop lists/bestseller charts)
- Sales consistency (key for pop lists/bestseller charts)
- The sample of people who buy your book (key for Amazon’s automated emails, also-boughts and on-site merchandising
The algo also factors in:
- Newness: promotes new content more readily than old backlist
- Sales history: if a book has a consistent history of poor sales, it’s harder to revive it than one that has a steady history of solid sales. Don’t worry; if your book is in the cellar, you can market it. Just understand that a book with steady sales is going to be easier to revive/boost up than one camping in the telephone number ranks.
This works for two reasons:
- Amazon’s algorithms treat consistent sales as organic buying activity, and will start to recommend your book if it sees evidence of this (provided you’ve targeted your book toward the right buyers).
- You’ll maximize the impact of those sales on your sales rank, generating visibility for you on Amazon’s genre and sub-genre bestseller charts.
It’s also important to know that Amazon counts each borrow via Kindle Unlimited (or Prime Reading) equal to a sale in the sales ranking. That’s why Kindle Unlimited books tend to dominate the Top 100 in most sub-genres. Ranking, however, doesn’t necessarily mean profit – people can borrow a book and never read it (thus not generating any sweet, sweet KU reads). Don’t fall into a trap of chasing rank and sacrificing significant profits.
Sales volume and velocity are self-explanatory: sell more books in a short period of time and you’ll rank higher. What’s a little counter-intuitive, however, is that Amazon’s algorithms also reward consistency, rather than massive spikes. As such, you want to spread out your marketing efforts to mimic the right-hand curve:
To be absolutely clear, sales will fluctuate, and you’re not screwed if one day your sales dip slightly; the underlying principle is what’s important. You want your sales to trend upward over the course of your promotions to maximize visibility and the tail. This way to ensure this happens is simply to spread out your traffic over multiple days, instead of doing a “one shot” blast that results in a massive spike and then…nothing.
You can massively amplify the effect of those sales when you spread them out over multiple days, rather than firing all of your promo sites, newsletter, social media and blogging efforts on a single day.
It’s important to note that spikes aren’t “penalized” – selling a lot of books in a day is never a bad thing. You’re not hurting your book’s long-term chances or anything like that. A spike without any follow-up, however, will drop right back into the sales cellar. Thus, the core reason we avoid spikes and focus on consistent sales is because the same amount of sales spread over a 4 – 10 days works better, both in the short term and for long term post-promo sales. I prefer 5 – 7 day promos and launches, for the record; that’s enough time to show the algos consistency and establish a solid sales history, but short enough to generate the velocity/condensed sales mass required to really move the needle. It’s also doable even if you have a modest (or practically no) existing platform.
Longer launches and promo pushes totaling 10 days or even more aren’t uncommon, but usually they have patchy areas where your sales drop precipitously, thus killing your momentum.
Curious about the math behind this? Let’s dive into that and explore why we should strive so hard for a consistent upward trend.
Why Consistency Matters: How to Maximize Your Sales Rank (The Math)
This is the math that explains why you need to spread your sales over the course of a promo. You can skip the math, but working it out will give you a deeper understanding of what’s actually going on behind the curtain.
First, two points of clarification:
- Sales rank (e.g. the # you see on every book page – Amazon Best Sellers Rank: #67,339 Paid in Kindle Store) is purely driven by sales + KU borrows.
- Popular list rank (now known as “Featured”; thanks to PhoenixS for explaining this to me) is a mixture of sales + KU borrows and 1/10 of any free downloads over a rolling 30-day period.
Higher ranks grant you better placement on Amazon’s popularity & bestselling charts.
Here’s a free Excel spreadsheet that auto-calculates the math outlined below. It also has a chart which details how many sales are required to hit a specific rank (big thanks to PhoenixS for letting me use her rank data).
- For simplicity’s sake, let’s say it requires 40 “rank points” to rank at 5,000 in the Kindle Store.
- 1 sale = 1 rank point, 1 borrow = 1 rank point.
- Today’s rank point score = 1/2 of yesterday’s score + today’s sales/borrows.
- Let’s say, given our budget, available promo and mailing list size, we can reliably generate 80 total sales @ $0.99 over a five day promo window, after which the book will return to full price. How should we spread these out to hit the top 5000 and maximize our visibility/ROI (return on investment)?
OPTION A (start big, taper down): At the end of our promo, we have our lowest score, lowest rank, and we didn’t hit our goal of hitting the top 5000.
OPTION B (start small, scale up): At the end of our promo, we exit at peak rank (beating our goal), breaking the top 5,000. We come back to full price at maximum visibility, and with our strongest rank point history, thus enhancing our chances of getting “sticky” at a higher rank even as sales naturally decline to an equilibrium point.
More in-depth math analysis can be found here.
Remember, we’re not spending more to deploy OPTION B, nor are we generating more sales: we’re merely scheduling things differently. With just a little planning, we massively increased the effectiveness of our promotional efforts. This is, of course, imperfect in practice; you can’t know exactly how many sales you’ll get from a certain traffic source beforehand. Using historical estimates, however, we can schedule our efforts so that they gradually increase, with the heaviest promo push toward the end.
The general principle is simple: backload your biggest promos toward the end of a promo, if possible, and create a consistently increasing sales curve.
Why does sales rank even matter? Because readers browse the sub-genre Top 100 lists—and hitting your sub-genre’s top 20 is a great source of free traffic. Structuring your promotional schedule correctly can be the difference between receiving that organic boost—and a nice sales tail that lasts weeks after your promo is over—and plummeting back into Amazon’s cellar.
Now that we’ve covered the first two key principles (volume & velocity and consistency) to leveraging Amazon’s algos, it’s time to address the third principle: the sample of people purchasing your book.
The Data Set: All Traffic is Not Equally Beneficial
We’ve talked about how to rank higher on Amazon’s sales charts.
We’ve even talked about the Holy Grail: a book selling on its own for weeks or months at a time, otherwise known as stickiness. Indeed, Amazon’s recommendation engine can push far more books than you can ever hope to—and all for the low, low price of $0. Alas, however, stickiness is a fickle beast.
But what if we could increase our chances of getting sticky?
We can. To be clear, the algos are a tempermental, unpredictable beast under even optimal circumstances. But the more we can tilt the odds in our favor, even if it’s slightly, the better chance we have at winning this book publishing game.
And all that it requires is an understanding of how to train Amazon’s data set.
As we’ve previously discussed, Amazon’s site is essentially one big machine learning organism. You can consider it a baby AI: it’s constantly recording your actions and trying to predict what you’ll buy next—before you even know. It does this via advanced statistical analysis of massive amounts of data. As it combs through all this data, Amazon’s baby AI searches for patterns amidst the hundreds of millions of customers in the database, like a pig searching for truffles. Based on your browsing and buying history, it will then recommend things that other folks with similar data profiles purchased.
All this sounds well and good and has already been covered: Amazon works hard to recommend us shit we like to buy. So what?
By feeding Amazon the right data, you can get them to recommend your book. But not just any readers. Exactly the right readers—whether you write political thrillers or culinary cookbooks. Result? When your book is shown to the right readers, it produces a higher conversion rate, better reviews, and more sales. Best of all? Amazon’s algorithms interpret this behavior as “customers enjoy this book” —so then the AI returns to its little data storage vault and recommends your book even more. This reinforcing cycle has massive upside, given Amazon’s hundreds of millions of customers. Imagine them hand-picking customers who love urban fantasy or bad boy romance out of this database, then aggressively marketing to those people—e.g. your core fanbase—automatically and totally for free.
What does look like in practice?
In short: if you have an urban fantasy book, and you feed Amazon some voracious urban fantasy readers—who have purchased dozens of UF books on their Amazon account—Amazon’s recommendation engine searches through the customer base to find other people who fit this “voracious urban fantasy reader” profile. (Hat tip goes to Chris Fox for putting these pieces together in Six Figure Author.)
The rest is basically as easy as counting your money. Because when voracious urban fantasy readers are recommended new, cool urban fantasy books, what tends to happen? They buy. And boom: Amazon is now selling books to hardcore fans that you never could have reached on your own.
Let’s dive further into the specifics of how to find the right initial readers to properly train Amazon’s data monster.
How to Train Amazon’s Data Monster
If you’re worried that getting Amazon recommending books to the right readers is going to be difficult, worry not.
You train Amazon’s algorithms simply by selling books to your core readers. This means that if you have a 1st person female main character urban fantasy mystery, finding 150 people who have bought lots of similar books will be more beneficial than 1000 more general book buyers. Because those 150 sales will teach Amazon exactly who your target market is.
This is when the magic starts: Amazon sends out highly targeted recommendations to voracious urban fantasy readers. Your book sells well, they recommend it more, it sells, and so on, as previously outlined.
But what happens when you get a random or more general sample of 1000 people to buy?
Well, this broader collection of buyers will also trigger recommendation emails. After all, your book is selling; and sales volume is a big factor in Amazon’s algos. Let’s say I like thrillers and purchased a UF book once – randomly. I might get an email from Amazon because I fit their vague, uncertain profile of people who might like your new UF book. Unfortunately, I don’t buy, because I don’t really like books featuring wizards. The result of these broader recommendations is catastrophic and also self-reinforcing: your conversion rate plummets due to your book not being relevant to a ton of the people it’s been recommended to, review scores drop, customer interest metrics are low (clicks on the book in emails or in-store merchandising placements/bounce rate on the page). All of which leads Amazon’s friendly neighborhood AI to an obvious conclusion after it sorts through the aftermath.
Buyers just aren’t interested in your book. And boom: Amazon stops recommending it. Your book disappears into the Kindle cellar.
The main idea here is that you confuse Amazon’s algorithms when you get a ton of sales that have no clear data pattern behind them. And when Amazon doesn’t know who to recommend the book to, its automated marketing efforts will be substandard.
All this means in practical terms is simple: prioritize highly targeted sources of traffic that hit your key audience over broader ones (even if they result in slightly more sales).
The real reason you want to target your buyers actually has nothing to do with fancy technology and everything to do with commonsense. If you advertise your thriller novel to people who like thrillers, they’re more likely to buy. This means your marketing costs are going to be cheaper and more effective.
On the other hand, putting your thriller in front of contemporary romance readers is an uphill battle. You’ll need to spend more to achieve less.
Thus, when you’re considering your traffic sources, remember:
- PPC is the gold standard for laser-targeting. Facebook, BookBub, Amazon AMS and other venues allow you to narrow your target to specific sub-genres and authors.
- Your fans are the best way to train Amazon’s data set, provided you don’t genre hop. If you do genre hop, then you need separate lists (e.g. one for your thrillers and another for your wizard books) so that you don’t muck up the data set.
- Highly-targeted, genre-specific promo sites “train” Amazon’s recommendation engine better than more generalized ones. Unfortunately, highly-targeted promo sites are rare; thus, you often want to save such sites until your book has a more established sales history and Amazon already knows who to recommend it to. Otherwise, you can confuse the baby AI out of the gate, and it can have a lot of trouble recovering later on.
Okay, but how sensitive are Amazon’s data systems? This is a question I’ve gotten a few times, so let’s clear up some misconceptions before we continue.
Important Note: Mucking with the Data
Some people are so concerned with misleading Amazon’s AI that they don’t want to send any off-target traffic to their book page, lest the algos get permanently confused.
While Amazon’s AI isn’t a genius (yet), it also isn’t that easily duped. A few sales from your friends/family or a well-meaning associate won’t scuttle your chances of getting some recommendation love. In fact, even a bunch of untargeted sales won’t skew the data set, provided a large chunk of the sales is coming from the heart of your target audience.
How much? I’d aim for 75%+ targeted sales as a rough rule of thumb. But that leaves substantial room for people outside your core target audience to pick up and enjoy your book without the algos getting confused.
So you can stop worrying about leading Amazon’s AI down the wrong path when Grandma Beatrice buys your latest opus. Your book’s data set will be just fine.
How Do I Know If My Data Sample is Off?
If you think you might have shotgunned your book out to everyone with a pulse (instead of your target audience), how can you tell if you’ve confused Amazon’s algorithms? To be clear, there’s no surefire way of determining this, outside of going under the hood of Amazon’s algorithms. Since the chances of them allowing us to do that are less than zero, however, we have to rely on a simple rule of thumb.
We look at our book’s also boughts.
Basically, your book’s also boughts should match your book’s genre. If you have a bunch of cookbooks in your thriller’s also boughts, you’ve confused Amazon’s targeting. This can happen when you use more general promo sites, or if you market your book to a more general audience without targeting any genre-specific readers. It can also temporarily shift after, say, a BookBub Featured Deal, since your book will often be “linked” to the other books in that day’s newsletter since readers will pick up multiple titles in different genres at once. The shift in also boughts following a BookBub tends to sort itself out after a month or two; in any event, BookBub Featured Deals are the best promotional tool in the business, so it’s worth enduring any temporary and minor confusion of the Amazon algos (should it occur).
This is what it looks like when your targeting is correct and you’ve trained the data monster well (this is for a dystopian/post-apocalyptic book):
And this is what it looks when the targeting is a bit off for an urban fantasy book (though not completely in this case):
Most of the also boughts are also urban fantasy books, but you can see at #4 we have a self-help title. This occasionally happens when you run a lot of more general promo sites (which I have for this book). In this case, I wouldn’t be worried about it; the other books are supernatural/urban fantasy related, which suggests Amazon has a pretty good idea of who to recommend by book to.
Note that you’re also more likely to see a bit more disparate also boughts for backlist titles (this book is over two years old at the time of this screenshot). This is because Amazon refreshes the also boughts for each title. If your book isn’t moving a ton of copies, that can mean that some random sales and downloads during that timeframe can slightly skew the sample. This isn’t cause for concern when it comes to backlist; it happens naturally, and your older, unadvertised books typically aren’t getting a whole lot of organic recommendations from Amazon, anyway.
And what about fixing the targeting if it’s gone awry? Simple: run some targeted promotions to the page. This could be your newsletter (e.g. run a deal and let them know about it) or using pay-per-click ads targeted to your specific audience (say, urban fantasy readers in this case). Or you could select some highly targeted, genre-specific newsletters that serve only specific sub-genres (e.g. a sci-fi/fantasy focused newsletter with an urban fantasy or supernatural suspense sub-genre segment). It can take a dose of a few hundred sales to recalibrate Amazon’s algorithms to get your book recommended to the right people again. So it’s generally better to get this right from the beginning.
Not a Silver Bullet
All this talk about algorithms may be conjuring up visions in your mind’s eye of riding Amazon’s recommendations straight to El Dorado on a cyborg thoroughbred fueled by NOS.
Sadly, I’m here to temper expectations a bit. Yes, leveraging the algorithms to your advantage is critical to maxing out your marketing efforts. But the gains are mostly modest. Remember, this bookselling game is one of compounding and brick-by-brick increases. This is merely a piece of your marketing toolkit. Effective? Yes.
But still just a piece. And relying on the algorithms to lead you to the promised land of sales not recommended.
It’s much harder to get sticky in 2019 than it was even two or three years ago. As mentioned earlier, Amazon’s algorithms like new content (primarily because people like new content—again, remember that their core focus is customer satisfaction). As such, their algorithms reward new books with visibility boosts. Colloquially, these are referred to as the 30/60/90 day cliffs. Much of your book’s new release mojo wears off after 30 days (when it’s no longer eligible for the Hot New Release lists, among other on-site placements). This dips a bit further after 60 days, before the rest vanishes after 90 days.
Practically speaking, this makes it harder for an even four or five month old book to keep riding high in the charts. Amazon’s algos just don’t shower these “older” books with the same love as new titles.
Indeed, most promo runs see books rocket up the charts, only to crash back into the ranking cellar like they’ve got lead strapped to their ankles, even when you design them with Amazon’s algos in mind. Thus, if you’re building your marketing strategy around perpetual organic sales, that has unfortunately become much less viable. I have easily dropped a healthy five-figures in ill-advised marketing spend in pursuit of this white whale. Those marketing dollars were used for one-off promos instead of building an author platform.
This comes back to strategy. Promotion is a critical part of your marketing mix. But you must consider how every promotion feeds into your overall strategy and brings you closer to your core objective.
I searched for a silver bullet for years. I found a couple bronze ones, but they never had the firepower I truly sought. That’s because, generally speaking, the key ingredient to igniting the algos is a rabid group of fans (e.g. your newsletter and other loyal readers). No ad spend or marketing blitz can compete with the power of word of mouth and a few hundred (or thousand) people who will buy your book during launch week.
While it does you no harm to optimize your marketing efforts for Amazon’s algorithms, and I certainly recommend you do so, this is not a replacement for the difficult business of finding your fans. Your core marketing focus should always be building your own fan base. This is a pain in the ass, because it demands years of patience. The alternative—an instant vault up the charts, and subsequent full-time authordom—is alluring. But chasing rank or burning massive piles of cash as a loss-leader to prod the algos is usually a fool’s errand. With 3,000+ books being released every day, organic sales can (and do) disappear overnight. Expect to do the heavy fanbase lifting, and consider any selling Amazon (or another retailer does on your behalf) as an amplifier to your own efforts.
I’m Wide—What About Other Retailers?
We touched on this in the beginning: Amazon has essentially staked their business on developing automated merchandising technology that recommends readers exactly the kind of books they like. This has been wildly successful for them by any measure. Other retailers, however, work on a much more old-school model: retailer-curated merchandising.
If you enter a brick and mortar bookstore like Barnes & Noble, you’ll find books on the front table, on end cap displays (the little racks at the start/end of aisles), and in other prominent locations. A handful of these will be top-selling books, but the majority of them will be either hand-picked by the staff or, more likely, paid to be placed there by the publisher. Obviously the front table is valuable real estate—every customer who enters the store must go pass by it. As such, publishers pay extremely good money for their hottest books to be stocked in these prominent locations.
Similarly, certain merchandising placements on retailers (even Amazon) are purchased by large publishers.
However, there is hope for smaller indies. Other retailers have teams that curate hand-picked selections featured prominently on their site: things like First in Free, exclusives to that retailer (for example, making a pre-order exclusive to that retailer), and other manually organized offerings. If you have a rep at one of the other retailers, you can petition them directly to be included in these merchandising opportunities. But the rep-less among us (like myself) aren’t out in the cold. If you submit via Draft2Digital, you can email D2D’s team and ask them if there are any merchandising opportunities available. They can sometimes get you included for these various promotions.
And if you’re direct to these sites, you can just cold email the support staff. Yes, they might say no. And yes, I understand as an introvert myself, that the thought of doing that might be akin to swallowing a handful of pine cones. But make them an appealing offer, and you can get an actual human to take notice. It only takes a minute or two, and there’s no downside (or cost) to you. And the rewards can be substantial, keeping you selling for weeks or months after the promotions are over.
Application: Promo Stacking
We’ve talked about sales rank and training Amazon’s data monster. But, if you’ve been following along, you probably have a question: how do I get enough sales to rank high and trigger Amazon’s automatic recommendations? If you’ve looked at the Top 20 or Top 50 of most sub-genres, you need at least 50 – 100 sales in a single day to crack that visibility threshold. That can be a tough mountain to climb, even for a mid-list author.
Subsequently, one of the most powerful tools in your marketing arsenal is promo stacking.
This term originally applied to using multiple promo sites to create a large boost in sales. Here, however, I’ll use it to describe combining many separate traffic sources—whether they be promo sites or PPC ads—to push your book higher up the charts. This “stack” is spread over multiple days. Normally, 1 + 1 = 2; but with a promo stack, 1 + 1 = 11.
Or it can, with a little luck and some planning.
We know from our discussion of the algorithms that they reward sales volume and velocity. They also reward consistency. But there’s a balance that must be struck between these factors, for a series of sales that looks like this:
8 – 7 – 3 – 10 – 5 – 4 – 3 – 8 – 5 – 2 – 9
will be less effective than:
15 – 16 – 19 – 22
The first scenario might be relatively consistent, but won’t get you near any charts if your sub-genre is competitive. And it probably won’t trip Amazon’s automated recommendations, either.
To maximize your chances of getting the algorithms behind your book, you must critical mass with sales. No one knows what the exact tipping point is for Amazon to send out merchandising emails and start pushing your book to buyers. But we do know they reward books that are already selling—and their recommendation mechanisms favor books that are selling big.
Despite its intricate nature, Amazon’s AI is still a baby mammoth that, while sophisticated, responds best to brute force. It likes sales. And it likes a lot of them in a short period of time (3 – 10 days; again, as mentioned before, I prefer 5 – 7 days for launches and promos).
But getting a lot of sales in a short period is difficult for authors. Hence the promo stack:
- Day 1: promo site 1 + promo site 2 + promo site 3 + promo site 4
- Day 2: first part of personal newsletter + $10 PPC campaign
- Day 3: second, larger part of personal newsletter + $20 PPC campaign
- Day 4: $40 PPC campaign + five additional newsletter sites
- Day 5: $80 PPC campaign + six additional newsletter sites + last chance email to your newsletter.
This is not a prescribed order—your traffic sources and stack will differ. It is merely an illustration of how you can combine and spread out multiple traffic sources to harness the power of Amazon’s algorithms. Whereas each individual source by itself might have only generated ten or fifteen sales, in tandem, they form a powerful push.
Of course, like fishing for bass, there’s no guarantee the AI will bite. But the more enticing and tasty the bait, the greater your chances.
And sales are the tastiest bait of all.
Done correctly, each element of the stack adds up to more than the sum of its parts due to tripping Amazon’s charts + automated recommendations. So you might get 50 sales from the 4 promos on Day 1 – but this turns into 65, because you hit the Top 20 of your genre and got additional traffic for free.
Super basic, but super effective.
We’ll talk about all the promotional sources you can use to drive traffic to your books in Part 5: The Ultimate Guide to Traffic Sources. This is actually the easiest part of marketing, so if you’re struggling with finding places to advertise or locate readers, that will provide you with plenty of methods of attack.
- Amazon’s algorithms are the most powerful bookselling force on the planet.
- Amazon, unlike other retailers, primarily uses automated merchandising based on customer data.
- These automated recommendations appear all over the site in places like the also-boughts, bestseller lists, and pop lists, as well as in emails.
- Understanding how the algorithms fuel these recommendations can massively amplify the rest of your marketing efforts.
- Despite the algos’ power, they aren’t a magic bullet. Building a career is primarily about building your reader base brick-by-brick.
- The algos are also unpredictable, even if you understand the principles and follow them perfectly.
- The recommendations focus on three primary factors: sales volume & velocity (most important), sales consistency, and a tight sample of people purchasing your book
- They also look at newness and the book’s sales history.
- When scheduling your promos, you want to hit critical sales volume mass
- Always backload your strongest traffic sources to end promos strong and set your marketing efforts up to generate a gradually increasing sales curve
- Target people who read your genre/sub-genre to get Amazon to promote your book to readers who match a similar profile in its customer base
- You can recalibrate Amazon’s targeting if you accidentally corrupt the data sample by getting a few hundred sales from readers in your sub-genre.
- Having a day where your numbers don’t move up isn’t a death knell when you’re trying to get sticky. General upward trend is the idea, not perfection.
- You can promo stack your various traffic sources (newsletter, social media, promo sites, PPC, et al) together to produce the necessary sales volume and consistency to trigger the algorithms. By combining your traffic in this way, 1 + 1 = 2; instead, 1 + 1 = 11.
- Design a promo stack that properly leverages the organic visibility from Amazon’s charts + recommendation engine.
- Determine how many sales it will take to break into the Top 20 & Top 50 of your sub-genre. Use the rank chart in the Excel spreadsheet below as a rough guide; here are Amazon’s Kindle Bestseller Lists (nicholaserik.com/top100).
- Based on this, pick a rank target for either a new release book or a backlist book that hasn’t gotten any promo love. E.g. if book #20 in paranormal vampire romance is #550, your target rank is 550 – and your rank point target is 300 sales + borrows in a day.
- Schedule your three traffic sources over 3 – 10 days, stacking them in a way that will create a gradually increasing sales curve and allow you to hit your target rank. You can download the free rank calculator Excel sheet (nicholaserik.com/excel) to automatically do the math. If you don’t have any traffic sources to use, read Part 5 first, then return to this step.
- BONUS: Book the promo stack and record the results.