A real horror story has been slowly building for the last year or so and I’m getting a lot of emails on the topic so it’s time to deal with this head-on: what the hell is going on with Also Boughts?
For those unaware, the strip of books right which are usually placed underneath your product description on Amazon, headlined with “Customers who bought this item also bought” are popularly known as “Also Boughts” and have become the subject of much attention lately, as our knowledge of their importance grows in tandem with Amazon’s seeming desire to muck about with them.
First their importance: if you have read Amazon Decoded you will already know just how critical Also Boughts are and can skip ahead. For the rest of you, Amazon’s system is always trying to determine what kind of products you personally are most likely to purchase, so that it can display more of those to you. One thing it looks at very closely is the connection between products. People who buy printers tend to buy ink, for example, and recommending a printer-buyer some ink to purchase will elicit a lot of clicks.
But it’s not just obvious pairings like leathers and feathers, Amazon’s system is constantly analyzing what everyone purchases to predict what everyone else will purchase next. Each book (that has been purchased by a certain number of people) will have a number of connections in its system, which can be visualized by a tool like Yasiv.com — which isn’t perfectly accurate, but does a good job of showing how certain groups of products tend to be purchased together, which in turn gives us a good idea of what books Amazon will recommend a customer after they buy any given title.
I’ve been stressing for the last year how Also Boughts are central to the entire recommendation engine. I wrote a post called Please Don’t Buy My Book in 2017 which explained how critical it is to protect your Also Boughts in those first days of a new release — so that Amazon’s system doesn’t get an inaccurate idea of what kind of person might like it, just because some (well-meaning) family and friends bought it out of support.
While the Also Boughts on your page are an indication of the health of those connections in the Amazon system, what really matters are the connections to your book, not the ones springing from your book. In other words, the Also Boughts pointing at you are what really matter, and I teased that out in a follow-up post called Who’s Pointing At You?
(Note: this is a good time to point out that if you want to explore the archive of like 400+ posts here it’s much easier to do that now. Search works great, finally. Everything is also now tagged by category, and you can just click Amazon for example and get all my posts on that topic alone.)
After getting distracted for a couple of weeks by a pair of clickfarmers hitting #1 on the Kindle Store — remember that? — I wrote the final part of this Also Bought breakdown called When Reader Targeting Goes Wrong.
All that was just the beginning though, and I dug into the topic in a more holistic way in Strangers to Superfans — which is basically an entire soup-to-nuts system for targeting the right readers and turning them into superfans who do the selling for you.
OK, that’s a lot of stuff about Also Boughts.
In all those words, though, I appear to have miscommunicated something rather key, which is contributing to a bit of fretting about Also Boughts, and it’s important to clarify a few things.
Let’s deal with the panic first. To anyone who hasn’t noticed, Amazon has been playing around with various iterations of product pages for over a year now. In fact, it’s always toying with the design of these, it’s just that more people are noticing now as Amazon cycles through various iterations a little faster.
In many of those iterations, which it split-tests regularly so that different segments of its customer base will see different things, Also Boughts have either moved to a position of far less prominence near the end of the page, just above reviews, or have vanished altogether. To rub further salt in the wound, in most iterations the strip of Also Boughts is replaced with (yet another) set of pay-to-play Amazon Ads.
The latter strand of split tests is engendering mild terror right now but I think I can set your mind at ease, at least partly. While such a change would be unwelcome — if adopted, more on that in a moment — it’s not the Also Bought Apocalypse it is being painted as. For these reasons:
- The set of Also Boughts on your page aren’t what is inherently valuable here. And it’s not even those pointing at your book. It’s what they represent.
- The connections between books are what is truly valuable, and what Amazon uses to decide what to recommend next. The Also Boughts themselves are just signposts.
- If a signpost disappears, it might be harder to find the town you are looking for, but the town itself doesn’t disappear.
- Much in the same way, it’s not Amazon’s system that will get confused about a book’s connections in the recommendation engine, it’s us.
In other words, if you write grimdark Epic Fantasy and share an audience with David Dalglish and Brent Weeks and have assiduously developed your readership and been disciplined with your targeting and Amazon’s system recognizes that fans of Dalglish/Weeks also love your books, and then recommends your books regularly to those readers, and this is a real driver of sale for you… don’t panic. Those recommendations won’t change in any way even if Amazon takes Also Boughts to the woodshed and we can’t see them anymore. The connections won’t disappear, just the signposts.
Yes, it will still suck. Looking at our Also Boughts is a quick and easy way of telling us if something has gone wrong with our reader targeting leading Amazon’s system to have a muddle idea of what kind of book it’s dealing with, a clear sign that we need to point our ads in a different direction, or target a whole new group of readers. Maybe tools like Yasiv.com will still work, but we really can’t say at this point — that all depends on what information will still feed out to the API.
Also Boughts are handy in other ways too — for example, they are great for harvesting targets for ads or identifying authors you might wish to cross-promote with, because they are often a classic indication of who you share an audience with (in sometimes surprising ways, as writers tend to think in terms of who they write like — not always the same thing).
Some writers can be ambivalent about Also Boughts because they often feature “competing” titles and can tempt readers into clicking away. But remember that even if Also Boughts disappeared tomorrow, and even if they weren’t replaced with yet more ads, there would still be 100+ other books advertised on our Amazon pages (I know, I counted them). And the key thing about Also Boughts is that they cut both ways — your book is appearing on just as many other pages, drawing readers towards you.
Readers seem to love them too, as it gives them ideas of other books to purchase. Which makes sense. If they are attracted to Hunger Games the overwhelming likelihood is they will be into Divergent also.
And it may be readers, ultimately, that save Also Boughts.
There seems to be a gloomy tendency to assume that we’re always on the worst timeline, that the least advantageous iteration of these split tests is what will indubitably come to pass. But I know from studying the algorithms pretty closely for five or six years that history doesn’t really support that conclusion.
Even the Great Algo Wipeout of 2012 — which nixed freebies and put an end to the first crazy goldrush of KDP Select — wasn’t the very worst iteration that Amazon was testing at the time. In simple terms, Amazon twiddles with knobs and pulls levers and strives for a kind of reader-positive balance (but one that still makes them a pile of money, of course).
People seem to be assuming that a strip of Amazon Ads instead of Also Boughts makes Amazon more money. I wouldn’t be so sure about that.
Remember that conversion rates on Amazon Ads are abysmal — if we get 0.1% CTR on an Amazon Ad we’re onto a winner. By contrast, to show how broken Amazon Ads is at a fundamental, structural level, a CTR of 1% on Facebook or BookBub might be considered a failure.
On top of that, Amazon has a history of foregoing the short-term cash reward to build long-term user trust in its recommendations. That’s central to the core promise that Amazon makes its customers: that it will always seek to show you the products you are most interested in.
Yes, you could argue the entire existence of Amazon Ads runs counter to this promise, and I’d agree with you — particularly in its current iteration which just encourages bad advertisers and deep pockets to keep spamming the same ineffective ads with crappy CTRs, slowly eroding whatever user trust remains in those ads.
Amazon Ads have been getting their day in the sun, and we’ve seen them shoehorned uncomfortably into all sorts of places — a very un-Amazon approach, trying to trick customers into clicking on them instead of doing the harder, long-term work of making the ads more relevant (by encouraging better advertiser practices, engineered by making relevancy/CTR a part of the auction, but I digress).
But it hasn’t all been one way-traffic. One unpopular change/test has already been rolled back, and Amazon Ads no longer appear at the top of search queries. Presumably readers voted with their wallets, and Amazon decided that maintaining the prominence of organic search results was better.
Amazon may well decide the same with Also Boughts, but even if it doesn’t, just remember that it’s only removing the signposts. The connections which still matter will be unaffected.
We’ll have a harder job figuring out what those connections are, and whether they are healthy, but that just means we’ll have to double down on making sure our reader targeting is on point, that are books are being pushed to the right eyeballs, that our packaging is eliciting excitement from our core audience, that our books are jam-packed with all the tropes they devour, and that we turn those buyers into superfans by engaging with them regularly.