If you’ve ever done any sort of marketing, chances are you’re familiar with the concept of affiliate programs, but for everyone else, they may be unfamiliar. Simply put, affiliate programs are a way of making money by referring your own site’s readers to other websites that sell products. Those companies pay you a small percentage of the purchase price of products purchased in exchange for that referral.
As an author, you can use affiliate programs to increase your share of the final purchase price of your books, at least for people who follow the links to those stores from your website. Additionally, you can use affiliate programs to help you determine the effectiveness of your ad campaigns. More on this in a bit.
Analytics is a related concept—so much so that you can’t usefully talk about one without talking about the other. Analytics refers to looking for patterns in large amounts of data. While wearing our marketing hats, we use analytics tools to learn more about the people who visit our websites and, more importantly, to learn more about how they interact with our websites. For example, with analytics tools, you can learn:
and so on. This, in turn, can provide valuable insight into how effective your website is at selling products and into what you can do to make it more effective.
At the core of analytics is the concept of tracking—using specially formatted URLs, referrer information, and other tools to figure out where a user came from and where a user is going. Many of the affiliate programs provide basic analytics support, and you should take advantage of that support to the maximum extent possible, so that you can find out how effective your advertising is. For example, if you’re doing a big advertising campaign, you could add the name of the ad campaign into your ads’ URLs. Later, when you’re looking at your sales, you can see how many people bought your book as a result of following links from those ads versus following normal links from your website.
This article—the fourth in a series articles on self-publishing—tells my experiences with affiliate programs from several popular U.S. book chains. (Stay tuned for future articles about print publishing and advertising. And if you missed the first parts, be sure to read Part I: Writing Over the Long Haul—What Went Right And What Went Wrong.)
After I released my electronic editions, my last task (other than shepherding the dead tree editions through that process) was signing up for affiliate programs so that buyers who followed links from my website (and/or Facebook posts) would net me a small additional bit of income. This sounds straightforward, but things weren’t always perfect.
One thing that was problematic for every affiliate program (and thus worthy of mention here) is that none of their link generators produce valid HTML. If you’re familiar with HTML, you know about HTML entities—things like — for an em dash, for a non-breaking space, and so on. Because the ampersand character has special meaning in HTML, whenever you want to show an actual ampersand on the screen, you must instead write &. They didn’t. (show details)
The behavior of an ampersand outside the context of one of those named HTML entities is undefined. As a result, if you fail to represent your ampersands with &, you’ll get quirky behavior that varies from browser to browser. Worse, the behavior will depend on what text comes after the ampersand. (This is literally one of the first mistakes that most people learn not not to make—the hard way—when they begin writing HTML by hand.)
What many people don’t realize is that those rules about ampersands apply everywhere, including inside HTML attributes (for example, in URLs that you include as part of a link or image tag). So if you create a link that looks like this:
<a href="http://example.com/foo?tag1=value1&tag2=value2">
then that’s not valid HTML. Why is it invalid? Because the browser reads that and tries to figure out what the HTML entity &tag means, or possibly what the HTML entity &tag2 means, and then handles that failure in a browser-dependent way. Usually, when a browser reaches the first character that’s illegal in an entity name (a semicolon for valid HTML, or the equals sign in this case), it stops, gives up, and assumes that the ampersand should have really been &. But if what comes after the ampersand happens to be a valid entity, most browsers assume that you merely forgot to include the semicolon, and replaces the entity with the appropriate character anyway. So a particularly troublesome example might be:
<a href="http://example.com/foo?tag1=value1&num=3">
because a browser could reasonably interpret that URL as http://example.com/foo?tag1=value#=3, which is a very different URL than the one you intended, and would be interpreted completely differently, both by the browser and by the web server.
Unfortunately, a lot of sloppy web developers ignore HTML’s entity escaping rules, and assume that if a URL works today, the problem doesn’t really exist. The problem with that reasoning is that in the 1995 HTML 3.0 spec, by my count, there were only 96 named entities. In the current spec, there are well over 1,400. So not only can the standards body add new HTML entities at any time, they do so with alarming regularity. Thus, it simply isn’t sufficient to look at the current list of named entities and avoid using them in URLs.
The only correct way to handle ampersands is to properly entity-encode them in the URL string. For example, the first URL above should be written as:
<a href="http://example.com/foo?tag1=value1&tag2=value2">
These big web-based companies really ought to know better. It’s even more troubling when I report the problem, and they say that my validator is being too strict. Somehow, it gives me an almost irresistible urge to recommend that the W3C and/or WHATWG deliberately create HTML entities for all of the commonly used tracking URL query parameters, just so that all those links would break horribly en masse. But maybe that’s just me.
Fortunately, it is easy to fix the URLs—just change all the ampersands to & before pasting them into your web page—but it does get tedious rather quickly.
Now, on to the per-company details.
Compared with Apple, Kobo, and Barnes & Noble’s affiliate programs, the Amazon Associates website is a bit of a headache to use, because Amazon requires each link to contain a link ID that is specific to that link. As a result, you have to actually use the link generation tools on Amazon’s website to construct the links, rather than simply appending a consistent, fixed string onto the URLs of the products you’re linking to.
For most people, that limitation probably doesn’t matter, but for folks who like to be able to pre-create URLs based on ISBNs so that their website will be ready to go live at midnight on a release date, that means some very late-night website editing for no apparent reason. (Of course, even if they changed that, it would still not be possible to link to Kindle eBooks prior to publication, for reasons previously mentioned in part II.)
Additionally, I had some pretty significant problems joining the program—multiple rejections with baffling reasons, many of which appeared to stem from them not knowing about CloudFlare’s automatic spam captcha system. As a last-ditch effort to get back on track, I wrote an email to their CEO, and their upper management stepped in and fixed the situation reasonably quickly.
Amazon has assured me that they’re going to carefully study what I experienced to ensure that it doesn’t happen to anyone else. I’m cautiously optimistic, but if you use CloudFlare, it might not be a bad idea to turn it off until they accept your application (assuming that doing so doesn’t make your site unbearably slow), just in case. At a minimum, disable Browser Integrity Check and crank down the Security Level setting to “Essentially off”.
Teething pains aside, though, on the whole, the Amazon system provides fairly good analytics capabilities, showing me the number of people who clicked through my links (by tracking ID, which you can use similar to a campaign ID), how many units were sold, how many free samples were distributed, earnings from the sales, etc.
The only feature I wish their site offered is a way to drill down by HTTP referrer and/or by arbitrary campaign ID, to find out precisely which of your website links was effective. You can get the same information by creating a large number of tracking IDs (e.g. one per website), but because you have to do that at Amazon’s website, and because you have to create all the links on a per-ID basis, it is a much bigger hassle to track sales precisely than it would be if you could just edit the campaign ID in the URL and paste it somewhere as you can with, for example, the iBooks Store.
Additionally, Amazon’s data visualization tools lack the polish of some of the other sites (no graphs). The data is there, of course; it is just a little harder to interpret it. (This is probably one of the disadvantages of building your own, home-grown affiliate program software instead of outsourcing it like everybody else did.) I hope that Amazon eventually adds similar data visualization tools, because it does make it a lot easier to understand what’s going on at a glance. In the meantime, it gets the job done, and is considerably better than many other affiliate programs, despite its shortcomings.
The Apple affiliate program “just worked” for the most part, though when I first joined earlier this year, their tools page was a mess, and the tools doubly so. Thankfully, they completely redesigned it in the past few weeks, and I’m happy to report that it actually looks like an Apple site now.
In particular, I really have to give Apple credit for their data visualization (though I suppose I really should be giving most of the credit to Performance Horizon Group, the third-party site that Apple uses). Apple’s affiliate site gives you an amazing view of your data that far exceeds what you get from any of the other affiliate programs. You can see graphs of each campaign to see how it did on any given day. You can also easily compare the top-performing campaigns at a glance, look at a given campaign’s click count or purchase count over time (graphically), look at what country the buyers, clickers, and downloaders came from, and so on.
The only real flaw I’ve found is that their link generator creates links that automatically open iTunes/iBooks and doesn’t provide a way to turn that feature off without manually editing the URL. Of course most of us manually edit links frequently whenever we change campaign IDs, but it would be nice if they offered more configurability for people who are new to the affiliate marketing world.
I’d be remiss if I didn’t make another mention of their store and affiliate program’s best feature. As previously mentioned in part II, Apple’s store provides you with direct links to hidden pages for each of your titles even before they start taking preorders. This feature makes it remarkably easy to create affiliate links for upcoming products before they appear on the store. No other affiliate program offered that capability, as best I could tell (though with the exception of Amazon, you could fake it for the other sites by constructing the deep-link URLs yourself and hoping for the best).
The Kobo affiliate program was relatively painless. They use a a third-party site, Rakuten LinkShare, that happens to be the same site that Barnes & Noble uses for their affiliate network. The biggest quirk with this network is that they won’t let you sign up for direct deposit until after they send you your first check. Très bizarre.
Additionally, it looks like there’s a lot less information at your disposal through LinkShare. As far as I can tell, there’s no way to drill down beyond the most basic numbers (number of units sold, revenue). And although they let you specify a campaign ID, I have not found a way to actually look at a breakdown of clicks or sales based on that campaign ID.
Basically, you should probably think of this as a way to earn a few extra cents per sale, but don’t expect to get any useful analytics data out of it. It just isn’t nearly as useful from a marketing perspective as Amazon’s site or Apple’s site.
The Barnes & Noble affiliate program tried hard to not answer my emails.
No, I’m not joking. The Barnes & Noble affiliate program became a private, invite-only program a couple of years ago. You’re supposed to join that same third-party affiliate program company that Kobo uses, and then send an email to an address at Barnes & Noble. I did, and nobody answered. A few days later, I sent an follow-up email. Nobody answered that one, either.
That’s when I decided to crank up the volume by sending email to their Nook Press help desk asking if they knew what was going on. Out of all the emails I’ve sent to the Nook Press help address, that’s the only one that they didn’t answer. (And it wasn’t an issue of spam filtering or anything; I received emails from them before and after that one.)
So then I sent an email to their general help email address—the one that normal folks use for customer service. Yup, you guessed it. They didn’t answer. I don’t think I even got an autoresponse from any of those emails.
About a month and six or seven emails later, I finally reached a human being who apologized, and was completely baffled. I think the only reason I was successful the last time is that I ignored their instructions completely, and instead, clicked the “Contact” link at the bottom of the affiliates.barnesandnoble.com website.
I hope now that they know about the problem, they’ll fix it. Either way, Amazon’s teething pains are in good company.
Once I finally got into the program, the tools I have at my disposal are the same tools that I have with Kobo. (Barnes & Noble also uses LinkShare.) I do like having the data from both vendors at the same website, but I wish the data were more thorough.
When it comes to affiliate programs, one company that is notable in its absence is Google. Their affiliate program ceased to exist a while back.
However, what Google does provide are general-purpose analytics tools that you can use to track how people use your website. You’ve probably heard about Google Analytics, and if you’re like me, you’ve probably blocked them at your firewall because websites randomly take a performance hit when Google’s servers aren’t fast enough. But what do these tools actually do for you as a website owner?
The short answer is, “it depends”. For a large e-commerce site, GA is great. It lets you do purchase tracking, discover which sites users came from when going to your site, discover information about your users, and so on. Of course, if you are willing to dig through your server logs, you already have that information, but GA presents it in a more easily digested format. GA also lets you discover which links people follow when leaving your site—something that your own servers can’t really tell you (unless you add some custom JavaScript).
For a small-scale site, GA is a train wreck. You see, there’s a category of abusive idiots called “referrer spammers” who go to your website and deliberately pretend that they are coming from their own website, so that their website will show up in your server logs and Google Analytics logs. These companies, in turn, sell services for search engine optimization and other things that might be of interest to site owners if it weren’t for the fact that people who advertise like that probably aren’t worth doing business with, but I digress.
The biggest problem is that Google doesn’t do anything about this. Many of these spammers don’t even visit your site; they just send bogus data directly to Google. Yet in spite of the fact that the requests don’t even contain your domain name, Google doesn’t bother to filter these clearly bogus hits out. And their filtering tools only work in certain parts of their site and not in others, so basically, their analytics tools lie to you frequently.
These sites actually send hundreds of bogus hits for each site to GA every week. So if your real traffic is in triple-digit territory or less, their spam may actually exceed the level of the real referrers that you’re trying to study. And they can make all your other data useless, because you can’t really say whether those eight MSIE users are real or referral spammers without digging in really deeply into the data.
To make a long story short, the number of people who abuse GA has made the data quality so low for small sites like mine that I would strongly recommend trying alternatives—ideally, server-side alternatives, where all the data collection is entirely under your control, and where you can filter it in whatever way you want. If and when I find one that I like, I’ll blog about it.
In conclusion, if you’re selling books, you should join every affiliate program you possibly can, and use affiliate links everywhere on your website and in any web-based marketing that you do. They’re basically free money that you can earn just by advertising your own books on your own website, and you’d be crazy not to take full advantage of them.
There is one caveat, though: Be sure to read the rules about when and where you are allowed to use those affiliate links. In particular, using Amazon affiliate links in email blasts is grounds for termination of your affiliate program membership. You do not want to go down that path.