Yesterday I published a new post on The Architecture of Information. It’s about the importance of maintaining structural consistency across a family of products.

Specifically, I wanted to put in writing an example I’ve used when talking about this stuff: the location of the “Sync to Furthest Page Read” button/link in the Kindle user interface. Bottom line: this is an important feature if you use Kindle in several devices concurrently. However, the way you access it depends on what platform you’re using. That can be frustrating. If you haven’t done so already, please read that post.

What I didn’t cover there is why this sort of inconsistency creeps into product families. (I’m focusing The Architecture of Information on examples, and not on speculating about how they’ve come about.) That said, writing yesterday’s post made me revisit my interview with Christian Crumlish. In that conversation, I asked Christian about the sources for this type of inconsistency, and even brought up the same Kindle example. This was part of his reply:

So, in theory, they’re all the same experience. And there should be someone saying, “Hey, we have a fundamental experience and you can express it differently, but we all agree it has to XYZ in common.” There are usually efforts to do that. And when I was doing the pattern library stuff, that was a version of that kind of thing.

Nowadays, design systems are a version of that kind of thing, but often they’re still about the interaction and not how it all fits together or how it works. But there are natural tensions. Teams are going to say, “Yeah, but that doesn’t work for my device,” or, “But I have reasons for this,” or “It’s always been this way on our sub platform. You bought us and now you’re trying to make us be part of you.”

It’s non-trivial — especially in a larger organization — to just, you know… Everything’s constantly shifting. It’s a system. You could gradually maybe bring it into harmony, but I think you just have to have some tolerance, therefore. The consumer has every right to expect it to be perfect. But I, know, from being inside the sausage factory, how much that can almost never happen, especially in large organizations that have probably completely different orgs making those things, and maybe not enough cross team alignment.

Every big organization I’ve ever been in is literally either in the process of becoming a little bit more decentralized or more centralized, or it’s finished doing one of those things and it’s about to start doing the other one. And they never find the perfect amount of decentralization and centralization for all these different overlapping things. So, you get matrix reporting. I have my boss, but I also have my practice leader. And then one day my practice leader is my main boss and I’m embedded in a team and we’re a service bureau. And it’s like, none of these models are right or wrong, but they produce software like that or experiences.

In other words, it’s a complex situation. Multiple teams are involved, each with separate requirements and goals. Each teams’ needs are continually evolving, and likely do so independent of each others’ needs. If the teams focus on particular platforms, they may want to explore new features that aren’t available on the other platforms. As a result, the navigation structure for each diverges over time from the “ideal.” (As far as there is one.)

But ultimately, it may be that achieving consistency across products in the family doesn’t move the needle — i.e., it won’t make much difference from a business perspective. As Christian went on to elaborate (in relation to another Kindle structural issue):

I also think sometimes you get into the difference between power users and ordinary users. I’ve worked on software where we burned a lot of cycles at times thinking about how to make the switching between your two accounts’ experience better, or the managing your multiple accounts. Until somebody looked at the data and saw that only 2% of the users have even the second account, let alone multiple.

So, I hate to say it, but maybe the long tail of Kindle readers don’t have more than one screenful of books or whatever, and investing in a great system for organizing your huge Kindle library just isn’t going to satisfy big enough fraction of their user base.

That’s an excellent point. If the incentives aren’t there, then why bother? Maintaining consistency requires effort, which comes at the expense of doing other things that may be more important, such as building new features.

I find Christian’s arguments compelling. Whenever I fail to quickly locate the “Sync to Furthest Page Read” link on my Kindle Oasis, I wonder: how many other people are frustrated by this issue? Is it costing Amazon money? Would fixing it be likely to make them more money? If not, it’s unlikely to get fixed. Ultimately, achieving and maintaining structural consistency requires a compelling business case — it’s unlikely to happen otherwise.