Over the past few years we’ve noticed a pattern on the corporate websites we’ve worked on:
- Most of the visitors enter through search or referrals. By most I mean 80% to 90%.
- Most visitors will see three or fewer pages. In other words, they will click (or tap) twice.
- Less than 10% of visitors will see the home page, and fewer will start there.
We’ve started calling this the two-click (or tap) rule: Most people will enter in the middle of your site, click (or tap) twice, then leave. Unlike the three-click rule there’s at least some evidence for it. 
The two-click rule has important implications for how we think about content and information architecture for corporate websites. Here are three issues that we’re grappling with on a project right now:
Design in media res
From the data we’re looking at, we now assume most people will enter a site in media res—in the middle of things—so our thinking about the site must start there too. And while this is not exactly a new idea for us , it presents a challenge for our information architecture practice. In the world of corporate websites it’s still quite common to focus on the home page, global navigation systems and questions like “what goes on the tabs?” No matter how data-driven our IA practice becomes, it might take years before executives grok this idea.
Designing bushy rather than branchy structures
Instead of a site “tree” where everything connects to a main trunk, we’re starting to think of sites as collections of bushes. These are small, dense bundles of content that exist semi-autonomously. They’re connected through common design and, of course, hyper links, but the central trunk is no longer essential.
If most visitors are having rhizomatic–non-hierarchical, in and out—experiences–the top or two or three levels of site structure may be mostly irrelevant. In fact we’ve started asking ourselves if those top two or three levels, which often feature abstract labels that only make sense if you approach the content from a top-down point of view, are there just to make vestigial navigation and content management systems work. Could we get rid of them altogether? 
Design for Teleportation
For many of the corporate sites we work on, search not only drives a lot of traffic, but it’s the most common link between a user’s need and the content or functionality the site offers.
We need a way to incorporate search into our early-stage evaluation methods. We’re big fans of Treejack, and it’s great for evaluating tasks where someone starts at the home page and navigates through a site tree to complete their task.
But we need to test Star Trek scenarios–someone teleports into a site, hopefully landing close to their destination, finds what they need nearby and exits. This means that to evaluate our IA work we need to have our participants start at Google, Facebook, Yahoo Answers or some random referrer and test their ability to teleport to the right place. We know these external manifestations of a website–how and where it appears in search results, and what appears alongside it–are just as important as the content itself. But we don’t yet have tools to effectively test them. (If you do, let us know!)
- Ammneh tells me that while this is true for many of the corporate sites we’ve worked on, sites with high brand engagement have more people coming directly to the home page through search. It’s always good to check these “rules” against real data.
- There’s very good thinking about this subject from as far back as the 80s, for example:
- The Design of Browsing and Berrypicking Techniques for the Online Search Interface (Marcia Bates, 1989)
- Home Alone? How Content Aggregators Change Navigation and Control of Content (Josh Porter, 2004)
- Core + Paths: A Design Framework for Findability, Prioritization and Value (Are Halland, 2007)
- It’s probably worth noting the specific challenges faced by content-heavy corporate websites. They have to support user tasks, group and label content in a way that’s consistent with the organization’s brand and overall sense of itself, and then convince every internal stakeholder that those two goals were achieved (despite the stakeholders’ differing perspectives on their users, their goals, and the organization’s priorities). Many don’t have the transactional imperative that exists for ecommerce sites, and the clarity of purpose that sales, revenue and conversion metrics provide. In these situations, IA is politics as much as design.
Thanks to Ammneh Azeim and Andrew Wright for giving me feedback on a draft of this post, and to Dave Robertson for lending me the phrase “transactional imperative of ecommerce”