Intranet Information Architecture (IA) Methods

by Jakob Nielsen on November 26, 2007

Summary: In analyzing 56 intranets, we found many common top-level categories, labels, and navigation designs, but ultimately, the diversity was too great to recommend a single IA.

Information architecture (IA) poses a tremendous challenge in designing any navigational system. Historically, intranets have had little in terms of systematic IA efforts; designers typically "structured" intranets according to the organic growth of pages and features provided by different departments. Employees suffered the consequences, repeatedly getting lost in confusing structures with inconsistent navigation options.

Luckily, many companies have begun taking intranet IA seriously, launching systematic efforts to maintain consistent navigation systems using a deliberately designed structure, rather than one that evolved haphazardly.

We initiated an effort to document intranet IA processes and the resulting designs, both in terms of the visible user interfaces and the underlying structures.

We analyzed the IAs for 56 intranets. Our analysis encompassed a wide range of organizations in 12 countries:

  • 33 companies from a variety of industries, including financial services, utilities, and technology
  • 11 government agencies
  • 5 healthcare providers
  • 4 educational institutions
  • 3 non-profits

Of the organizations, 11 were small (500 employees or less), 30 were mid-sized (501-20,000 employees), and 15 were large (more than 20,000 employees).

Architecture without Architects

Ideally, an intranet team would include a professional IA specialist to handle the IA component of the user experience. But that's like saying an intranet team should include professional interaction designers, graphic designers, writers and editors, software engineers and system architects, as well as dedicated usability specialists to do the user research — plus, of course, a brilliant manager with full executive support. That's indeed the dream team. Most companies, however, can't afford full-time specialists for each of the many roles needed to design and build an optimal intranet user experience.

Among the organizations in our study, only 25% had a full-time person dedicated to the intranet's IA. We can assume that companies participating in a project about intranet IAs have an above-average interest and commitment to the topic, so the percentage across all intranets is no doubt much lower.

All is not lost, however. While it's better to have a large, multi-disciplinary intranet team, a smaller team can do good work if some members take on multiple roles. People in other jobs can learn the basics of IA, just as they can learn the basics of usability. Just as all design teams should do user testing — even if they don't have a specialized usability professional at their disposal — all intranet teams should take IA seriously and take systematic steps to improve it, even if they lack an official "information architect."

On such teams, a designer, usability specialist, or writer typically takes on the information architect's role. Given a little training in the most important IA principles — and the resources to base their design decisions on user data — teams can certainly achieve great IA results without full-time, dedicated information architects.

One of the most important goals on an IA project is to institute a consistent user experience for two key elements: the visible navigation user interface, and the underlying — invisible — structure (where things are found on the intranet). To successfully achieve this, teams must:

  • Decide to proactively design the IA instead of letting it evolve.
  • Ensure that management supports the central IA designer's authority to provide guidance and structure to other departments' intranet work.
  • Ensure that management won't second-guess the design team and impose the awkward structures or navigational terms that individual executives happen to like.

Generalist Tools, Simple Research Methods

In our study, the only tools that IA projects used with any frequency were good old Microsoft Office and Visio. Only a handful of organizations employed anything more specialized.

For example, only 4% of organizations used computerized tools for card sorting studies. This certainly makes sense: you can get most of card sorting's benefits by sticking to the simple technology that the name implies (that is, actual index cards).

Our study's teams used only four research methods frequently: surveys, card sorting, traffic analytics, and user testing. These are certainly the most important methods; any organization that employs all four will have a sound foundation of empirical data on which to construct a good IA.

Only a few organizations used more advanced methods, and most would have benefited from a significant increase in systematic user research. For example, only a few organizations conducted field studies (also known as ethnographic studies), where users are observed in their natural environment, such as an office or the factory floor. A few organizations used search log analysis to great effect to improve the IA and identify short-cut menu candidates. After all, when users search for something, they: (a) want it, and (b) don't see it in the navigation.

The good news is that most organizations in this study employed some kind of user research as a foundation for their intranet IA. This is a huge advance over the early days of intranets.

Common Categories

The median number of top-level intranet categories was 7, but the number ranged from 3 to 31. There was absolutely no connection (a correlation of 0.006) between an organization's size and the number of top-level navigation categories on its intranet. So, just being big is no reason to have an expansive top-level navigation. It might work just as well to have a tightly focused navigation system.

Only three topics attained top-level navigation status on the majority of the intranets:

  • human resources (HR) information (66%),
  • company information, (63%), and
  • news (59%).

Information about departments or divisions was a top-level category in 46% of intranets, and there was a very long tail of additional categories found in a smaller proportion of intranets.

When we started this project, we had hoped to produce a recommended IA for intranets. Although structural diversity ultimately made this an impossible goal, we did identify an IA skeleton that projects can use as a starting point and adapt to their local circumstances.

Many intranets follow several general patterns. Certain types of companies also tend to follow particular trends. For example, manufacturing companies often include a product-related category in their top-level navigation, whereas companies with a focus on intellectual property often present a top-level knowledge management (KM) category.

Personalization and Customization

Many intranets offered a Quick Links feature with shortcut navigation directly to popular destinations. Some even had more than one such feature. This choice is almost always overkill, and can actually reduce usability as users lose track of which shortcut menus contain which links. With one Quick Links area, there's no doubt about where to look, which saves users considerable time.

Users should be able to customize the Quick Links list. In our study, some intranets definitely made doing this easier than others. It's always hard to get users to customize features, so it's worth spending time to make the default presentation as useful as possible right out of the box. And, if your customization UI doesn't have incredibly polished usability, it might as well exist for geeks alone, as no one else is likely to use it.

Because users don't like to spend time on customization, many intranets in the study had invested in personalization, which is generally getting better over time. For large intranets in particular, simple personalization can make the navigation more manageable for average users by emphasizing the things they need the most.

Still, don't forget the simplest, most "Web 1.0"-like navigation aid: cross-reference links. On many intranets, Related Links lists were immensely helpful, particularly when they were designed consistently so users always knew where to find them and what to expect.

Timeless Design

In analyzing the intranet IAs in our study, it was striking how often the user experience got into trouble because of changing circumstances. In some organizations, reorgs were almost an annual phenomenon, dooming any IA that relied on department-associated content. In such cases, navigation menus and URLs were in constant flux.

Although easier said than done, it's important to try and envision future changes and make the IA sufficiently resilient. Doing so will help the intranet remain reasonably stable despite reorgs, corporate mergers, new company directions, new projects, new features, and simply the inevitable growth of content over time.

In our study, task-based structures often endured better than intranets organized departmentally. In our user testing of intranets, we've also found that task-based navigation tends to facilitate ease-of-learning. Thus, the benefits for IA durability are just one more argument in favor of adopting a task-based structure for your intranet.

One change that can be disastrous for intranet navigability is requests from top management to change navigation labels to reflect the latest buzzwords or corporate fashions. It might make sense to make up a fancy term to brand the CEO's latest initiative or management priorities. But such neologisms should always be restricted to narrative content. You should never use them in the intranet's main navigation system, where they'll only wreak havoc and lead users astray.

Ask management if it's worth a few hundred thousand dollars in lost staff time to put made-up terms in the intranet menus. Usually it isn't, particularly because you'll likely be asked to remove the fancy terminology next year — probably just after people have finally gotten used to it. Stick to long-term terminology for navigation labels. These labels are the most costly words in the company in terms of the number of hours employees will spend pondering their meaning.

Fat Report

The full Intranet IA report is available for download. (Note that this link leads to a newer version of the report than the one described in this article. However, most of the findings were confirmed and are still discussed in the new report.)

At 1,350-pages with 1,108 unique screenshots, this is the largest report we've ever published. IA isn't really that complicated a topic, so what gives? The report is long because we included an unprecedented amount of supporting evidence for our analysis in the form of the complete IA trees for 56 intranets, including detailed screenshots of all their navigation systems.

We offer this extensive documentation because our research shows that intranet IA designers face a tremendous challenge: because of the very nature of intranets, designers can't see how other intranets structure their information and can't play with other intranet navigation designs. Although you still can't actually play with the 56 intranets in this study, by publishing their complete IA details, we do let you dig into them as deeply as you like and investigate how they've solved problems that might be too specific for us to comment on in our analysis, but crucial to your own work.

And don't worry. Even though the full report is rather big, we report our main conclusions and recommendations in the first 265 pages, which are a relatively quick read. So, first read through this entire overview, and then dig into the details to your heart's content.

Although there's no single solution that you can copy directly, there's much to be learned from the diversity of solutions that teams have designed to address varying IA challenges. Borrow from their strengths and avoid their weaknesses, and your solution can be even better.

The1,108 screenshots also bear close scrutiny. Their primary purpose is obviously to provide a gallery of design ideas for IA projects, but there are many other interesting design details to be found. For example, we're quite enamored by the way McDonald's presents its intranet usage statistics: "27893 Visitors Served Last Week," echoing its famous signs proclaiming the billions of hamburgers served. Even though intranets are a serious business tool, and a good IA should promote employee productivity, there's no reason we can't lighten up a bit in parts of the interface, especially when it aligns the user experience with the company culture.


The latest intranet information architecture findings are discussed in a new article Intranet Information Architecture (IA) Trends.

Share this article: Twitter | LinkedIn | Google+ | Email