Information Architecture Revealed!
Conducted via email by John S. Rhodes (24-May-99)
What is information architecture? In general, why is it important?
The field is relatively new and under development; therefore, the definition of information architecture depends on whom you ask. Here's one that works for us:
This means that optimizing your site's search engine to help users find what they're looking for is information architecture (IA). Optimizing your search engine for effective load balancing is not IA. Developing a system of labels to populate the options in a navigation bar is IA. Determining whether the navigation bar is green or blue is not IA.
Why is information architecture important? Well, imagine that you've just invested a few million dollars in your site. It's aesthetically beautiful, technically perfect, and full of wonderful content. But you're finding that users can't find the information they need, and that you can't determine where to put new content and when to remove old content. That's a pretty good justification for IA.
Or imagine that your employees have such a hard time using your intranet to answer customers' questions that they are printing off pages for fear of never finding them again. Of course, when that information changes, your employees don't know, and give customers outdated and therefore wrong information. This is a problem that a good IA can help you avoid. Heck, we've even seen situations where employees made up answers rather than use a poorly-architected intranet. Ouch! Better fix the architecture.
Great question. I'm not sure there is a good answer, at least not yet. But I'll blindly wade into these dangerous waters nonetheless.
I think of usability as a broad field that is as relevant to testing and improving the performance of automobile dashboards and automated tellers as it is to improving how a Web site performs information retrieval tasks. I know that as an information architect, I've learned a lot from usability specialists about users and how to evaluate their experiences in a general sense. However, I don't know that I could learn much from usability specialists about information itself: how content and its tendency to vary greatly affects retrieval performance. I also find that usability specialists often don't have a good feel for how greatly a user's information needs can vary, and how those different needs can have a huge impact upon retrieval performance.
So I guess I see usability as an area with extremely broad applicability, but incomplete if not combined with some domain specialization. We're experimenting with this very issue here at Argus; we recently hired Keith Instone, as well as another HCI specialist, so that we'd have top-notch usability expertise in-house. Next we'll see what kinds of services the market requests from us, and will mix together our information architecture and usability expertise as our projects demand. After a year or two of market testing, my answer to this question may be very different.
First, it's important to recognize that every information system, be it a book or an intranet, has an information architecture. "Well developed" is the key here: most sites don't have a planned IA at all. They are analogous to buildings that weren't architected in advance: design decisions reflect the personal biases of designers, the space doesn't scale over time, technologies drive the design and not the other way around, and so on. So we'll assume that a well developed IA is a planned one.
Assuming that the IA is developed in advance, it's hard to determine the best approach to take. There are so many ways to organize information that using them all in a site would be overwhelming for users. A well developed information architecture helps users by identifying and specifying the *subset* of all these possible ways that are most appropriate given:
At Argus, we believe very much in the 80/20 rule: if we can identify the few major ways of organizing a site that meet 80% of users' information needs, then we've reduced the chance that we'll overwhelm the user with too many choices, many of which are only rarely useful, and all of which have a significant maintenance overhead. Similarly, we often find that roughly 20% of an organization's information is sufficient to meet those major information needs. That makes for smaller, more precise retrieval sets and narrower, more focused and more successful browsing. Not trying to be all things to all users also means that system owners and managers have less to maintain, saving them time and money.
In more concrete terms, a well developed IA means that users will spend less time finding information, and are are less likely to miss finding what they need altogether. They are more likely to find that a relevant document is linked to other relevant documents. They can switch back and forth from searching to browsing more easily. They will have better cognitive models of what content a site contains. They will get along better with their peers, and be better mothers and fathers. You get the idea.
In general, information retrieval is really, really hard. Finding information is analogous to the Telephone Game: you are trying to convert your idea into words, then into text or a search query. Then you either manually browse through documents (very inefficient) or employ a fairly stupid piece of software (i.e., a search engine) to do it for you. What you find is text that hopefully represents the words in the author's mind, which in turn reflect his or her original ideas. Throw in the ambiguity inherent in human language, and you see that there are obviously a lot of places where things can get lost in translation.
Information retrieval research shows that this process of matching user's information needs with authors' ideas has performed poorly in traditional settings, such as using a specific, narrowly focused online database like LEXIS/NEXIS. The Web further complicates the matter: throw together myriad formats, broad topical coverage, and other variation to the body of content being served, and bring up the volume of content a few orders of magnitude. Then add every type of user under the sun (or, at least, anyone who can access the Web). Result: even worse information retrieval performance than ever imagined. The more heterogeneous the content and the user population, the worse performance will be. Apples and oranges shouldn't be mixed, but they regularly are in Web sites, and that's why we get 2,000+ hits on our searches, and why we can't browse the corporate intranet without becoming hopelessly lost.
What is Argus Associates? What is your role?
Argus is a consulting firm that specializes exclusively in designing information architectures. We try to tie together the disparate islands of Web-based content that are typical of the corporate environment in ways that make sense to users. Most of our consultants have backgrounds and advanced degrees in information and library studies, and the company has been very successful at porting the principles of this field to Web sites and intranets. I think that's what makes us unique: we're not only solely focused on information architecture, but IA from the librarian's perspective.
Most of our clients are Fortune 500 companies that are trying to manage "enterprise information portals". In this case, the client and the users want a common, centralized interface to access information from across the whole enterprise. But the content is managed and owned by highly autonomous political units. Argus' approach is to develop taxonomies and other architectural schemes for improving browsing and to configure searching for better performance, all while avoiding over-centralizing content management to the point that it is impractical given the client's culture, politics, and resources. We also analyze content at the document level, looking for ways to make it easier to move from one document type to another. In all cases, our goals are to make it easier for users to find what they need, and to reduce content management costs.
I'm one of the two founders of Argus and its President. Peter Morville is our VP, and we're the two principals of the company. Peter and I have very different and, fortunately, complementary personalities and skills; we've found this to have been a real strength for the company, and have continued to seek diverse skills in the people we hire. Did I mention that we're always hiring?
I'm a librarian. "Shh," and all that. I have a Masters in Information and Library Studies from the University of Michigan, and also did some Ph.D. work in that field. I also have a BA in US history. Being affiliated with the University of Michigan afforded me some fun opportunities over the years; besides working as a librarian, I taught courses in designing Internet-based information and searching for information on the Internet, both in the early '90s. And I also worked as a researcher, playing with Verity's search engine and developing push-based information systems.
I have a chip on each shoulder. Chip #1: I want the world to understand that librarianship has value in the new millenium. Let's face it: there will only be more and more disorganized, unstructured content in the future, and there aren't many fields that are better equipped to work on such problems. After all, librarians have been dealing with these issues for centuries.
Chip #2: I want librarians to get their heads up in the air and actually respect and value themselves. As long as some of my colleagues see themselves as people who work solely in rooms full of books, the profession is as doomed as some claim. But if librarians understand that their skills are highly relevant in any information-rich setting, be it library, Web, or wherever, then we're talking about the hot career of the third millenium. In sum: it's not about libraries, it's about librarianship.
What is the Information Architecture Guide?
It's the Web-based version of the Information Architecture book's bibliography. We try to keep it up to date, but it's hard to keep up with the recent explosion of writing relevant to information architecture. Not to mention the need to keep up with the (billable) demands of our clients.
It's a labor of love. I started the Clearinghouse in 1993 as a repository for topical guides to Internet information; it was intended to serve as a place for my UM students to both sample good examples of guides, and publish the ones that they were creating in my class. It grew far beyond that; there are about 2,000 guides today, each of which we evaluate and, if they pass muster, then index for inclusion in our site. We turn down about 90% of all submissions.
It's never made us any money; maybe because we weren't good enough marketers, maybe because we are more interested in information architecture consulting. Who knows? It'll be interesting to see if The Mining Company can profit from a similar model. In the meantime, we keep it going on a very shoestring budget because it's a good service to the Internet community and because it showcases librarianship skills.
We're experienced: we work with and learn from our clients, bright people who face difficult challenges at Fortune 500 corporations. And we've been doing it since 1994.
We're really good at what we do: the industry hasn't reached the point where an IA company can grow by taking out banner ads and making cold calls. Instead, most of our work comes to us by word-of-mouth, and we get a lot of repeat business from our clients. So, to grow at the rate we have, we must be doing something right.
We've got information architecture (constantly) on the brain: we wrote the top selling book on the subject, as well as dozens of other pieces (columns, book chapters, and so on). We eat, drink, and breathe this stuff: it's the only service our company provides. And specialization is a Good Thing in the ever expanding and more complex world of Web site development.
We're unique: besides the fact that there aren't many information architects out there in general, there are even fewer librarians working in this space. At least not yet. It's not surprising: we don't tend to be the most entrepreneurial profession...
Please briefly describe a few of the core ideas of your best-selling book, Information Architecture for the World Wide Web. Also, why do you think it has been so popular?
After the requisite introductory chapters, we have two major sections. First is coverage of what we consider the core areas of IA design: organization, navigation, labeling, and searching. The next major section covers process: how to develop an architecture in the context of a site's research, conceptual design, and implementation phases. Finally, we conclude with a brief case study. We don't tell you how to design your site; there is no one right way to do it, as every successful IA is built on the unique characteristics of the site's users, content, and goals. We do try to provide an introduction to core IA concepts and provide readers with terminology so that they can have better conversations about the subject, and hopefully design better architectures.
Why is the book selling so well? I'd love to claim that it's due to our revolutionary ideas and spiffy wordsmithery. But I know better: the real reason is timing. Web developers are on their fifth and sixth generation sites these days. They know more than they care to about such tangible issues as how to optimize image files, program functionality in Java, and so on. But their sites still have major problems that are related to more abstract issues of site design, such as architecture and usability. As Web developers began to recognize this on a broad scale, our book came out. Great timing!
I wish I knew. More applications for managing content (especially those based on XML) are coming out, and these will make for more flexible architectures, as well as more reusable content. This is very cool. But I'm pessimistic that there will ever be enough information architects or sufficiently powerful retrieval technologies available to keep up with the amount of heterogeneous, unstructured content that is being churned ever faster. This may be very bad for civilization as a whole. On the other hand, as a businessman in the IA field, it makes me very optimistic...
Peter and I will write a second edition of our O'Reilly book, and will add content dealing with "bottom-up" (document-level) information architecture, something we didn't know much about when we wrote the book a couple years' back.
But first, we'll be writing a book on how to improve the performance of your site's searching system, also for O'Reilly. Now that I've let the cat out of the bag, we'd better get writing... Comments and suggestions for both the new book and the second edition of "Information Architecture for the World Wide Web" would be greatly appreciated! Send them to email@example.com.
Thanks very much for the opportunity to
spout off; I hope to hear from
Now, what should you do?
If you want to know when
new content goes online,
Home | Moving WebWord | Cool
Books | Hot Web Sites | Reports
Newsletter Archive | Services | Interviews | News | About WebWord.com
© 1999 by John S. Rhodes. All rights
Do not reproduce or redistribute any material from this document,
in whole or in part, without explicit written permission from John S. Rhodes.