Information Architecture

What is information architecture?

Information Architecture [IA]

  1. The structural design of shared information environments.
  2. The art and science of organizing and labeling web sites, intranets, online communities and software to support usability and findability.
  3. An emerging community of practice focused on bringing principles of design and architecture to the digital landscape. (Adapted from EuroIA and the Information Architecture Institute.)
  4. The analysis and design of the data stored by information systems, concentrating on entities, their attributes and their interrelationships; the modeling of data for an individual database; and the corporate data models an enterprise uses to coordinate the definition of data in several distinct databases (in information system design).
  5. Some have suggested IA is analogous with "taxonomy." A contrary view is that the activities involved in the creation of a taxonomy are a subset of the activities involved in developing IA (since developing IA typically also involves articulating the objectives of the information, and understanding the intended audience). (Adapted from Wikipedia.)
IA — The key to a great web redesign effort

Unlike a physical building such as a house, where the architecture is directly visible and tangible, IA is more ethereal, and therefore sometimes hard to explain. (Raise your hands, those who understood all of the above definitions on first reading!)

When applied to a website, IA is more noticeable by its absence than its presence. For this reason, some people will tend to define IA based upon what is often lacking when the IA is not strong. I can't tell you how many of the RFPs for consultants made this mistake by equating IA with a site map, a menuing and navigation system, or some form of search optimization. This is a common, but false, equivalence.

Good IA is the blueprint and the foundation upon which the rest of the site is built. It is essential for the site map, navigation system, and search feature, as well as, for communicating a strategic message. Simply building a site where each of these are addressed independently, however, will not guarantee (and will seldom produce) a strong IA. Solid IA not only provides a framework for developing these features, but allows for the evolution of the site—enabling the site to weather the inevitable changes to message, content, design and navigational structure— thus providing long-term sustainability for the site.

Without a good IA, we risk deploying an aesthetically appealing, technically sophisticated site with great content which will not grow with us. After such a great investment of time and resources in this redesign effort, it would be tragic, a few years from now, to be left wondering why it has devolved to a state (similar to its current condition) which seems unsustainable—where content cannot be found, strategic message is lost, there is no clear context for the placement of new information, and old and stale content are prevalent. Under such circumstances, the administration who funded the project, not to mention those of us who are devoting untold hours in its implementation, will be hesitant to embark upon such an undertaking again!

To go back to the physical analogy of a house, over time we'll often rearrange the furniture, paint the walls, replace carpet with hard-wood floors, and much more. None of this changes the architecture of the house. We might do some heavier re-facing to change the curb-appeal, or update the kitchen from "country" to a shiny stainless steel look. Does this cost money? Sure. Does it take time, talent, and expertise? Of course! But it doesn't change anything about the underlying structure of the house—which walls are load-bearing, or how deep the foundation was laid. And it doesn't take as much effort or money as replacing the foundation, or changing the load-bearing characteristics of the house on top of cosmetic updates.

As it applies to a website, we can swap out the content, design and navigation system. We can even modify our strategic message. This will take time, effort, expertise, and even money, to do well. But with a strong IA underneath, these can all be done much more easily, without having to lay a whole new foundation. Lacking strong IA, changes that ought to be the equivalent of laying new carpet become more akin to adding another bedroom, or possibly leveling the old house and building a new one in its place—after all, that carpet was getting really dated looking!

IA—Grasping the intangible

Let's consider two case studies to try to get a handle on what IA is—or, at least, how we can observe its presence or absence.

Let's look at the positive example first. Here's a search for "red wine" on

Notice that a list of categories is provided which the items that meet our search criteria can be sorted. Whether I'm looking for a book or song about red wine, or a camera (electronics) that is wine red in color, the search is actually useful. Even if what I want is not on the first page of results, I can get there quickly by selecting a category.

What is not apparent, because it is made to look easy and transparent, is that the ability to provide information (in this case items for purchase) in context by category is only possible because the information has this context by category already associated with it. To generate context on-the-fly by some form of data analysis is both slow and inaccurate.

Additional information about our content (whether the content is items for purchase, news stories, or multimedia) is called meta-data. There are various systems for managing metadata, making use of taxonomies, ontologies, or the like... At this point the details are unimportant (that's why we have a consultant, and why the consultant will help us in our CMS selection process). The big deal is to realize that there is a system and process that manages this metadata!

IA—A diversity of examples

Sometimes a lesson can be reinforced by comparing experiences of interacting with differently designed sites—some more transparent and simple than others. In these examples, let's look for a "cordless drill" across a number of sites:

You can click through all of the above if you like—but I'll summarize the results as we discuss them, so don't feel compelled to look at them all. The interesting thing to note here is the diversity of results. Given our previous example, we aren't surprised by the results on—and we find and a little different in terminology but similar in usefulness.

Google is interesting, in that it is a generic web search utility. Nevertheless, it provides results that are surprisingly good in two ways. First, the results are sorted by relevance—a "secret formula" Google has for figuring out what you wanted from the terms, and what pages are most likely relevant to what you wanted. This is automated metadata analysis at its best; and Google pays top-dollar for programmers to keep their "secret formula" ahead of the competition. (There are instances where Google does less-well at figuring out the relevant sites, but in this case, the results are not bad.)

The second surprisingly good element to these Google search results is the ability to focus Google to one type of result. Google presents additional context against which it can make this query. If you click the "more" link, you find additional contexts! From here I built the link for Google Products, which again provides reasonably "relevant" results in a generally useful fashion. We don't have the breakdown by "category" or "department" as we do within a single vendor's site. But we have to give point for the fact that this data is being pulled, all but dynamically, from the web itself!

I have selected Black and Decker for the dubious honor of counter-example. Searching the site for "cordless drill" (a class of item they manufacture) produces a list of results that remind us of Google search results, minus the sorting by relevance or narrowing by context. Remembering that Google indexes the entire web, while Black and Decker's search has a much more limited scope, we have to deduct major points from the latter's effort. It is difficult even to write a meaningful comparison of this to the retailers linked above. Images, prices, descriptions, and categorizations or contexts are provided in each of the respective storefronts, but not within the Black and Decker results.

There are two possible reasons why the retailers above make use of strong IA to help users find what they desire. First, they are retailers—they want you to find what you are looking for so you will buy it! Black and Decker, on the other hand, is not a reseller. So there is less direct financial incentive to increase the usefulness of the search function. The second reason may be a lack of strong IA underneath the system. The failure of the search system to perform well does not necessarily mean that the IA and metadata needed for this do not exist within their product database. It is possible that the search system was simply thrown together quickly without focus on making use of the IA underpinnings. If this is the case, the effort required to code new search capabilities should be attainable, as using existing metadata is not nearly as involved as generating the data from scratch. If, however, the metadata is not accessible (or non-existent), and the IA under the hood is limited, then deriving greater performance from the search feature means one of the following:

  • manually redirecting common search terms to useful landing pages (a never-ending maintenance nightmare)
  • paying top-dollar to get a Google-class programmer to develop a "secret formula" (used to derive relevance and context from the pages themselves)
  • or rebuilding from the ground up with a new IA (one that provides metadata supporting keywords, categories, relevance and context)

In limited defense of Black and Decker, one might argue that customers more often come to their corporate site to browse products by category, or to find information about a particular product. If we search the site for a product number, we get fewer results, with one of them generally pointing us to the correct product page. This is not to say that the retailers do not offer this feature on top of their search capabilities.

IA—the "so what" of the web redesign

What do we learn from all of this? That some sites have better search tools than others? Why don't we just buy a Google appliance and be done with it? Google certainly has great programmers, and can pull some semblance of order from the chaos of the web, can't they do that for a single site? What if I like Yahoo! results better? Can't Web 2.0 just let us aggregate all of our data (or all of the search engines)?

Let's go all the way back to the purpose of the site we're on right now:


What are the goals of the re.web project? We see terms like "navigation scheme," "search feature," "content model" and "information architecture." But the unwritten goal, as we read between the lines, is to establish something that is very unlike the chaos we have now, and something which won't devolve into the chaos we have now. What we want is a web presence that can handle the incorporation of a fresh message, fresh content, fresh media—a site which can even handle the evolution of the web itself!

We can't predict the future. We can't guarantee that every scenario will be covered. But we know from experience that building a web presence in an unmanaged, organic fashion, without any underlying framework, doesn't work! The re.web project is our opportunity to build something which will work, and which will provide the infrastructure for sustainability.

One of the reasons mStoner was selected as our consultant is because the re.web RFP committee found evidence of a strong commitment to IA in the mStoner proposal, as well as, in the backgrounds of those assigned to our project. We believe they "get it"—and they'll leave us with a sustainable architecture for the future.