Information Architecture for the World Wide Web: Designing Large-Scale Web Sites

3.2. Information Needs

When a user comes to a web site to find something, what does she really want? In the "too-simple" model, she wants the "right answer" to her question. Indeed, right answers do come from searching databases, which store facts and figures and answer questions that really do have right answers, such as "What is the population of San Marino?" To many of us, database searching is the most familiar model of searching.

But web sites store much more than highly structured data. Not surprisingly, text is the most common type of data stored, and text itself is made up of ambiguous, messy ideas and concepts. When we go to a web site for advice on retirement investing, to learn about restaurants in Mendocino County, or to find out what's happening with the Manchester United football team, we are essentially looking for ideas and concepts that inform us and help us make decisions. The answer, if there is one, is an ambiguous moving target.

So back to the question: What do users want? Let's use the metaphor of fishing to get at the answer.

The perfect catch

Sometimes users really are looking for the right answer. Let's think of that as fishing with a pole, hoping to hook that ideal fish. What is the population of San Marino? You go to the CIA Fact Book or some other useful site that's jam-packed with data, and you hook in that number (it's 29,251, by the way). And you're done, just as the too-simple model would have it.

Lobster trapping

What about the times you're looking for more than just a single answer? Let's say you're hoping to find out about good bed-and-breakfast inns in Stratford, Ontario. Or you want to learn something about Lewis and Clark's journey of exploration. Or you need to get a sense of what sort of financial plans can help you save for retirement. You don't really know much about what you're looking for, and aren't ready to commit to retrieving anything more than just a few useful items, or suggestions of where to learn more. You're not hoping to hook the perfect fish, because you wouldn't know it if you caught it. Instead, you're setting out the equivalent of a lobster trapyou hope that whatever ambles in will be useful, and if it is, that's good enough. Perhaps it's a few candidate restaurants that you'll investigate further by calling and checking their availability and features. Or maybe it's a motley assemblage of Lewis and Clark stuff, ranging from book reviews to a digital version of Clark's diary to information about Lewis & Clark College in Oregon. You might be happy with a few of these items, and toss out the rest.

Indiscriminate driftnetting

Then there are times when you want to leave no stone unturned in your search for information on a topic. You may be doing research for a doctoral thesis, or performing competitive intelligence analysis, or learning about the medical condition affecting a close friend, or, heck, ego surfing. In these cases, you want to catch every fish in the sea, so you cast your driftnets and drag up everything you can.

I've seen you before, Moby Dick...

There's some information that you'd prefer to never lose track of, so you'll tag it so you can find it again. Thanks to social bookmarking services like del.icio.uswhich were primarily intended to support refindabilityit's now possible to toss a fish back in the sea with the expectation of finding it again.

This fishing metaphor is helpful because it illustrates four common information needs. When you're hoping to make the perfect catch, you usually know what you're looking for, what to call it, and where you'll find itthis is called known-item seeking. An example is when you search the staff directory to find a colleague's phone number.

When you're hoping to find a few useful items in your traps, you're doing something called exploratory seeking. In this case, the user is not exactly sure what he's looking for. In fact, whether he realizes it or not, he is looking to learn something from the process of searching and browsing. For example, the user may go to his employer's human resources site to learn something about retirement plans that the company offers. In the process, he may encounter some basic information on IRA plans, and then change his search to learning more about such plans. As he learns more about the IRA, he shifts his search again to learning whether the simple or Roth IRA plan is best for him. Exploratory seeking is typically open-ended; there is no clear expectation of a "right" answer, nor does the user necessarily know how to articulate what exactly he is looking for. He is happy to retrieve a few good results, and use them as a springboard for the next iteration of the search. It's not always possible to definitively determine when exploratory searching is finished.

When you want everything, you're performing exhaustive research. The user is looking for everything on a particular topic, hoping to leave no stone unturned. In this case, the user often has many ways to express what she's looking for, and may have the patience to construct her search using all those varied terms. For example, someone who is trying to learn more about a friend's medical condition might execute multiple searches for "AIDS," "HIV," "acquired immuno-deficiency syndrome," and so forth. Again, there isn't necessarily a "right" answer. And in this case, the user must be patient enough to wade through more results than is typical with other information needs.

Finally, our failing memories and busy schedules continually force us to engage in refinding a piece of useful information that we've happened upon before. For example, while you're at work, you might surf for a few minutes and stumble on a great but long explanation of Django Reinhardt's guitar technique. Naturally, you won't read it now and risk losing your job. You'll refind it later instead. It's no surprise that del.icio.us users often assign such tags as "readme," "toread," or "readlater" to their bookmarks.

Figure 3-2 illustrates these four different types of information needs.

These four information needs are by no means the only ones, but many of your users' needs will fall into these categories.