Large Sites and Information Architecture

I apologize in advance, because this is going to be more of an article of philosophy, than a technical how-to.  A while back I wrote a piece on the subject of breadcrumbs.  In it, I made a comment about how breadcrumbs are a tool that helps expose the information architecture (IA) of a site to a visitor.  A discussion ensued where a rebuttal was offered that visitors simply don’t care about a site’s IA.  David Poteet said about IA (or rather a breadcrumb’s effect on IA) that certain ideas “…make the assumption that users understand or want to understand the structure of our sites.”  I ceded that point based on the fact that I believe he was, in part, correct.  Today, however, I wish to expand the topic some, and explain that while I believe they don’t care, they also certainly DO care.  Your information architecture plan is very important, and you should understand why things get placed where they are.

Information architecture is not science, it is art.  Like conducting a symphony.  It’s one part knowledge, two parts skill, and three parts just saying stuff like you know what you’re talking about and hoping others will go along with you (How often have you had to defend why page X was in folder Y instead of subdomain Z?  And with all the XYZs, imagine you did the site for Sesame Street!  I think my head just exploded…).  But at the same time, if users don’t care about it, why bother with it at all?  I’ll tell you why.  Let’s take, for example, two sites: my blog and your university site.  Both have very different IAs (In theory.  If they don’t, one of us is in serious trouble, and my “I’m with stupid” t-shirt is at the ready.).  My blog’s IA has very little active thought behind it, and is almost totally informed simply by the functionality of the CMS that I use for it (WordPress).  That’s not to say I didn’t put actual thought into a couple areas, I did, it just wasn’t much as far as these things go.  A university site, on the other hand, is vast, and generally requires constant thinking and rethinking and questioning of why and where things go.  You need a blueprint to inform that construction process.  Even the best CMS can’t just spit out a functional IA based on how you put information in to it for a site of that scale.

In a way, this compares to two businesses - let’s say Bob’s Lumber and Home Depot.  Bob’s Lumber is your local, downtown lumber yard.  They’ve been there for 50 years (But unfortunately due to the economy, probably won’t be there for 50 more.  That has nothing to do with the example, just trying to build a whole picture).  Then there’s Home Depot.  They carry ten times as much stuff, and have eight times the square footage of floor space.  In terms of these two stores’ “information architecture,” no one would argue that Bob’s needs to have maps posted around his store.  No, you might look overhead for a sign (landmark) once in a while, but you would assume it’s small and structured enough that even if you’d never been there, within a minute or two of walking in the door, you probably would feel like you could track down what you wanted (this would be like my blog).  Home Depot, on the other hand, has a map as soon as you walk in the door.  Then they have more posted throughout the store.  There are big signs over every aisle, and each aisle has additional itemization of what’s there.  They still have stuff organized in a familiar and (semi)logical format, so why the difference?

compass 256 Large Sites and Information ArchitectureThe difference is wayfinding. Peter Morville talks about this at some length in his book (which, for the record, is a great book and I’ll be reviewing it here shortly).  The way we navigate space and the web, and the way we assign symbolic references to the web that mimics space is intriguing.  Also, Jared Spool from User Interface Experts has a nice mini e-book on this subject that David mentioned call Designing for the Scent of Information.  The reason I bring it up is that the idea of the way we apply “scent” to words, layout, and structure is absolutely part of the “instinct” of wayfinding we use.  David talked about scent in the breadcrumb article’s comments, explaining that to be useful, breadcrumbs need to be properly scented.  That notion applies to many things though.  Header/footer links, primary navigation, even the header tags and descriptive text.  Like a bloodhound, our brain digs into that page to smell out where it is the instant we land there, because assuming you need to continue on from there to accomplish your goal, or because your interest has been piqued about something else, you’re going to have to know where to go from there.  This isn’t something you do consciously, it’s just the reaction of our wayfinding instinct kicking in.  If you haven’t put a structured IA in place, that instinct will get lost, and the ability to use a site will go down.

Another example: imagine you are driving down the highway.  You have no map.  But at least you know you’re on a main highway (it’s four lane and has a 70mph speed limit) and that you’re headed where you want to go (sign just said “Kansas City 45mi.”).  But oh crap, you’re out of gas!  So, you pull off at the next exit to stop at a gas station to fill up, after which you get back on the highway, and continue your trip (eventually arriving safe and sound in Kansas City, even though a stupid bird flew right into your grill and cracked a headlight. Stupid birds.).  But you never looked at a map, so how did you do it?  Scent and wayfinding.  And more specifically the “information architecture” of the road system.  You didn’t get out a map to figure out where a gas station was, and you didn’t need one to get back on to the highway.  You didn’t even need it to know you were on the highway to begin with.  And the best partyou don’t care.  You don’t want to know all the intricacies that go into road planning and exit ramp design, at least not consciously.  Unconsciously, however, is another story.  The whole time you travel you’re noting signs, landmarks, inferring direction, relationships, etc.  Assuming it’s all well designed.  Try dropping a new person on the highways in and around Atlanta, GA and watch their confusion.  This is because their road system grew at such a frantic pace, it didn’t have time to be well thought out necessarily.

Back to one of my first points, I agree with David that users don’t care about how our sites are ultimately organized.  They don’t care how we set up folder architectures, page heirarchy, or content cataloging.  They don’t care if we have breadcrumbs, footer navigation, or tag clouds.  What they do care about, is that we’ve made some effort to expose enough markers to them that their instincts can orient them.  The bigger and broader a site is, the more important this is.  They come with a goal, and that’s that.  But that’s all conscious thought.  Unconsciously, it all matters a great deal.  We are constantly positioning ourselves in the web.  We like to think of it in spacial terms, because it allows us to reference locations easier and describe how we got places.  On a small site (Bob’s Lumber Yard), this is all less important, because there’s less to infer, and thus we build an accurate assessment faster.  But on a large site (Home Depot), you want as many maps, symbols, artifacts, and signposts as you can muster.  A breadcrumb trail, for instance, may never get used.  But, that doesn’t mean it isn’t a component of the visual cues we pick up and make mental notes of when we land on a page.  We don’t usually land on a page going “Where on this site am I?” as our first thought.  But our brain certainly does, because if you come with a goal, you can’t accomplish that goal without inferring aspects of the information architecture.

My point is, always keep in mind how your site grows.  Universities and colleges have been terrible in the past about letting things just happen willy-nilly.  That’s no longer effective.  We should have plans, and goals, and intent with respect to what happens where on our site.  The next time someone asks “Can I put this on the site?” be sure to consider if it’s best under their department, or maybe as a special area in their college site.  While it’s not about being right or wrong, it’s all about wayfinding.

Photo by

18 Responses to “Large Sites and Information Architecture”

  1. Says:

    My favorite line:

    “. . . three parts just saying stuff like you know what you’re talking about and hoping others will go along with you.”

    I feel like that is true for so many things : )

    This is the subjective art part of it. Everyone has their interpretation of where things should go, so department’s sections, where control is distributed, get willy nilly.

    Is centralized control / gatekeeping the only way to maintain a plan? That seems to have it’s own challenges.

    • Says:

      The larger a site becomes, clearly the more difficult good control over IA becomes. The things I’d make sure of: make sure people know there is a plan to the site, offer suggestions when you see things in a strange place, and enlist others to help support it. In our case, each college has a “techie” sort of person that helps with all the web sites under that college. We let them know what’s up and work with them to effectively manage sites.

      In that sense, there has to be a central brain running things, but you have to have a certain amount of trust in how people do things. It’s like that adage: You can have it fast, cheap, and good - pick two. We can’t have it all. But if we acknowledge that there is a problem with old IA gameplans, we can start changes to improve it.

  2. Says:


    Given that visitors to web sites increasingly are arriving below the home page — even on small sites like blogs — the need for wayfinding clues is even more critical.

    To use your examples of the Home Depot and the highway, the analog is that instead of walking through the front door or entering the highway through an on-ramp, you’re just suddenly there, in the middle of everything.

    In the words of David Byrne, “you may ask yourself, ‘Well, how did I get here?’”

    Breadcrumbs and a clearly-defined and uncluttered navigation scheme help the arriving visitor to figure out “You are here.” They also give the visitor an immediate sense of place as their eyes sweep up and to the left, looking for your logo, your tagline or other explanatory information, your search box and your navigation.

    Good IA is so important at that moment!

    • Says:

      You make an excellent point. Imagine searching for “wood” and ending up teleported into the 2×4 aisle at Home Depot suddenly. You must quickly assess you are in a lumber aisle at a home improvement store, because you might have actually been looking for golf clubs!

  3. Says:

    It could be a lot worse - you could have had your “Safe Search” filters turned off.


  4. Says:

    Hi. Came across this post from Digg. Interesting stuff, and agree that exposing portions of the site architecture aid in wayfinding. (Breadcrumbs can also serve to provide for “in case of failure fallback navigation” — when contextually surfaced navigation, or your primary navigation fail).

    That said, this:

    “Information architecture is not science, it is art. Like conducting a symphony. It’s one part knowledge, two parts skill, and three parts just saying stuff like you know what you’re talking about and hoping others will go along with you”

    I have to disagree with.

    The best design decisions you can make are the ones you can defend with evidence: quantitative evidence, from usability studies, task analysis, eye tracking studies, or analytics; or qualitative evidence from user research interviews, contextual inquiry, surveys, or other ethnographic research techniques. At my agency we sell and defend work based on the “guess - think - know” spectrum. When you design based on guessing (informed by standards & conventions, best practices, expert opinion), you have no way to defend yourself other than through persuasion. If you design based on what you think you know, from user research, you’re much better off. If you run a usability test and validate your design decisions, then you know your choices were right.

    The goal of any good information architect, user experience architect, ui designer, or what have you, is whittling down the number of assumptions and guesses you have to make.

    • Says:

      You are very correct. I agree, that if you are doing things right, you should be able to back up what you are planning and doing, and the right background and research ultimately plays a large part in that.

      But, where I think things differ is in execution. In higher ed, we are always defending decisions, sometimes to the savvy, more often to the not. These “nots” don’t know what a bounce is, they don’t get siloing, and they rarely care. It just comes down to whether or not you can spin it in a way they will care about. Maybe that means falling back on the research, or maybe it just means telling them what they want to hear to get the green light to do things your way (i.e. the right way).

      Absolutely I could take an afternoon to review them on usability testing, lexicon, flow charts, etc, but in the long run, it’s usually easier (and cheaper) to just convince them in a simple way and move on.

  5. Says:

    Let’s not forget that good IA is as much for the team that maintains and operates a site as it is for the end visitor to the site. That’s why, on a personal blog, IA isn’t as big a concern as it is on a large site. You’re the only writer on the blog, the only person maintaining it, the only person that has to determine any future changes. Since it’s a one person show, you can do, or not do, whatever you want architecturally. If it turns out to be a problem, change it and move on- case closed.

    Alternatively, for a large site, you have multiple people involved in the operation of the site. Any change or decision affects more than one person and any miscommunication or difference of opinion can create friction and, perhaps, hard to remove consequences like politics, power struggles and/or poor teamwork. IA, in that sense, comes to the rescue.

    A well crafted IA phase where key people collaborate and are empowered to contribute their best efforts serves as an agreed upon platform for everyone to work upon. If a usability problem crops up, then the team as a whole is responsible. There is less chance of finger pointing or other negative consequences because the team was involved as a group in laying out the roadmap. A problem is therefore shared, worked on and solved as a cohesive group. I could go on, but that’s the power of IA and why it’s important.

    The benefits that site visitors get from it can be regarded as icing on the cake if you care to look at things from a different perspective.

  6. Says:


    Funny, a coworker and I were discussing this exact thing today at lunch. I just finished a draft wireframe for our new home page and we were going over the contextual clues that the wireframe provided users and the flexibility that the layout would provide subsites with different contexts.

    Specifically, we were talking about how people instinctively wayfind on sites using clues that they are not even consciously aware of and how the wireframe leveraged those, particularly in cases where users teleported to a page deep within the site.

    Your post will go a long way in helping us explain/defend our decisions.


    • Says:

      Yay! It’s always nice to know that I can inspire the great @tonydunn! Hehe.

  7. Says:

    @fienen: I liked the highway analogy. When I read the first post, I immediately though about how when I navigate to any point, on the Web or otherwise, I personally like that sense of context. When I drive, it comes in the form of a Google Map. On the Web, it could be breadcrumbs or a site map or a search. Sometimes you need to look out over the big picture.

  8. Says:

    Fienen, you have me thinking here… as important being able to know where you are when you land on a site and having a clean and clear IA is we still have 50%+ of visitors who simply give up on trying to figure out where they are and just use your search option.

    I guess the point I’m making is that the IA for a large site is VERY important in the usability/accessibility and ultimately search engine optimization of a site.

  9. Says:

    We’ve decide to create our website with WordPress. Work is not too fast but easy and quickly.

  10. Says:

    Great post, Micheal. I’ve already read the “Ambient Findability” and it’s really a great book indeed!

  11. Says:

    Great post. I agree, my experience shows that there is something perhaps at a sub conscience level that users acknowledge as an IA-or at least IA clues.

    You can say that they don’t comprehend the IA, but test after test seems to confirm that they do indeed infer an IA regardless of how explicit it is.

  12. Says:

    Great post, Michael. It really is amazing how difficult good, clear information architecture can be when dealing with higher education.

    For a standard commercial enterprise, information architecture is fairly simple. You have products and/or services. You divide them up into categories, add an “About” page and a “Contact” page and you’ve got a well-organized, sensible site.

    However, for higher education, we have to consider our audiences more than we consider our “products” and “services” and have to come up with six or seven different architectures for basically the same information. That makes for a lot of necessary duplication, and can cause the architecture of the site to become extremely confusing.

  13. Says:

    You make a very good point in that universities have terrible insight on “future” development and usually do not jump into any web project with scalability in mind. I’ve seen this one too many times when they paint themselves into a corner and have to restructure their entire site only a year later. Good use of tuition money huh? Thanks for the information!

  14. Says:

    Thanks for crafting a very nice argument for well-thought IA. Obviously, it’s critical for the sprawling mess that is a higher ed site. The highway analogy is dead-on. I also appreciate what Tim Windsor was saying about being dropped in. Search is so central to the modern site that to ignore the fact that users don’t necessarily drill into your site from the top is crazy.

    Probably the one question I find myself asking when making IA decisions is “Is this content important to a user?” - and for much of our content the answer is no. Unfortunately, that unused content may contribute to a larger picture in some way (such as demonstrating program goals). The problem I find I’m often wrestling with is finding a balance between content “completeness” and content “simplicity”.

    Any suggestions?