Thinking Differently About Site Mapping and Navigation
January 17, 2005 | 
29 Comments
Summary: Thoughts about sitemaps, navigation, IA deliverables and the root problems with organization and content on the Web.
In my new position I’ve got to think a bit differently than I’m used to. For the most part I really enjoy that. As you may know, or if you’ve read my stuff for any amount of time, I’m into questioning convention. At the end of the day, I’m also very practical, so I try things out, see what works, and what doesn’t and adjust my thinking and way of working based on that. Recently I’ve been thinking about Information Architecture conventions and processes. One of these is the traditional hierarchical site map used for many sites.
(When I say “site map” I mean as a way to visualize grouping of information, usually for smaller sites, not necessarily as a navigation technique. Some people refer to this as page taxonomies, content blueprints and a host of other things.)
First off, I’m the first to admit, what I do is really “little IA” and I also realize there are many ways to do a site map, and honestly I can see how, for some projects a visual “site map” might not be the best way to represent information and content. I also realize I may be behind the curve a bit. In my previous job most of the projects were similar and I was asked to focus my attentions in many different areas. Now (when it comes to the actual work anyway—my attentions are still needed all over the map) I can concentrate a bit more on information design and I get to work on many projects with various types of information and all sorts of different goals.
So, back to the topic at hand.
Thinking outside the box and around the arrows
I’m really questioning the traditional “home down” way a site map is presented and how that hierarchical visualization (and often times the groupings themselves) drives our design, content and navigation. The concept of “home” is a valid one, although the idea that it’s first, or “at the top” isn’t really accurate in many cases. It makes more sense to visualize it at the center, as kind of a “hub”.
If you saw some of the latest site map deliverables we’ve created, they reflect this more and more.
But it’s really not about the deliverable, although I’ve got questions about that as well. It’s more about the way people search for information on the Web. We’ve talked recently about how some users don’t “see” navigation. I’ve questioned the concept of “home” and there’s been talk about if “Web pages” and where they live in a site’s organization is always relevant. Remember The Page Paradigm? Or Gene Smith’s Beyond The Page stuff?
These are things that, when talking to stakeholders, there always seems to be at least some difficulty. I’ve spent hours trying to choose the right labels to satisfy every internal audience (let alone visitors!) and make sure every bit of content is properly “bucketed.” They want to reach a consensus as to where everything lives, and often have heated debate as to where a page, or piece of content, is grouped.
Part of this problem stems from the idea that items on a Web site can’t live in more than one place within a taxonomy. This, my friends, is unrealistic, unpractical and well…silly. At least in many cases. It might seems obvious, but it can be a real challenge to get stakeholders to see that it’s ok to put things in more than one grouping, even if it’s just via related item linking. Or, maybe a larger and more common issue, to get them to understand that it’s ok to place something into a grouping where it might not fit 100%.
And then there is the Homepage. They want to know what “lives” on the homepage, and often have large internal struggles to get that sorted out. Struggles that can be an ongoing maintenance drain, let alone the effect of an ever changing home- or hub- page has on users.
They don’t realize that their stuff might get more visibility on internal content pages. And it’s not just internal stakeholders that have a problem with these ideas. For example, it always amazes me that people will pay more for one homepage ad than they will for a load of internal page ads. It’s doesn’t always make sense when you think about it, especially when there is an opportunity to relate the ad to the content on the page.
In addition, I can safely say that quite often these struggles are next to meaningless when it comes to helping a user find what they’re looking for. The fact that everything is conveniently grouped within a hierarchy and mapped down from a home page doesn’t help everyone (or even most people) coming to the site looking for information. It can be very helpful to those who begin an the homepage and browse through your site. However, as search engines become more accurate, and as Web services and syndication spread content around the Web and, in some cases, away from the Web browser, this type of behavior will become less and less common.
I’m not trying to imply that we don’t organize and group content at all, I still feel that it is important to do it. It helps searchers, browsers and everyone to have some kind of consistent and accurate organization and grouping structure. What I’ve got a problem with is the idea that one way of organization is going to make everyone happy. This will never be the case, and, as with many other Web challenges a compromise will need to be reached. Unfortunately, the nature of the site map deliverable, and the process behind it, don’t really lend themselves to compromise or consensus. This creates a bloated process and quite a bit of effort that is, ultimately, wasted if it doesn’t help people find what they are looking for.
Home isn’t where the start is
“Home” as a single entry point to a site isn’t really valid for most sites. However, it’s hard to get beyond the idea that “home” is where you begin your quest to find something. Even in Christina Wodtke’s highly recommended book Information Architecture: Blueprints For The Web she talks about typing in top level domains to begin her searches. This is perfectly fine for the examples she is talking about, but I question how common this behavior is anymore. I know when I’m looking for most things, I start with a search on google or a local search engine.
Think about this, for many sites most people enter a site via some page other than the homepage. (That is if they see the site at all, with the rise in syndication and alternate ways to access Web content, many people may never see the site at all.) I know for my own site, most people come in via individual archive pages. Any large content site would have similar entry points. It’s important to have a “home” or “hub” page, but it’s equally important to address a visitors navigation needs from more common entry points.
If you think of your Web site as a maze (which, to your visitors, might be a very apt analogy) with “Start” being the homepage and “End” being the content they’re looking for—most people are dropped in somewhere in the middle. It’s important to get these people back to the start, but more important to get them to what they’re looking for.
Look at Apple. I’d imagine they have quite a few people arriving from Google via searches for “iPod” and “mp3” player. Now, ideally they’d land on the page they were looking for directly from Google, but if not, it’s important for them to be able to not only get back “home” but continue their search from the page they landed on.
It’s about findability
Visitors don’t necessarily care where something lives as long as they have no problem finding it. Via traditional navigation, that reflects (usually) a site map and it’s hierarchy, is only one way people can go through a site and frankly I feel that in most cases it’s pretty straightforward and, if anything, designers and stakeholders only complicate things by trying to make sure everything is “living comfortably.”
The site map is important, but not as important as addressing the paths that people follow through your site in their search for information. Another thing stakeholders tend to want to do is make sure content is prioritized. This is fine when talking about internal goals, and has some relevance when it comes to a site’s visitors, but…and this is a big but…when someone is looking for content that piece of content they’re currently looking for is the most important bit. I guess what I’m getting at is that as business goals shift, and audience and user needs change the value placed on different sections and groupings of content will change as well.
It’s pretty hard to create a hierarchical site map that adjusts in real time to shifting priorities, goals and needs—regardless of where the originate.
Shouldn’t more time be spent on addressing the user’s real needs? We should be helping them to find the information they’re looking for and giving them options to keep them on track when traditional navigation fails. Things like:
- proper meta data (particularly labels and page titles) for search relevancy
- related item grouping and linking
- Indexes
- links within content
- faceted classification and corresponding navigation
- folksonomies
- personalized taxonomies
- a homepage that acts less like a landing page and more like an information hub (or site map - oh the irony)
A traditional hierarchal site map can’t always illustrate these things, yet often times it’s perceived by clients and stakeholders as the final say when it comes to organizing content. It seems to be rather difficult to get people beyond that, even with tools like wireframes, content inventories and page description diagrams. Unless a site’s visitors think of the content the same way as the internal stakeholders (and every other visitor) and they’ve got the same goals and needs, relying on a traditional site map might be expecting way to much.
Filed under: IA and Usability
Comments
1. Matthew Pennell said:
Excellent article, Keith.
I’m seeing more and more references in popular articles around this subject to the increase in feed aggregation, RSS/XML use, or whatever you want to call it. Personally, I think that our own methods of interacting with the web - specifically syndicated content from regularly updated blogs - have skewed our perception of the popularity, and the relevance, of this form of browsing.
Unless you’re running a business log, I just don’t see there being a customer need to subscribe for updated content - perhaps on a very narrow focus, such as book releases from your favourite author; but in general why would Joe Public want to know when content is updated on your site? Or mine (a bank)?
Posted on January 18, 2005 01:41 AM | #
2. Small Paul said:
Findability is key. On some sites, proper structure can be helpful, e.g. as Matthew just said, when I go to my bank’s website I’m much more likely to start at the homepage. If they’ve got some IA that makes sense, I’m likely to find what I’m looking for more quickly.
But you’re definitely right that no hierarchical structure can guarantee that I’ll end up in the right place first time, so related content linking is very helpful when I’ve just missed.
I guess you’ve just gotta explain to the stakeholders the different ways that people will be coming to the site. Possibly via demonstrations.
Posted on January 18, 2005 05:08 AM | #
3. Brian Sweeting said:
As far as content grouping goes, I’ve been thinking lately about how I could incorporate keyword tagging of pages (e.g. flickr & del.icio.us) in order to provide better related links. The hierarchy has become such a standard that it would be hard to move away from it all together, but if you can provide related content, it makes it easier for visitors to find exactly the information they are looking for.
Posted on January 18, 2005 05:51 AM | #
4. Mike P. said:
“These are things that, when talking to stakeholders, there always seems to be at least some difficulty.
So true - it can often be the largest hurdle I run into in a project.
Digital Web always serves as a great example when I want to demonstrate, in a tangible sense, how complex (yet simple) the organization of info in a site can be…
Posted on January 18, 2005 06:10 AM | #
5. david gouch said:
“(That is if they see the site at all, with the rise in syndication and alternate ways to access Web content, many people may never see the site at all.)”
Why do you think that, Keith?
You said that people find their needed content with Google searches, and, well, Google sends you to a webpage. If I wanted my friend to read this article, I would IM him the link; I would not tell him to subscribe to your RSS feed and read the January 18th entry.
What is with this prediction that RSS/Atom feeds are going to eclipse the web page?
Posted on January 18, 2005 06:33 AM | #
6. Joshua Porter said:
Thanks, Keith. You’ve articulated several ideas that I’ve been reading about lately in all sorts of places. How fast can “folksonomy” become a word in common use?
I see much of traditional IA as throwing a solution at the wall and seeing if it sticks. It is about designers using their best guesses as to what would make a good navigation.
Things like folksonomies are the opposite. See what best sticks to the wall and then use that. It is about designers creating navigation based on how users actually use the content.
Maybe that’s the ultimate in user-centered design. Users designing for themselves.
Posted on January 18, 2005 07:35 AM | #
7. Eric said:
Hi,
Yes, we should work on all the additional facets you listed (e.g., metadata, taxonomies, thesauri, etc.) and, yes, the traditional hierarchical site map is inadequate to represent the totality of all this work to a client.
However, I am not in full agreement when you state “Visitors don’t necessarily care where something lives as long as they have no problem finding it.” This may be the case for a certain sub-set of users; but others with differing cognitive styles do focus on labeling and browsing through the labels. If the visitor’s cognitive style involves finding out where something lives based on labels, then we must include those styles.
We should, of course, try to accomodate as many cognitive styles of information-gathering as possible without short-circuiting everybody’s experience in the process.
In my experience, the traditional site map still has utility when thinking about what content should be associated with what labels and to centralize a web/design team’s efforts. Most people in our team know that the site map will always be a limited, abstract entity – but it sure is referred to and adjusted often.
This stems from an ambiguity in the meaning of “where content lives.”
1. “Content lives” meaning the internal unit primarily responsible for maintaining and/or promulgating the resource.
2. “Content lives” meaning the spot on the site where the content can be found.
In well-designed sites, content “lives” in limbo - nowhere, really. It is linked to from the areas the user expects and tagged with metadata allowing cross-referencing and user-initiated filtering from anywhere.
Thanks for the great site.
Eric
Posted on January 18, 2005 07:44 AM | #
8. Mike D. said:
One of my producers has this system he uses when producing freelance sites where every page is a 404. There are no directories and no actual files. Every URL leads to a 404 and he has a big lookup table which silently redirects each request to the appropriate content. It’s an uncomfortable concept to directory junkies but apparently it works pretty nicely. It also has the effect of helping you fight the urge to put everything in a hierarchy.
I do something similar to this on Mike Industries, but it’s more of a hybrid approach. I have directories and files, but I also have a pretty frickin’ mensa 404 system. You can type anything into the address bar of your browser, and you’ll get to the right place. For instance, this link:
http://www.mikeindustries.com/keith robinson
… will take you to a listing of all pages on my site that include your name. If there’s only one page, it will take you directly to that page.
I wish all commercial sites had this functionality.
Posted on January 18, 2005 08:47 AM | #
9. Nick Finck said:
Mike D.: Digital Web uses this same technique. I can thank Cal Henderson for his handy work here.
Mike P.: Thanks for the kind remarks. We did have at least one thing going for us when we were creating that IA, and that is that our users “got” what we were trying to do and our clients worked in the web industry. That doesn’t happen often at the day job.
Eric: I think you are taking Keith’s statement a little too much to heart… I think he didn’t really mean the user doesn’t care at all, but its relatively a much smaller concern than finding the information itself. Am I right here Keith? However, I do agree that there is still a place for a sitemap in the IA’s toolbox. Digital Web had a sitemap even with all of that crazy IA navigation. But, like I told Mike P., our clients were much easier to convince.
Posted on January 18, 2005 09:47 AM | #
10. James Melzer said:
What if you had a background process that automatically parsed user’s search keywords from the referrer URL and re-tagged your content with metadata from their search? You would have an implicit folksonomy for your site without your users needing to actively provide any information to you. This would provide some of the collaborative benefits of del.ico.us without all the work. It would also give you a wealth of data for creating your own implicitly user-driven site map.
Excellent post and great comments, all! Very thought provoking.
Posted on January 18, 2005 10:20 AM | #
11. Jonathan Snook said:
The problem with developing any site architecture is trying to guess how users will find content on your site. Folksonomies aren’t bad on a large scale but I would find impractical on a smaller scale. This may change as the popularity of engines like del.icio.us increase and develop APIs that allow for a greater integration of user-developed architectures.
Imagine your site displaying not only a description of a page as developed from within the organization but also displaying a list of keywords (tags) pulled in from a folksonomy engine (del.icio.us) and displaying related content both on your site and external sites. Oh, that’d be hot. (I gotta start developing that…)
Anyways, back on track. Multi-faceted navigation is definitely a fantastic approach. Having a site index instead of a site map can also help break away from more of the “one bucket” mentality.
Having a site search that actually makes use of this keyword functionality would also go a long way to making content more findable.
Posted on January 18, 2005 10:23 AM | #
12. Eris said:
Mike D, Nick: What is the percentage of people visiting your site who would type something into the address bar hoping that it takes them somewhere? It seems like an odd approach at finding something on a website to me, but I recognize that doesn’t mean it’s odd to other users.
Posted on January 18, 2005 10:24 AM | #
13. Nick Finck said:
Eris: If they can’t find the information they are looking for? I’d guess at 60% Unlike what Mike D. is talking about Digital Web Magazine uses the same mechanism more for redirecting *known* URLs that don’t exist. Even more so, if you look at every URL on Digital Web Magazine, and compared it with the actual file structure under the web root, you’d see that none of these *pages* exist… it’s all using the 404 mechanism Mike D. described.
Posted on January 18, 2005 11:25 AM | #
14. Eris said:
Redirecting a series of known URLs that don’t exisit, I get that and absolutely agree that it’s a good practice to have. But, what I’m really curious about is how many users, when they go to a site and can’t find what they’re looking for, will use the address bar as a substitute search mechanism? This particular user action is just new to me and I’m trying to figure out how common it is so that I can take that behavior into consideration with my future projects.
I understand ‘how’ and ‘what’ is being done, but how often is it utilized by a lost person at a website? Are they more inclined to type their query into the address bar than into a search box when one is available?
Posted on January 18, 2005 11:59 AM | #
15. Mike D. said:
Eris: I don’t think it’s common to randomly type stuff into the URL field at all actually. I just put that in there because it’s nice. Here’s where it helps:
1. If I want to search my own site quickly I can just type in “http://www.mikeindustries.com/whatever”.
2. If someone tries to link to me but doesn’t get the URL exactly right, my system will often “self-heal” it by redirecting to the correct URL.
3. Sometimes (but not often), people will indeed assume you have a clean URL structure and type in something like “http://www.mikeindustries.com/sifr” if they are looking for sIFR.
4. Often times, people will want to link to *all* of my sIFR-related content (The Wolf does this alot), so instead of finding whatever the newest article is, they will just link to the URL in item #3.
If I really wanted to draw attention to the feature, I’d have a line near the top of my pages that explains how to search via the address bar, but I just don’t think enough people use the search feature of my site to warrant giving up that sort of screen real estate.
Posted on January 18, 2005 01:34 PM | #
16. Eris said:
Thanks Mike!
Posted on January 18, 2005 01:49 PM | #
17. William Murray said:
Mike D and Nick’s address bar functionality is interesting, but I believe that there are better ways to implement it. Microsoft does a fantastic job on their site. If you append a specific product to their domain (e.g. http://www.microsoft.com/office), then you are redirected to the product page. In this instance, the behavior is both graceful and helpful.
In most instances, though, it would probably be more useful to display a traditional 404 page with a search component. If someone is looking for a specific topic or keyword, why force them to navigate to a separate search page or simply redirect them to a page that may not appear to be related? (When I clicked on Mike’s “Keith Robinson” link, I got a page about Fantasy Football.) Integrate the 404 message with a search component and provide search results that are augmented by a page description or summary, and the user will probably find the functionality infinitely more useful.
Posted on January 18, 2005 03:21 PM | #
18. Thomas Baekdal said:
Keith: Interesting article!
Mike D: Your approach is very good. I have been using a similar technique (the 404 version) for all the sites I have made in the last year or so. In my case all page requests including news feeds are picked up by a 404 type of page (no physical pages exists), which then calculate the request, contacts the database server, and returns whatever it finds (without a redirect).
Additional trickers could be added - like /article/whatever - would limit the result to articles.
Building a site with no physical pages provides a lot of advantages - one is that faceted navigation is suddenly the standard. In most cases we have to add code to limit how you can navigate.
I think the future is to extend this to create websites that is build around specific knowledge (not pages or structure). Knowledge is often several pieces of information and today we are not helping our readers turning this into knowledge. We might create better navigational methods - but we are still just moving from one piece of information to another. Thus forcing the reader to mentally turn this into knowledge (called short-term memory overload). We need to reverse this, and instead of moving the reader from one piece of information to another - we should incorporate new information into the existing information.
So what you start out with is a full article (like this one), when you request another bit of information, the first one would summarized, similar messages would be summarized too, and what’s left of the second article is those paragraphs that contain truly new information. Same thing happens if you request a third bit of information.
When the readers are finished, they have a summarized information of “what they have learned (knowledge)”, and during that time we have not wasted their time with duplicate information.
Posted on January 18, 2005 07:33 PM | #
19. Mike D. said:
William: Yes, the only thing missing from my system is a primary lookup table to intercept the obvious vanity urls I’d want. So in your Office example above, because “office” would be specified as a “super-word”, you’d go directly to the office section and the built-in smart searching wouldn’t even have to fire.
I do, however, already do a bit of what you’re talking about with regards to integrating search with the redirect. In the “Keith Robinson” link, you are being redirected to the Fantasy Football page because it’s the only page on my site which contains that exact term (Keith won the Fantasy Football Championship… wooohooo Keith!). I do, however, provide you an out if you look near the top of the screen on that page. Using PHP, I give you the original URL you typed in case you’d like to change it. Also, in cases where there are multiple page hits, you get all the search results. Like so:
http://www.mikeindustries.com/apple ipod
Posted on January 18, 2005 08:56 PM | #
20. Joshua W. said:
This 404 redirect system sounds awesome!! I’m curios how you approached this. Do you use the .htaccess file with a rewrite rule… which calls some search script based on the url? Or do you have another approach?
I’m not looking for exact details, just the overall concept of how you accomplish this
Very inspiring…I could see some enormous benefits for a system like this especially on an ecommerce site…
Posted on January 19, 2005 09:41 AM | #
21. Shaun Inman said:
Weren’t you paying attention Joshua? http://www.mikeindustries.com/smart 404 ;)
Posted on January 19, 2005 11:06 AM | #
22. Joshua W. said:
Thanks Shaun!
Pure Genius!!
I’m surprised this is not a more wide spread technique.
Posted on January 19, 2005 11:53 AM | #
23. Jeremy said:
Keith, excellent article. It is my belief that to grow as designers we must focus far less on the appearance of information than its organization or architecture. This article is a great observation into this line of thinking and a great way to re-examine traditional web design trends. Well said.
Posted on January 20, 2005 12:54 PM | #
24. Mark Thristan said:
I agree, this is a nice article, Keith, raising some interesting thoughts on some current IA philosophy. I think that the top-down mapping comes from the fact that this is a user scenario which is easy to imagine: “user arrives at home page”, where “user arrives at any page” might seem to be harder to control from an IA perspective. Nevertheless, there are IA strategies, such as contextual Read More/Related Items areas which do address this.
Additionally, I am interested in seeing how faceted metadata combined with search might blur the boundaries of searching and browsing - imagine the Digital Web categories being used to refine “finding” activities - is this hierarchy browsing or searching? Well, it’s a bit of both, really, isn’t it…
Posted on January 21, 2005 08:05 AM | #
25. P.J. Onori said:
This is definately a subject worth thinking about. The days of domain-centric browsing seem to be dying and therefore our methods of organizing data needs to be re-evaluated. It could be that the most accessed page (either by link, search or bookmark) is the “top” page in the directory. Wow, this is a very interesting concept.
Posted on January 21, 2005 12:00 PM | #
26. Jon Berg said:
I believe that the navigation should be as obvious as possible. You should be able to figure out how to navigate the site in less than 1 sec. Fancy menus are not always the easiest to use. It would be very frustrating to navigate the web if you had to “learn” to use every page you visited.
Posted on January 24, 2005 11:01 AM | #
27. rose said:
Navigation should be easy, however, many webmasters take search engines into consideration. In otherwords, sometimes the best navigation for people isn’t best for search engines
Posted on January 25, 2005 04:54 PM | #
28. Steer Ethelston said:
I think this is one of the biggest issues on the web that very few people are concerned about - the ‘findability’ of pages within your site surely relates to your return on investment through whether or not people find what they want and buy. So in that sense the issue here should be more important that SEO!
Posted on January 27, 2005 07:42 AM | #
29. Tim Parkin said:
If anybody is thinking of using this technique, they may wish to check whether search engines index 404 pages. When I looked at the possibility of using this technique, my research led me to believe they do not (looking at Google.com in Q3 2004). Also, is it really good practise to send a page to a browser with a status code of 404 that isn’t an missing page. Perhaps rewriting the status code is possible on some platforms (also bear in mind that http redirects can have negative consqequences wrt search optimisation)
Posted on January 30, 2005 11:34 AM | #
Comments are now closed