Ted Ulle aka Tedster On Information Architecture

Author: Gab Goldenberg
Ted Ulle

Ted Ulle is a veteran member of WebmasterWorld, where he’s better known as Tedster. He moderates the HTML and Google SEO forums there, as well as serving as an Administrator. He’s on Pubcon’s advisory board, the chief search strategist at Converseon, a global online marketing agency for enterprise-level companies, and is on Twitter as TedUlle. You can find out more about Converseon and Ted’s professional affiliations at the end of the post.

I saw Ted give a remarkable and controversial presentation at Pubcon Las Vegas 2010, where he notably criticized most implementations of dropdown navigation as overstuffed. We spoke afterward and Ted consented to give me the following interview. For answers that required clarification, I inserted an asterisk at the appropriate spot and added the clarification question and answer immediately after.

1) Shari Thurow criticizes SEOs who organize information architecture (IA) around keywords. (http://www.youtube.com/watch?v=qyNCxltHrxk ) Assuming you have content that matches those keywords, what is the problem with that practice? Isn’t that “siloing” or “theming” your navigation?

I’d say Shari is right. Information Architecture (IA) needs to be designed around the VISITORS’ needs.

If you start from a foundation of “I need to get these particular keywords into the main navigation”, then you’re very likely to miss the optimum way to bucket your content. What you want is a natural, self-explanatory set of buckets that make browsing the site a joy instead of anxiety.

There’s a major difference in context going on here. Keyword data is generated by people who are using a search engine. That context is pretty much unconstrained, semantically. But site navigation is offered to visitors who are already on your website. The context is now much more defined and constrained.

That said, once you do have visitor-oriented buckets, you certainly should use keyword research. It can inform you of how best to label those buckets for navigation. And I wouldn’t only use search engine keyword tools, either. I would mine social media, customer emails, and I would strongly advise involving customer service personnel in the vocabulary choices.

But again, themes and silo decisions are not, at their most fundamental level, IA issues. They are more like technical SEO issues. When it comes to creating the IA, SEO needs to be on the back burner and any compromises should lean toward visitor needs.

2) When a product can fit into more than one category (e.g. black shoes and New Balance shoes), should it be listed in both? Are alternative suggestions preferable? (E.g. Items in this category overlap with items in {category A} and {category B} ({links}) ) What’s your favored solution?

Human minds make extensive use of cross-categorization. That’s an essential survival trait for our species. The challenge is that no two people categorize in the same exact way. Any team who has worked on Information Architecture for a large site certainly has experienced this challenge, just among web team members.

So it’s inevitable that many schemes for classifying content are going to come up. However, it is essential to choose just one for the sake of not confusing the visitor. The visitor does not know your inventory or content the way that you do. If they navigate to the same item in two different ways, they will not be sure that it actually IS the same item.* Then they begin to feel that the site itself is confusing – and they lose the confidence needed to motivate conversion.

So the main navigation – the core Information Architecture – needs to be a fundamental beacon that the visitor can depend on to stay oriented. The job of the Information Architect is to create a consistent structure that most people will relate to easily.

That said, there are certainly other ways to offer categorization. Faceted navigation systems of all kinds are possible through the sorted site search, or via browsing categories. However, these options should clearly be offered as alternatives or supplemental avenues. They should not be presented in a way that is confused with the main system of buckets that forms the backbone of the site’s IA.

There are significant SEO challenges to these alternative offerings, too. But that’s a topic for another day. The essential point is to pick just one structure for the Information Architecture and then supplement that as needed to serve the minority needs that some visitors may have.

*) I don’t see how people are going to be confused by finding a pair of black shoes from browsing that category and seeing the same shoes when they click under “Brands -> New Balance.” Why would they doubt that it’s the same SKU? Perhaps this is more true for undifferentiated items/commodities?

Yes, I do see what you’re saying. I was trying to cover all types of websites in my comments, not just e-commerce. And that’s probably why it seems like we’re not on the same page. Pure shopping sites are often more like ordering utilities and customers can arrive with a pretty clear idea of what they want. However, when the product mix is diverse, then IA becomes important for getting more items per sale, as well as customer retention for repeat purchases – and that’s a major factor in ongoing success.

I’ve seen the problems that cross-categorized menu choices can make the first hand, especially on corporate sites that are more by information than by direct selling. For a small e-comm site with a very homogeneous selection, users can discover almost anything they want quickly.

For shoes, you probably could have “browse by brand” and “browse by color” with no problem – but is that ideal for the main menu? How do you integrate the non-purchasing information pages, for example? And how do you draw the line on creating more splits? Browse by sport also seems important. Browse by gender? Browse by price? Browse by material? Laces versus velcro? The menu is getting chaotic now and I’m sure there are other divisions possible than the ones I named. What happens if the site expands into casual shoes and dress shoes too?

It’s been fascinating to watch Amazon’s IA evolve over the years. They do extreme testing and metrics, so each iteration is driven by results more than opinion. And they’ve now dropped the tabbed menu at the top completely – pretty much-expecting people to use site search. Effective site search is another major challenge, but it’s not Information Architecture.

I guess Amazon became so huge that they couldn’t find a traditional IA resolution. How do you sell a Kindle book, a convection oven and a Steinway concert grand on the same site, you know? It is interesting to notice that the site search department choices are MUCH more limited than the department list on the left side.

3) Which popular shopping cart currently available makes it easiest to get IA right?

The fundamental job of the Information Architect should be done BEFORE the shopping cart is set up and products entered. That job boils down to two steps. They may sound deceptively simple, but they can hold a world of pain.

1. How should all the bits of this content inventory be grouped?
2. What should we call each group?

As you can see, those two sets of decisions need to be made prior to any technology choice. Most so-called “shopping carts” are a lot more than just a shopping cart. They are also a kind of CMS or content management system. Because they are hoping to be useful for many kinds of end-users, they have grown in flexibility and complexity over the years. However, they still tend to embed various problems for both the Information Architect and the SEO.

My favorite third party shopping cart has no pretension of being a CMS – and that’s Americart. The code runs on a third party server and it is therefore platform-independent. All you need to do is understand their tagging system and then tag your buy buttons. The entire structure of your site and all its pages are completely up to you. You can generate the entire cart with a spreadsheet.

When it comes to the vast array of self-hosted carts that are available, I’d suggest that the Information Architecture be established first, along with other needs. Then the various choices can be assessed according to how well they match up.

Most of the shopping cart challenges I see are SEO related rather than IA related. But a website should never be designed to accommodate the shortcomings of any tool.

4) Jakob Nielsen argues that rollover/dropdown navigation annoys users. Yet dropdowns (when not used as a crutch) can help flatten a site’s architecture, thus improving the indexing of deep pages. What guidance can you give people to get this right?

If a deep page needs more frequent indexing or link equity (pretty much the same thing) that does not mean that page changes its position in the information architecture. There are many other ways to drive the spiders besides getting shifty with the main menu – including links within the content, secondary call-out boxes, most popular lists and so on.

Nielsen is right. Changing the navigation means shifting the visible Information Architecture.* That is extremely disorienting. Imagine a highway that doesn’t standardize its road signs. Think of a hospital that uses inconsistent signage. That is a big problem for hospital signage because the IA of a hospital building needs to serve many different audiences: basic staff, professional staff, visitors, patients, and even investors.

Solving such a challenge is not easy. It often requires a trade-off or three. But we should always trade in the direction of simplicity and stability, rather than doing what seems opportunistic. Otherwise, we confuse the visitor and they abandon our site. And well they should – we are messing with them!
There are a lot of tools in the toolkit. So don’t try to throw every possibility at every possible application. Let your visitors calm down and breathe a bit so they feel welcome and at ease.

*) I don’t see how dropdown menus are like inconsistent navigation. I’d say a better analogy is the digital signage you increasingly see to explain upcoming road construction’s location and alternative routes to the now-blocked destination, in that they add more detail to the main message. Can you clarify your view?

In either case, whether it’s inconsistent navigation or dropdown/hover menus, visitors cannot easily compare their possibilities side by side – they are only seeing one at a time. This violates Steve Krug’s “don’t make me think” principle.

One of the challenges in any Information Architecture discussion is that the principles are highly abstracted – and they need to be applied in very concrete situations to have any hope of reaching consensus. Real differences between web team members are extremely common, just as you and I are having differences in how we see these principles right now. And we don’t even have a real web development project in front of us!

I don’t really understand your detour metaphor, for instance. As I said earlier, people do not all build categories the same way, even though we all build categories. If you and I were working together on a project, we would need to explore the mismatch between our thinking styles. That discussion could be difficult, and that’s why IA is so often abandoned – people want to avoid the feeling of conflict between team members so they buy into some agreement or other.

The end result of persevering through those difficult spots to arrive at a real agreement is the creation of a great website, rather than just another standard website. The last IA project I worked on took 8 weeks of almost daily meetings before we arrived at a stable place.

5) It was common to see a site’s main navigation being largely ignored by users, who focus instead on the active window. Is that still the case today? Or is the ever-increasing familiarity with the web making that a moot point?

Yes, it’s still common to see visitors not clicking on the main navigation. But just because they don’t click a lot doesn’t mean they are ignoring it. It’s a quality signal, a grounding influence that they depend on if they should ever become lost as they browse.

Focusing on the active window is natural – and it means that you’ve actually engaged the visitor. They caught the “information scent” and are at ease. You’ve done your job. On a great website, the main navigation is a kind of “In Case of Emergency” sign.

There is such a thing as navigation blindness, too – something like banner blindness. That can often be the result of a design flaw. If the navigation looks or reads like and ad, then it wasn’t created with the visitor as top priority, but rather with some mistaken idea of SEO or boosting sales.

The best way to boost sales is to keep the visitor comfortable. Every visitor needs to learn a new interface every time they go to a new website. Don’t make them work too hard. A comfortable visitor is more likely to become a customer or client.

6) It is increasingly popular in e-commerce to use dynamic slideshows in the homepage’s active window, instead of offering various links to main product categories (which solves the problem of users ignoring the main navigation). What experience do you have with this? What’s theoretically the best solution? Does that differ from reality?

I’m currently working with a website that recently introduced a jquery slideshow approach so I can see the numbers for a present time implementation. The slide show certainly made the home page look snazzy as hell. The problem is, it isn’t improving sales – it seems to have hurt them.* Even worse, several home page rankings have also slipped since the slideshow’s introduction.

So the current debate is about dropping the whole thing and going with the old-fashioned approach again. That is what I would have predicted had they run the idea past me first. It’s good to experiment as long as you watch the metrics and make quite changes if the metrics show a problem.

It looks to me like this particular experiment had a negative result. As long as we don’t fall in love with our ideas, almost anything is worth a try. My friend at Converseon, Mike Moran, wrote a business book called “Do It Wrong Quickly.” That title just sums it up.

As a general rule, I’d say that slideshows are like any other dynamic navigation – even hover menus. The challenge is that visitors cannot see the possible selections side-by-side and compare them. And with any eye candy approach, you risk distracting the visitor from even noticing what your full offering is. I know that’s what happens to me, and I don’t like it. In fact, I get quite impatient with it.

*) Regarding the slideshow, is it the case that categories that are later on in the slide-deck are getting lower CTR than previously? Have you guys tested a slideshow that used previews like Yahoo.com’s news section?

In the case I’m currently looking at, it’s actually the first slide that’s getting the lowest CTR – no matter which option they put first.

Sorry, I’m not familiar with slideshow previews in the Yahoo news section – and I couldn’t find an example to make an informed answer.

7) How can someone get a deeper understanding of Information Architecture?
There are painfully few resources available, even after all these years of ecommerce. IA is hard work and most people would rather go into avoidance. There are a few resources I do recommend:

1. O’Reilly has a solid book – the best sourcebook I know of, now in its third edition. It’s called “Information Architecture for the World Wide Web“. See

2. If you want to go the free route, Webmonkey has a long-standing IA tutorial. It offers you least a top-level survey view of what IA is all about.

3. And finally – come to PubCon in Austin TX this March, just before SXSW. I’ll be offering an “Expert Spotlight Session” on Information Architecture, expanded from the most recent Las Vegas session. For more on that see http://www.pubcon.com/austin2011-expertspotlight-conference.htm.

Converseon‘s been working with enterprise clients since 2001 and offering a unified approach to all things web. They were recently featured in the Forrester Wave report as having one of the top social listening technologies in the world. We are one of only a very few agencies to have access to the full Twitter Firehose rather than just an API.

If you enjoyed this interview, add my RSS feed to your reader for more like it!

Author: sroiadmin