In our work of building and maintaining e-commerce taxonomies, we often run into the problem of products not fitting nicely into one single category. Although this problem is not specific to e-commerce taxonomies, their use for navigation and browsing presents a special categorization challenge; the need to lead a wide range of customer types down an intuitive path to the product they are looking for.

Think about something as simple as a pair of headphones: where do they belong in the following hierarchy?

Audio Products

• CD Players
• MP3 Players
• Home Theater
• Speakers
• Stereo Receivers
• Accessories

First of all they are an audio product, so from a purely taxonomic standpoint they could exist as a direct child of audio. However if there is one thing we know from endless hours of user testing, its that people love accessory categories and anytime you ask them to find a product that is small in size they will immediately gravitate to accessories.

Ok, so what if we put them under audio, and then, under audio accessories. Great, but people often look for products in categories which they feel are related in more abstract ways. For example, it is not uncommon for users to say things like:

“I Iisten to my MP3 player with headphones so I would look there”

Now imagine other branches of the taxonomy such as Computers or Video Game Consoles. There are lots of headphones made specifically to work with these types of products. Do we place headphones there as well?

I am sure you are starting to get the idea, headphones are becoming ubiquitous in our taxonomy, but is that a problem?

There is no real hard and fast rule or best practice around polyhierarchy. It is an important part of the taxonomists’ toolbox and an essential feature of navigational taxonomies. However in our quest to accommodate a wide variety of users and achieve maximum findability, it can easily be overused.

One of the biggest dangers of overused polyhiearchy is that the principles that govern the framework of the taxonomy become diluted. This can result in the placement of products in categories based on overly abstract associations, and can lead to a whole slew of governance problems as the taxonomy evolves.

Another problem with polyhierarchy has to do with its implementation on websites. In some e-commerce applications of content management, polyhierarchy is achieved through a single master location of a given product that gets linked to from other locations in the taxonomy. This can lead to some very disjointed navigational experiences, especially when the polyhiearchical treatment involves multiple branches of the taxonomy.

For example if the master location of headphones is as a child of Audio, then the following pathway could be possible

Computers > Computer Accessories > Headphones

(select and be directed to a new breadcrumb trail)

Audio > Headphones

All of a sudden the user finds himself in an entirely different branch of the taxonomy. This can be a jarring experience that detracts from the learnability of the taxonomy structure.

The next question then is: does the average user really care? I would argue that most users are more concerned with finding products than the purity of the navigational pathway involved. They can always resort to the ever present “back” button.

That being said, the best way to determine where polyhiearchy should be used is with user testing. Products that are candidates for polyhierachy should be tested with a broad spectrum of users. Analyzing their chosen access points into the taxonomy will give you a clearer idea which associations warrant polyhiearchy and which ones do not.

As a general rule we tend to err on the side of too much polyhiearchy, even though our taxonomist spines may shiver at the thought of some navigational pathways. At the end of the day our goal is to help people find what they are looking for. If that means including headphones in computer accessories… so be it.

]]> http://sethearley.wordpress.com/2008/07/21/the-case-of-the-ubiquitous-headphones/feed/ Mike http://sethearley.wordpress.com/2008/07/08/the-pain-and-gain-of-taxonomy-user-testing/ http://sethearley.wordpress.com/2008/07/08/the-pain-and-gain-of-taxonomy-user-testing/#comments Tue, 08 Jul 2008 18:29:45 +0000 stephanielemieux http://sethearley.wordpress.com/?p=70

As a taxonomy consultant, I always recommend (rather, urge with great gravitas) to my clients that they reserve some time and budget for adequate user testing. As they say, the proof is in the pudding: there’s nothing better than quantitative data to tell you whether you’ve built a structure that really resonates with your core audiences and facilitates their tasks. Creating a taxonomy without testing is putting a lot of faith in guessing - albeit, usually pretty good guessing, based on industry experience and knowledge of best practices if you have a good taxonomist.

Having done user testing on taxonomies I’ve built a few times, I compare the feeling to what I imagine it’s like a being an actor or actress watching yourself in a film.

Imagine you are Scarlett Johanssen (I often do), watching yourself in a film… You cringe at the sight of yourself on the big screen, pick apart your performance, thinking “I could have done it this way…”, hear the audience laugh (or not) at jokes… And the pressure of the success of the movie weighs heavily - will it be a bomb? Will the production company go bezerk? Will I ever work again?

Ok, so taxonomy user testing is not quite that dramatic, but there are parallels. Watching users navigate a taxonomy I’ve built is always teeth clenching for me - I am constantly thinking, why are they clicking there? Are they blind? I should have gone with my first idea for that label… The client is going to freak when they find out that one of their star products is unfindable… Will I ever work again?

Of course, that’s the whole point of user testing - to prevent the taxonomy from bombing.

I am always surprised at what I discover in these tests. Categories or labels that I thought were no-brainers can turn out to be black holes of findability. Often this is related to how the taxonomy plays out as a whole in a users’ eyes, rather than the specifics of a particular category. When we build taxonomies, we can get very focused on individual labels and categories and neglect the interplay between different terms across the entire structure. The “stickiness” (or lack thereof) of a particular concept or label can severely affect the performance of other seemingly unrelated categories - a ripple effect of sorts.

For example, a recent test on a toy taxonomy had an interesting ripple effect… One of the labels in the taxonomy included the word “pet”. This term turned out to be so sticky that users looked for any product that even remotely resembled an animal in this category, regardless of how well it matched the product.

User testing also tends to enlighten stakeholders around the dangers of using internal terminology or organizing principles on the customer-facing web. Sometimes clients can be very reluctant to change categories or terms that they have been using for a long time or that represent how they understand their product line internally. They believe customers think the same way and can’t commit to change until hard data proves that users don’t care, don’t understand, or are just plain wrong about particular terms.

I am big fan of Steve Krug’s testing motto: test early and test often. From the taxonomist’s point of view, user testing can be a mix of cringe-worthy moments and fist-pumping “I knew it”s. But there is nothing better than getting real users to point out the flaws and successes in your taxonomy.

For more on taxonomy usability testing, you can download our recent webinar on the topic from our site.

Enterprise Search Survey
Sue Feldman, research VP, content technologies at IDC, and Michelle Manafy, editor of EContent magazine, the Enterprise Search Sourcebook, and conference programmer for the Enterprise Search Summit Invite you to participate in a short (2-4 minute) survey about enterprise search tool selection. They will present the results at the Summit and on the Enterprise Search Center.

Links:
IDC: Http://www.idc.com
EContent: http://www.econtentmag.com
Sourcebook: http://www.enterprisesearchcenter.com/SourceBook/default.aspx
Summit: http://www.enterprisesearchsummit.com
Survey: https://vovici.com/wsb.dll/s/2b4cg3261b

From time to time we organize a free educational conference call series on search, taxonomy or content managment. Next month, we’ll be running our Search Series.

Register at http://www.earley.com/Searchjumpstart2008.asp

Search is complex:
There are lots of tools and techniques that can be applied to solving the challenges surrounding the “findability” of information. Some people think you just put up a search box and that’s it, but in reality effective search is much a more complex issue. Effective search strategies leverage metadata and taxonomies and include a variety of tools integrated as search applications and search systems.

Learn about:

• The value of effective search – making the business case
• Issues and challenges in deploying enterprise search solutions
• The nuts & bolts – understanding and evaluating search tools
• Issues in implementing search and building search applications
• Tuning for improved precision and recall of search results
• Evaluating search infrastructure
• Integrating with taxonomies, tagging systems, etc.
• Faceted search and guided navigation
• Complex federated search
• Natural language queries and semantic search
• Future developments in search

Vendor Sponsorship:
Each session is sponsored by a vendor who will present a 5 minute ‘commercial’ about their offering. In this way we separate educational content from commercial content.

Presenters include:

Lou Rosenfeld, Rosenfeld Media
Rennie Walker, Wells Fargo
Chris Modzelewski, ChartSearch
Mike Ferguson, Intelligent Business Strategies
Ted Sullivan, Raritan Technologies
Theresa Regli, CMS Watch