UIEtips: Designing for Faceted Search

Jared Spool

April 28th, 2009

If you’re like me, (and hopefully you’re not,) you have books and magazines scattered all over your house. For reasons I can’t completely explain, I always want reading material in arms reach, so I’ve haphazardly distributed my library in every possible room. There’s even reading material in the bathroom.

(I once bumped into the editor of my favorite magazine and told him his publication lived on the back of my toilet. His response? “That’s the highest praise you can ever give a magazine editor! Thank you! You’ve made my day.”)

While it’s great to always have something to read nearby, finding a specific item is close to impossible. There’s no order or organization that even comes close to making anything easy to find. Not ideal for someone who has made their living helping people create usable information resources, eh?

Now, imagine if every other asset in the house (such as clothing, dishes, or financial records) was equally as randomly distributed. The house would grind to a halt.

Yet many organizations find themselves almost in that situation daily. Every week, we hear from clients who have Intranets where every user complains how hard it is to find the things they need to do their job.

Enter the taxonomy. Once you start to organize the information, you need to identify the right way to classify and store that information. And taxonomies go well beyond just the categories on a web site. Done well, they become a tool that you can use repeatedly to structure and optimize almost every business practice.

In today’s UIEtips, we have an article about one such application: using a taxonomy to create faceted navigation. Stephanie Lemieux,from Earley & Associates, shares her tips on what facets are and how teams can implement them effectively. If you’ve been wondering about this guided approach to navigation, this article is a must read.

By the way, we’re very excited about Stephanie Lemieux and Seth Earley’s upcoming UIE Virtual Seminar, New Ways to Think About Taxonomy: The Role of Taxonomies in Your Organization. This is our first seminar on this critical topic — a must for anyone who needs to improve the way their business is managing their critical information assets. Space is limited so register early for the May 7 seminar.

Have you implemented faceted navigation in your web site? What challenges did you run into? Share your experiences below.

4 Responses to “UIEtips: Designing for Faceted Search”

  1. Elizabeth Pek Says:

    Hi Jared, I work at one of Australia’s leading media organisations. We’ve recently implemented faceted search through out our major websites including our rich information news sites (eg. http://www.smh.com.au) as well as our utility sites (eg. http://www.drive.com.au).

    Your design dos and don’ts definitely resonated with us and definitely agreed that you need to build your taxonomy with faceted search in mind. One key learning that we’ve gained from the project on drive.com.au is that choosing the right interface for the search starting point is also very important in helping users discover the ‘right’ content.

    To help people find their dream car, we initially implemented a single keyword search interface (like Google) together with faceted navigation as our new search model on drive.com.au. We thought that the single keyword search interface would allow flexibility (eg. search for types such as ‘Sedan’, or by makes and models). A single keyword box was also considered by the team to be intuitive and familiar – given that it’s synonymous with Google’s search.

    At the time of the project, there were much debate amongst the business owners whether this was a right decision. One group felt that this new ‘Google’ style interface model was too different from the other search interfaces in the automotive category in Australia. They felt for the car category – giving users a more ‘structured’ starting point such as drop downs of makes and models would be better.

    We decided that the best way to solve the debate was by testing with users in real contexts and in real time. So we conducted a multivariate test of these two interfaces in our live environment.

    To our surprise, the Google style interface produced the least business value (as measured by the number of pages consumed and number of leads generated to the advertiser) and that the hybrid version of keyword search box as well as drop downs of makes and models produced the best overall net revenue result.

    We also conducted one-on-one usability testing to see what was going on here. We observed that the hybrid version is successful because it catered for users’ various levels of car knowledge and familiarity with a faceted search model.

    I presented this as a case study at one of our industry conferences last year where I talked about user’s readiness for faceted search. My slides from this conference are available here:

    Just thought I’d share this experience with you and your readers.

    Elizabeth Pek
    Head of User Experience
    Fairfax Digital
    Sydney, Australia

  2. Melissa Cooper Says:

    Hi Jared,
    I’ve just read your article and the one by Stephanie Lemieux regarding faceted search. I’d have really liked to comment directly on her article… Loads of thought provoking stuff.

    I’ve been involved in various faceted search projects on classified sites, and have since begun really thinking about employing my learning on the mobile platform. Obviously the solution needs to be quite different, but I think some of the considerations are the same. Are you aware of any articles on this?

    One thing I think tends to get overlooked with faceted search is designing for many *and* few results. Most articles I’ve seen on faceted search really focus on narrowing search results. However, we should bare in mind that faceted search also allows users to broaden their results as needed. This is really important if a single facet dramatically changes the range of content available in combination with another facet. Faceted item counts are really important as well as designing suggestions for no results, or very few.

    In summary, I think more thought in this area needs to go towards designing for few results. Especially as there are more entry points with navigational elements such as tagging that can bring a user to search result formats.


  3. Stephanie Lemieux Says:

    These are really great comments – appreciate the feedback…
    I also got a great question about ordering facets that I answered on my blog (sethearley.wordpress.com).
    I’ll be sure to take these two points into consideration in a follow-up piece!

  4. Brian Robson Says:

    It seem to me that if you have three facets and each has five possibilities, that gives 125 combinations and not 243 as given in the article.

Add a Comment