Well over a hundred information architects, interaction designers and UI programmers gathered this week at Lillehammer for the Yggdrasil 2008 conference. It was a pleasant and social event, with unexpected and charming entertainment by Dataforeningen’s very own singer/songwriter Henrik Wigestrand, singing about digital beauties with brains.
The business value of usability was the topic of Jeff Johnson’s key note presentation. Investments in usability may reduce maintenance costs, and boost revenue growth through a simpler, and more focused product implementation. The result is shorter time to profitability, which occurs when accumulated revenue exceeds accumulated development and maintenance costs. Jeff also made a reasonable argument for usability testing as a risk management strategy, especially for web sites with external and diverse users, and where usage is optional and transactional of nature. I would think most e-commerce sites are right in the sweet spot for this type of strategical investment.
User-oriented information design, a presentation by Torunn Brandt and Peggy Thoeny, gave us a thorough introduction to various techniques for visualizing large data sets. We all use and abuse tables, graphs, maps and diagrams, and I think it takes some careful consideration choosing the right representation for your data.
Tables are by far the most common way to present search results, with one line for each hit linking back to the original document. Sure, it’s a little boring, but remember that the SERP, together with the search box and “Search” button, is basically the definition of Search (with a capital) in the minds of most users. You shall deviate from this convention at your own risk.
Map presentation of search results is clearly a good idea for documents with geographical coordinates, such as classifieds and yellow pages. It works well because the conceptual connection between a map and the physical world is very strong. The meaning of distance is clearly defined. This definition is less clear for cluster representations like Grokker Map View, where distance is somehow connected to similarity. You also hide a lot of information, making it more difficult to determine if documents and clusters are relevant or not.
A timeline graph may be a good alternative representation for a range navigator, like price on Globrix. Search result filters/navigators are usually presented as tables, preferably with five entries or less per navigator (7±2 as always), and a “View more” option to see them all. Try dragging the Globrix price range markers around; it’s actually quite fun.
Subscribe to this blog, and come back for a more in-depth discussion on search result visualization.
Personality matters, according to Jon Gunnar Wold and his presentation entitled “Koselig eller psykopatisk?” (Pleasant or psycho?). It seems to be true for Finn.no and Google, who attract loads of traffic in spite of plain visual design and an apparent lack of bells and whistles.
Jon makes a good point about irrational and obnoxious behaviour in software, like Clippy (Microsoft’s infamous office assistant) and the annoying User Account Control in Windows Vista. I think it’s a great argument for personalization, as I wrote about in my previous post about personalized search. A bit of implicit learning would make most productivity software soo much better, and my life a little easier.
[update] Online presentations from the conference: