Innovations in Search Results Pages

by April 13, 2005

I asked tonight’s Recent Innovations in Search and Other Ways of Finding Information Panel (Peter Norvig, Google; Mark Fletcher, Ask Jeeves; Udi Manber, A9; Ken Norton, Yahoo!; Jakob Nielsen, NN Group) about the ubiquity of the search results list, what makes it the right interface solve for displaying search results, and why experiments in search result visualizations have met with little success.

  • Peter Norvig explained that clustering technology is not yet advanced enough to provide meaningful data visualizations to users.
  • Ken Norton pointed out that any search result needs to be really fast and what’s faster to send down the pipe than a text list?
  • Udi Manber said visualizations of search data might become increasingly important for niche audiences. He also expressed concern that we may be standardizing an imperfect solution (with the results list) and used the metric system as a metaphor. The metric system is a better measurement solution than the English system, but we are so tied to the original way we did things (the English system in the US) that changing now is exceptionally hard.
  • Jakob Nielsen leveraged user observations to argue that a prioritized list is basically a simple and usable interface solution: most users simply click on the first link (assuming it is the most relevant) and titles of search results are easy to scan quickly.

However, the panel also touched on a few key issues (prior to my question) that illustrated significant opportunities for improvements to the “de facto” search results page.

  • We are faced with more information (overload) each day. A9 has over 28 million images of streetscapes in its database. BlogLines indexes 1.6 million blog entries a day. This makes information about information increasingly important. A list view though, only communicates prioritization (listing order) and high-level data (title, excerpt, and url).
  • There’s a lack of context around search queries (esp. social context). Information about information (beyond prioritized relevancy) on the search results page could introduce much-needed context for users.
  • Consumer needs are increasingly being better met by “vertical” search functionality (products, local, travel, multimedia, personal, etc.). Search results pages are addressing this by surfacing vertical search paths like “book results for…” and “local results for…”. This change illustrates a need to move beyond prioritized item lists into different organizational structures (i.e. vertical paths and prioritized relevancy).
  • Search queries are becoming more focused on the “long tail” of information. Ask Jeeves has a continually increasing number of unique queries per day. This may mean users are looking for more specific information and would benefit from more contextual information.
  • Search engine abuse, search engine optimization efforts, and paid search placements may begin to dilute the relevancy of the first page of search results driving users deeper into result sets.