Skip to content

Rhaptos Software Development

Personal tools
You are here: Home » Developer Blog » Kathi's Blog » Search results tips from Jarod Spool -- (more scent, most relevant first, no pogosticking, no wacko results, more result per page)

Search results tips from Jarod Spool -- (more scent, most relevant first, no pogosticking, no wacko results, more result per page) Search results tips from Jarod Spool -- (more scent, most relevant first, no pogosticking, no wacko results, more result per page)

Document Actions
Submitted by kef. on 2008-07-15 13:09. Development

Jared Spool's article, http://www.uie.com/articles/search_results_part2/, has some interesting results from researching various companies' search results pages.

  1. Provide a lot of info in the results to support user tasks. He gives the example of price, distance, and room amenities for hotel searches, to support various types of customers. For us, I think that means providing axes of information about the content in the initial results page, which we do fairly well. At some point, including lens information, ratings and reviews will probably be important.
  2. No one ever complains that the page has too many results -- 10 per page can be limiting, but I think right now performance issues force us to limit the UI.
  3. Most relevant results should be first. Well that is obvious. But the interesting result is that people stop searching down if irrelevant results appear (relevancy threshhold). As we get textbook searchers, we really need to reconsider how often modules are obscuring collection results that may be more relevant to the searcher.
  4. Eliminate wacko results. Just thought the psychology here was interesting. Programmers see wacko results as a weird corner case of an algorithm, but searchers see them like Tourette's syndrom (Spool's term) -- random results inserted into an otherwise sane response -- and it makes them not trust the other results.
  5. Prevent pogosticking -- The more times someone goes to look at content and comes back to the search results (which Spool calls pogosticking), the less likely they are to ultimately find what they want. (I think that this would be a good stats problem for our new book -- probability of success given Y pogos) Preventing this is related to the "Provide a good scent in the initial display).
Developer Blog
« November 2008 »
Su Mo Tu We Th Fr Sa
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30            
2008-11-10
13:39-13:39 Suggestion for live site slowness reports
Categories:
Content (55)
Copyright (0)
Deep Code (3)
Development (203)
Markup (22)
Metadata (1)
Printing (7)
Style (9)
Testing (2)
Usability (6)