7. Thanks to Google (and Bing of course)
• Search technology is wide-spread
• Search is highly adopted
• Search is conceived as an easy tool to explore and find relevant information
8.
9. All I want is
How complicated can it be?!
False expectations
10.
11. Enterprise data is complex
The information needs within an
organization span a wide variety of
information types, sources, formats,
...
Popularity and the number of
referrals are less important in
Enterprise search compared to
Internet Search.
Google builds its metadata from
millions of users searching for
content, the enterprise is a much
smaller case.
In an Enterprise lot’s of people
create content with little attention
paid to information governance.
12. The user is complex
• “This is a huge change to the overall user
experience. It transforms the way we think
and opens opportunities to use search in a
disruptive fashion. I love it!”
• “Personally, I think people will get annoyed
with it. The interface itself isn’t anything
new, and it’s an outdated concept. When
you think about state-of-the-art search, it
should be less about searching and more
about finding.”
15. Expertise significantly impacts how
we seek information online.
The effects on search are
determined by
• Domain knowledge
• Technical knowledge
http://bit.ly/1pQe5dv
The User
How would you
take a picture?
18. Serialists concentrate on the individual parts rather
than the whole
Holists focus on the cohesive whole rather than on
components
Draw a vertical line inside the rectangle
Rod-and-Frame test (Witkens & Ash)
Serialists versus holists
- Spend 50% more time
- Visit twice as many pages
- Are more likely to use the browser’s back button
BUT: the performance gap vanishes if technical
expertise is equally high
SerialistHolist
Source: Kim K. Information seeking on the web: Effects of user and task
variables. Library & Information Science Research. 2001;23 233–
255.6,8.
19. Source: Paivio A. Imagery and verbal processes New York: Holt:
Rinehart and Winston; 1971.12.
Mayer R, Sims VK. For whom is a picture worth a thousand words?
Extensions of a dual-coding theory of multimedia learning. Journal of
Educational Psychology. 1994;86 389–401.11, 13
26. Search is a continuous improvement process
• Small iterations with PDCA cycles
• Requires Management buy-in
• End-users involvement
• Good communication
• Means to contribute
Plan
Do
Act
Check
27. Recall
Definition: RECALL is the ratio of the number of relevant records retrieved
to the total number of relevant records in the database. It is usually
expressed as a percentage.
29. Precision
Definition: PRECISION is the ratio of the number of relevant records
retrieved to the total number of irrelevant and relevant records retrieved.
It is usually expressed as a percentage.
32. How to start?
Capture the user requirements using traditional analysis techniques
or analyze the existing data to analyze search performance and
behavior.
Work like Google is not a requirement!
35. Backwards oriented behaviour
• Autosuggest – help express specific terms and
suggest queries of other users
• Related searches – stimulate novices to explore
related searches
• Avoid zero-results – by using spelling correction,
query expansion, query reformulation
• Breadcrumbs – to navigate back to a previous
query if one is unsuccessfull
36.
37. Designing search user interfaces that are easy to
learn can help bridge the gap between novice and
expert serialists, progressively training them how to
use the application
Source: Spool, J. (2005). What makes design seem “intuitive”? User
Interface Engineering. Retrieved June 8, 2012 from
http://www.uie.com/articles/design_intuitive/.9
Design for learnability
• Descriptive text in search box
• Contextual popovers
• Guidance
• Full-screen overlays
39. When information scent is strong, users are
confident that they’re headed in the right
direction. When it’s weak, users may be uncertain
of what to next, or they may abandon their search
altogether.
40.
41. Examples
• Tonal Patterns (swine flu <> h1n1)
• Synonym patterns (mail <> email)
• Time-based patterns (traffic @eod)
• Question patterns (categorization)
• Answer patterns (content types)
• Find common usage patterns, trends, and
outliers
• Start with queries and their relative frequency
counts
• Eliminate search log “junk”—meaningless
queries—as best you can to improve your
analysis.
43. Examples
• Tonal Patterns (swine flu <> h1n1)
• Synonym patterns (mail <> email)
• Time-based patterns (traffic @eod)
• Question patterns (categorization)
• Answer patterns (content types)
• Try to understand what people are looking
for based on the query cluster
• Works best when done by multiple people
44. Examples
• Tonal Patterns (swine flu <> h1n1)
• Synonym patterns (mail <> email)
• Time-based patterns (traffic @eod)
• Question patterns (categorization)
• Answer patterns (content types)
Try to find what type of content users expect to
find to identify potential content types.
45.
46. • You aren’t offering the content that
your searchers want.
• You offer it, but the search engine isn’t
finding it.
• A difference exists between how you
and your searchers describe the same
content.
Diagnose problems and determine what to fix
or improve for your site’s searchers.
47. Example:
http://www.amazon.com/s/ref=nb_sb_noss?url=search-
alias%3Daps&field-keywords=nike%20sneakers%20ruskie
• Don’t be afraid to say you did not understand
to prevent trashing (changing the query
without resolving the problem)
• Focus on providing a way out. Make sure
every control on the page does something
productive to help resolve the no search
results condition.
• Focus on the customer’s goal. Provide the
most relevant recovery content first, while
staying as close as possible to the customer’s
original intent.
48. If you have access to information about who
searched what and when on your site, conducting
session analysis will help you gain deeper insight
into what searchers do and how their needs change
over a short period of time.
49. Audience analysis will help you better understand
how information needs and searching experiences
differ between audience segments.
Challenge the assumption that your users are all
alike.
Audience analysis can beef up your personas or
boost your organization’s existing segmentation
analysis.