Mobile Usability Update

The user experience of mobile websites and apps has improved since our
last research, but we still have far to go. A dedicated mobile site is a
must, and apps get even higher usability scores.

Jakob Nielsen's Alertbox for September 26:
  > http://www.useit.com/alertbox/mobile-usability.html

----------------------------------

Usability Week 2011 conference

  * San Francisco: THIS WEEK (Twitter #nnguw)
  * Austin, TX: October 3-7
  * London: November 13-18
  * Las Vegas: December 4-9

Full program:
  > http://www.nngroup.com/events


DEADLINE to save $200 on Austin advance registration:
THIS FRIDAY, Sept. 30:
  > http://www.nngroup.com/events/austin/agenda.html

----------------------------------

DESIGN COMPETITION: APPLICATION DESIGN SHOWCASE

   ** Call for Entries **

Think you have a great application, or a particularly compelling feature?
Enter the Nielsen Norman Group's Application Design Showcase competition
that will recognize the world's best applications.

The winners will be announced in February 2012, and recognized in a
published Nielsen Norman Group report and on our website. Winners will
also be invited to attend a Nielsen Norman Group Application Usability
seminar in Las Vegas (with registration fee waived).

  ** Who Should Enter **

Any well-designed application is eligible, from complex productivity
applications to lightweight Web applets to programs intended for
entertainment. We will consider apps designed for any platform including
web-based, native desktop and mobile applications.
 
While no application is perfect in every way, the winners will include
some features or elements that exemplify innovative and usable design. The
complexity of the problem space and the history of improvement will be
considered when evaluating entries.

  ** How to Enter **

Please see:
  > http://www.nngroup.com/reports/applications/contest

  ** Deadline **
Wednesday, October 12, 2011

----------------------------------

RECOMMENDATION SYSTEMS: matching users' needs only half the job

Recommendation engines are the part of Web pages that say "if you like
this, you might also like these" or that recommend new things for you to
try.

The key challenge is to find things that users will actually like. But
there's one more job, as outlined in a recent paper that studied how
people use these recommendations:

  > http://dx.doi.org/10.1145/1864708.1864717
	
Sylvain Castagnos and colleagues from Ecole Polytechnique Federale de
Lausanne in Switzerland tested users shopping for perfumes.

In the second part of the study, users bought 50% more items found through
the recommender than items found through the site's multi-criteria
filtering tool. This finding is obviously relative to the respective
quality of the filtering tool and the recommendations and will not
necessarily generalize to other sites.

Of more interest is the finding that it was important to provide DIVERSITY
in the recommendations. You might think that the only goal is to find
items that are as close as possible to the user's interest. But that
results in a list of recommendations that are all very much alike and thus
don't help users enough.

It's better to show recommendations that are fairly close to the user's
(predicted) needs, but are quite different from one another. This will
help the user zero in on their true need.

A detailed analysis of users' use of recommendations shows a more complex
picture: in the beginning, diversity is more important, because that's
when users are considering a wide product range and need help on getting
closer to their needs. Later in the process, diversity is less helpful and
it becomes more important to show products that match users' (by now
narrower) interests as closely as possible.

Tracking users' evolving sophistication within a product space is more
easily said than done, so most sites probably can't implement a two-stage
recommendation strategy. However, all sites can tweak their
recommendations to showcase more diversity instead of presenting a bunch
of things that are virtually identical.