Data Visualization of Web Stats - Log Charts & the Drooping Tail

Jakob Nielsen's Alertbox for August 14 is now online at:
  > http://www.useit.com/alertbox/visualizing-traffic-analysis.html
	
Summary:
Using a linear diagram to plot data from website traffic logs can lead you
to overlook important conclusions. Sometimes advanced visualizations are
worth the effort.

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

Intranet Design Annual, call for participation

Think you have a good intranet? Enter it for the competition to find the
world's ten best-designed intranets.

Deadline: NEXT MONDAY, August 21

  > http://www.nngroup.com/reports/intranet/2007

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

User Experience 2006 conference

  > Seattle, October 22-28 
  > London, November 5-11 

We have expanded the conference to 7 full days. I don't expect that
everybody will come for the full week: registration data from previous
events show that most people attend 3 or 4 days. But some attendees have
requested a longer event:

(*) Some people tell us that their boss will only pay for one conference
per year, so they want to make the most of their one chance.

(*) Some people travel for 20 hours or more from Asia, South Africa, or
Latin America. They naturally want to distribute this overhead across as
large an event as possible.

The registration system allows you to come for 1 day, all 7, or any number
in-between.

Full conference program:

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

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

What People Say vs. What They Do

At last week's Search Engine Strategies conference a company (that will
remain nameless to protect the guilty) released the results of a survey of
Web users. Respondents claimed that they rarely go beyond viewing 3 pages
of search engine results, but that they often do go past the first SERP.
This conflicts with empirical evidence, when we observe users actually
using search engines: people very rarely go beyond the first page.

Why do people SAY SOMETHING DIFFERENT in a survey than what they actually
do? In this case, there are two plausible explanations:

(1) People *sometimes* do click to page 2 or even page 3. It's rare, but
it happens. Human memory is notoriously fallible, and exceptional cases
stand out in memory. Thus, respondents remember having looked at more than
one page of search results and over-estimated the frequency of such
searches when answering the survey.

(2) People are ashamed of their superficial behavior and exaggerate how
"responsible" and thorough they are in their Internet usage. It's
well-known from social psychology that respondents provide the answers
that they believe to be the most socially acceptable and that people try
to please the interviewer by saying something to cause this other person
to judge them favorably.

If you relied on a survey, you might think that a placement on page 3 of
search results would be enough to drive traffic to your site, whereas in
fact you need to be on page 1.

This is a cautionary tale to stop using surveys as the primary way to
collect data about the Web. Yes, it's easier to ask people than to run
actual studies, but you have to watch users' behavior to find out what's
really going on.