Will Max Get Mad When He Surfs Your Website?
Geller - April 24th, 2000
It's more than axiomatic that surfers have low tolerance for sites that
load slowly or make it hard to find information - it's actually true.
Companies need to know whether their websites suffer from this kind of
problem. Most traffic analysis tools tell what pages have been visited
and how long the pages were viewed. By looking at the last page that a
surfer visits a company can deduce what parts of a site surfers don't
like, but this requires that the company first lose those surfers (and
takes a different approach. It relies on determining the behavior of real
surfers against a variety of web site styles, and then rolling the results
into a model of a "typical" surfer. This model, named Max, examines the
pages of a site and reacts to slow loading graphics, placement of links,
the amount of a page that is actually visible on a screen at one time,
the language style in the text, and other factors. MAX is based on a psychological
model called GOMS first developed by researchers at Xerox PARC; the model
incorporates information about such human behaviors as reading comprehension
and motor skills. It has been verified against real surfers and found
to be statistically accurate by an independent testing laboratory.
new study that is being undertaken by the company will expand the heterogeneity
of the model. Max, like most research in psychology, is based on a population
of college students. The new study will enable the company to build a
family of Maxes (and Maxines) that cover other specific demographics,
specifically age and browsing experience. The company could also, depending
on the data collected, create goal-oriented Maxes, such as one that captures
the behavior of surfers shopping for particular kinds of merchandise.
The reports from the product show detailed information about Max's browsing
experience on a URL-by-URL basis, so that a company can zero in on problem
pages and understand why they have problems. The product also provides
comparative analyses of competing sites. The company publishes aggregated
statistics for various vertical segments for free on its website.
WebCriteria's approach should interest website marketers. As a result,
the products should do well. This will not be an easy product to replicate,
although there will probably be some look-alikes if WebCriteria seems
to be catching on. And, especially if they can build interesting models
from their new research, we expect them to make significant penetration.
However, we do not believe that they will replace existing traffic analysis
tools. These will still be necessary on a daily basis even after a site
has been analyzed by Max and optimized. Traditional traffic analysis attempts
to answer questions about what happens when people look at a site; Max
wants to explain why it happens.
are other ways to model human behavior besides the one taken by WebCriteria,
so don't be surprised to see articles comparing the validity of different
psychological modeling theories appearing in the Internet trade press
before too long. Such academic discussions may be a breath of fresh air
compared with the usual news and analyses.
About half the company's sales are made to large corporations, which seems
reasonable. Max is an easy purchase for a large company that needs every
advantage that it can gain in a competitive web market, and may be unsure
about the quality of its Web development. Smaller dot-coms (a quarter
of current sales) probably have more confidence in their abilities, but
if they are entering a competitive market they will also find this a useful
tool. The remaining sales are to consultants and agencies that use the
product for their customers or resell it. We think that any website that
is in a tough competitive situation could find a friend in Max.