You Say 'Page View,'
I Say 'Visit': How to Count Web Traffic? (August 26, 1998)
by
Lisa Napoli (New York Times)
[I] n press releases
touting traffic to news sites on
the Web last week,
more than record numbers met the
eyes. There was a
jumble of data, too.
On the day after President
Clinton's address on the
Monica Lewinsky matter,
for example, several of the most frequently visited Web sites reported
numbers that
brought into focus
the difficulty of measuring -- and
comparing -- Internet
traffic.
ABCnews.com
said its Web site received 1.6 million
"visits" last
Monday, while CNN reported 20.4 million
"page impressions."
Meanwhile, MSNBC said its Web site had 1.13 million "unique visitors."
Later in the week,
Relevant ------------------
Knowledge, a third-party
Internet Related Article
measurement service,
released its
figures, which reported
over 3.5
million unique visitors
to 11
different news sites,
including those
listed above, last
Monday and ------------------
Tuesday.
What do all those disparate
numbers and qualifiers add
up to?
"The measurement conundrum
in a bottle," said Rich
LeFurgy, president
of the Internet Advertising Bureau, a
trade group formed
in part to promote standards in
Internet advertising.
"Basically you have different
sites telling you
different metrics that are not being
confirmed by the
measurement companies."
Most Web sites collect
traffic data from their own
server logs, allowing
them to monitor how many different people visited
the site during a given period (referred
to
as "unique visitors"), how many times those
individuals
returned (known as "visits"), and the total
number
of pages those visitors looked at on the site
(commonly
described as "page views" or "page
impressions").
In addition to the
Web publishers' own numbers, there
are several independent
companies, like Relevant
Knowledge and Media
Metrix, that measure Internet
traffic. In general,
these third-party services report
traffic in terms of
visitors, but even when they use the
same unit of measurement,
their numbers often differ
from the figures reported
by Web publishers themselves.
Part of the reason
for the difference can be attributed
to the way these services
gather their data.
Most independent measurement
services monitor the
surfing habits
of a "representative sampling" of online
users, similar
to the way the Nielsen rating service
measures television
audiences. These companies provide
their sample users
with proprietary software that tracks
their Internet use,
then use the data gathered to
project overall Web
traffic.
However, Web publishers
often criticize these ratings
services as failing
to accurately track important
segments of the Internet
audience, including
international users
and those who surf the Web at work.
"It would be like
Nielsen only looking at the workplace
and saying no one
really looks at Seinfeld," said Mark
Bernstein, vice president
and general manager of CNN
Interactive.
Making matters more
complicated -- for both publishers
and measurement services
-- is the issue of proxy
caching, a
method by which some Internet service
providers store pages
of frequently visited Web sites.
Caching speeds download
times and alleviates
bottlenecks, but it
can also interfere with traffic
measurement. If an
ISP has cached a Web page seen by one user, the next user at the same ISP
won't be counted by
many measurement systems.
All of this translates
into confusion for anyone trying
to determine the breadth
and depth of the Net audience.
Largely, that means
Web publishers, and the advertisers
they are trying to
attract.
"However you cut it,
there's not a lot of accuracy in
the numbers. When
a potential buyer of media tries to
compare, and those
numbers don't make sense, it's hard
for the media company,"
said LeFurgy, of the Internet
Advertising Bureau.
"Say there's an ad agency that wants
to make an online
media buy. The sites go nuts if [the
agencies] go to measurement
companies because they don't believe the numbers are accurate compared
to the
[publisher's own]
server data."
Publishers agree that
some sort of standardization is
necessary. Jim Kinsella,
general manager of MSNBC on the Internet, said he understands the frustration
of
potential advertisers
who are presented with disparate
methods of traffic
measurement. "I personally know
they're frustrated.
It's totally confusing. We've got to
get our act together,
got to stop the silliness."
But that means Web
publishers would have to agree on how to represent their traffic -- and
some say that will be
a challenge.
For instance, MSNBC
is committed to reporting unique
visitors, Kinsella
said. CNN, on the other hand, would
rather represent its
traffic in terms of page views.
Each ardently prefers
its own system -- and criticizes
the other for using
a different method.
the other for using
a different method.
"The fact that
someone comes
to our site and reads
10
pages shouldn't
be
dismissed,"
said Bernstein,
of CNN Interactive,
who is
also critical of the
measurement
services for
MSNBC on the their
emphasis on visitors
Internet (as opposed
to page views).
"I don't object
to using
Relevant Knowledge
and Media
Metrix," he said.
"But I do take issue with the
utilization of those
numbers as a correct reflection of
our audience."
For their part, the
measurement companies say the data
supplied by the Web
publishers has serious limitations
as well, because it
doesn't provide demographic and
psychographic details
about the people surfing. They
also acknowledge that
their own competing methodologies don't make things any less confusing.
But Mary Ann Packo,
president of Media Metrix,
attributes part of
the problem to the newness of the
medium, noting that
it took "10 to 15 years" for media
measurement to
evolve for television.
"The great news is
there's tremendous information
available. There's
so much opportunity for this medium
-- it's not just a
frequency and reach medium, it can be
direct response,"
she said, referring to the two factors
commonly used to measure
advertising's effectiveness.
"Because it [the Internet]
is so measurable we can
collect so much more
information than NBC would ever
have about its TV
lineup, and the way people interact
with its lineup."
Chris Charron, an analyst
with Forrester Research,
agrees that Internet
traffic measurement ultimately has
to go beyond the standards
used to measure advertising
in other media.
"Reach matters
to content providers and to advertisers,
but the definition
of reach is bursting at the seams
with the Interent,"
he said. "We need to shift from
measuring pure viewership,
and put more emphasis on
measuring response
and attention and frequency of focus.
That's a better signal
of how compelling content is and
it's also a better
signal for advertisers to judge how
loyal, how active
is the consumer. These are things we
really need to get
at -- not just do they have the
computer turned on."
The fact that Internet
traffic measurement is in the
early stages of development
is probably the one factor
that everyone agrees
on. "We have to start somewhere and work toward the height of accuracy,"
said Mark Bernstein
of CNN Interactive.
"I don't think there's going to be a
simple solution for
the next year or two."
--------------------------------------------------------
Related Sites
Following are links
to the external Web sites mentioned
in this article. These
sites are not part of The New
York Times on the
Web, and The Times has no control over their content or availability. When
you have finished
visiting any of these
sites, you will be able to return
to this page by clicking
on your Web browser's "Back"
button or icon until
this page reappears.
* ABCnews.com
* CNN
* MSNBC
* Relevant Knowledge
* Internet Advertising
Bureau
* Media Metrix
* Forrester Research
--------------------------------------------------------
Lisa Napoli at napoli@nytimes.com
welcomes your comments and suggestions.
--------------------------------------------------------
Copyright 1998 The
New York Times Company |