The debate over which site had the most November page views reflects
the difficulty of tallying Web traffic, and billions of ad dollars are
at stake.  On the Web, the competition for most popular site can be as intense as
the race for the pennant or even the Presidency. So news that
up-and-comer MySpace.com beat incumbent Yahoo! (YHOO) for most monthly page views for the first time in November understandably grabbed headlines.

According to comScore Networks, News Corp.'s (NWS)
MySpace racked up 38.7 billion page views in November, compared with
38.1 billion for longtime leader Yahoo. But the dispatches concerning
the upset were the online equivalent of the famous headline "Dewey
Beats Truman."

Almost immediately, the results were called into question. UBS (UBS)
analyst Benjamin Schachter left voicemail messages for investors and
reporters warning against making decisions on comScore's potentially
unreliable data. Yahoo was still the undisputed leader, measured by
other key metrics, including unique visitors and time spent on the
site. Besides, Nielsen//NetRatings (NTRT),
comScore's main competitor, still had Yahoo leading page views in
November: 33.4 billion, vs. 29 billion for MySpace and the rest of News
Corp.'s Fox Interactive Media properties.

Focusing on Ad Impressions

The discrepancy has revived complaints about the accuracy of
reporting agencies' results, which often differ from companies' own
audience measurements (see BusinessWeek.com, 10/23/06, "Web Numbers: What's Real").
It also underscores the rivalry between comScore and
Nielsen//NetRatings for recognition as the most trusted source for
Web-traffic data. The winner, if one emerges, may set the standard for
how site popularity is measured, influencing how marketers dole out
billions in online ad dollars each year. Recognizing the high stakes in
that tussle, comScore and Nielsen//NetRatings both are refining their
tactics.

For starters, in the first quarter of 2007, comScore plans to change
how it determines the amount of advertising a Web company shows its
audience. To do that, comScore will focus on ad impressions, or the
number of times an ad shows up on a page, rather than the number of
times a Web page is viewed. The results can vary, depending on how a
Web page is designed.

For example, Yahoo and Google (GOOG)
are among companies that use a technology called Ajax that can change
an ad, influencing the number of ad impressions, even if the user
doesn't refresh or click to a different page. Yahoo attributed the 9%
drop in its page views, which allowed MySpace to overtake it, to its
inclusion of Ajax. Thus, page views—long used by industry analysts to
estimate how many ads a company can serve and, thus, potential
revenue—is no longer a reliable measurement.

Time Spent on a Site

ComScore is also planning to launch a set of entirely new ad metrics
in the second quarter of 2007, according to comScore Chief Executive
Officer Magid Abraham. "We are developing some proprietary metrics that
are a much better replacement for page views and are actually a better
measure of engagement," says Abraham.

Abraham is tight-lipped about specifics, but he says the new
techniques will better reflect the influence of new technologies. The
company will focus more on time spent on a site than page views. It's
worth noting that, by that estimate, Yahoo is the clear winner. Its
users spent 42.7 billion minutes on its pages in November. MySpace's
users spent a total of 13.8 billion minutes, says Abraham. 

Nielsen//NetRatings is also working on its measurement techniques in
response to new technologies. The firm developed a proposal several
months ago on how to track Ajax and improve tracking on other
technologies such as streaming media, says Manish Bhatia, NetRatings'
vice-president of global operations and U.S. sales.

Getting Certified

Changing the metrics alone, however, isn't enough to satisfy
everyone. The Interactive Advertising Bureau, an organization of
advertisers and major Web companies, is calling for an external audit
of both comScore and NetRatings by the Media Ratings Council, the body
that certifies measurement practices of media ratings firms such as
radio's Arbitron (ARB).
"At this point, the only thing we know is that the results are
consistently different. What we don't know is why," says Sheryl
Draizen, IAB's senior vice-president and general manager. "The
differences between the numbers reported by comScore and NetRatings are
obviously a big issue for the industry as a whole."

Both comScore and NetRatings say they have begun working with the
MRC to get certified. But Draizen says the process is not happening
fast enough. "It should be a major priority for them," she says. "It is
a real issue to have such different numbers in the marketplace."

For Internet companies and advertisers, determining whose numbers
are correct is more important than just crowning the rightful platform
king. At stake are the tens of billions of advertising dollars destined
for the Web. Internet advertising is estimated to reach $16.4 billion
this year, according to research firm eMarketer. The number is expected
to grow to $25.2 billion in 2010. Which companies get that money,
particularly dollars earmarked for promoting brands and not just
directly selling merchandise, will depend largely on who can verify
their ad inventory and the size, composition, and engagement of their
audience.

May the Best Metric Win

ComScore and NetRatings each believes its own methods the best.
ComScore insists that its method of allowing new measurement methods
will establish it as the clear innovator. It also boasts that it is
able to better measure college students because it offers its software
for download over the Internet in exchange for games, screensavers,
security software and other items. "We feel we have the advantage,"
says Abraham.

Nielsen has an edge with its study panel, which was selected via
random telephone polls and thus, according to Bhatia, more
representative of the general population. He argues that only a
particular kind of computer user is comfortable downloading free
software over the Internet. "Fundamentally, the quality of the result
comes from the quality of the underlying panel," says Bhatia.

Who will win? That will depend largely on who is able to respond
fastest to the new technologies that impair the relevance of old
metrics, says Peter Daboll, a former president of comScore who now
heads Yahoo's global market research department. "Internal server log
data, panels—all have their limitations," says Daboll. "I think frankly
we just need more innovation."