magazine publisher has two Web sites. They both attract
roughly the same number of visitors, who view roughly
the same number of pages.
But one of them is failing. Can you tell which one?
CondeNet's answer was swoon.com. The relationship hub,
until recently the online content conduit for Conde
Nast titles Mademoiselle, Glamour and GQ, was shut down
despite the fact that both its audience and usage came
close to that of sister site concierge.com, a travel
site that remains a firm fixture in CondeNet's online
plan. In one month, in fact, the "dating, mating
and relating" destination site of Swoon.com received
5.9 million page requests--100,000 more than concierge.com.
Welcome to Web metrics, where a few numbers can mean
many things, and a lot of numbers can add up to very
"There are lots of things that translate into
success and revenue on the Web," says CondeNet
president Sarah Chubb, explaining how a strong audience
demographic enables concierge.com, with 676,000 monthly
unique users, to virtually match ad revenues with CondeNet's
most visited site, epicurious.com, which draws 900,000.
But for swoon.com, the revenue potential simply wasn't
there. "In fact, both of those sites had managed
to attract and retain a really loyal audience,"
Chubb said of swoon.com and phys.com, the health site
that was also shut down. "But when we Looked at
Swoon we saw it didn't belong to a category that had
revenue attached to it, and there was no real e-commerce
Divining a Web site's future from the morass of usage
statistics available may be a bit like reading tea leaves,
but for many sites--especially those lacking the customary
measures of revenue and profit--indicators like page
views and visitors may be the only ones to go on at
"Practically no one is making money, says Rob Garrett,
president of AdMedia Partners. The situation forces investors
to look elsewhere for signs of a site's viability--usually,
Garrett says, at how much its audience is growing and
how long they're staying. But there is some disagreement
even there, along with a general confusion over what success
on the Web means in the first place.
"The question is, 'What is the strategy?'"
says Dana Chinn, director of Internet business projects
at 101communications LLC and a member of American Business
Media's Electronic Media Committee. "If the strategy
is for the Web site to be its own business unit, or
if the strategy of the site is simply to serve the print
subscribers, the metrics of success can be very different."
Likewise, success is measured differently by different
parties. "We have many constituencies," says
Mitchell Prayer, president of Nationalgeographic.com.
"For advertisers there is one set of metrics, e-commerce
there's another set, and for editorial there's another
THE NEW TOOLBOX
The common metrics of Web successor lack thereof--revolve
around measures of unique visitors, the number of different
people that access a Web site over a given period, page
views (sometimes called page impressions) and visits.
Those are the primary metrics audited by ABC Interactive
(ABCI), the Audit Bureau of Circulation's Internet division,
whose reports are utilized by both advertisers and investors.
"A site will want to show itself in the best light,
so they'll pick the combination of statistics and the
presentation that shows them off the best they can,"
says Dick Bennett, ABCI's senior vice president of auditing.
"That can be in conflict with the way the advertiser
or investor wants to evaluate them, and that's why the
ABCI report renders the information in a consistent
But unlike pure-play Web companies, magazine publishers
have been slow to jump on the auditing bandwagon, says
Bennett. One reason is money. Internet auditing, usually
done on a monthly basis, costs well over twice that
of an ABC print audit--a tough bill to swallow when
so many Web sites already cannot cover expenses. And
while ABC endorsement is a must-have in print, advertisers
are not yet demanding audited Web figures.
Yet Web metrics are not always what they seem. While
unique visitors, or unique users, is often broadcast
as a significant indicator of success, the two methods
for calculating that number yield wide discrepancies.
Both--representative audience measurement as used by
Media Metrix and Nielsen/NetRatings, and log-file data
gathered at the site's server--pinpoint the same metric
from opposite directions.
The first does so from a user's perspective, by monitoring
a sample audience and then applying the results to the
entire online universe; the latter does so from the
perspective of the Web site, by tallying actual requests
to the site's server. The differences lie partly in
methodology and partly in each method's shortcomings.
Log-file data, for instance, can be inflated by "spiders"
or "bots," automated programs that access
Web page--for various reasons--or, conversely, deflated
when pages are "cached," that is, stored and
accessed from another Web site. Representative sampling,
on the other hand, accurately measures only the largest,
broadest sites, for which a sufficient sample can be
Illustrating the difference in metrics is CondeNet's
epicurious.com, where log-file activity indicates nearly
twice the unique visitors estimated by Media Metrix,
says Chubb. She attributes the difference to Media Metrix's
relatively small at-work sample, noting that site data
indicates many of epicurious.com's visitors log on during
"In the ad sales world we sort of have to live
with Media Metrix, but when we're looking at the site's
ongoing success or what the business opportunities are,
I look at our numbers," Chubb says.
Nationalgeographic.com also utilizes both log data,
monitored with WebTrends software, and representative
measurement from Nielsen/NetRatings, which Prayer says
allows him to compare his site's performance against
that of other sites.
Niche sites with highly defined audiences, such as
101communications' flagship MCP.com for Microsoft Certified
professionals, don't even show up on representative
radar, says Chinn.
A Web site's relevance to its audience is better measured
in page views than audience figures, says Chinn. The
pure number of unique visitors doesn't necessarily mean
you have a successful Web site, she says, but rather
that you've constructed a useful and easy-to-navigate
site. "If you want to draw a comparison to magazine
circulation," she adds, "page views is probably
Page views can work in the same way as reader-response
mechanisms. At Nationalgeographic.com, high page-view
figures tell Praver that visitors want functional, detailed
features--such as its interactive mapmaking machine--and
are willing to spend time using them. Nationalgeographic.com's
average user views between seven and 13 pages per visit,
Praver says. "That indicates when people come to
the site they're exploring and digging deep. It's not
the type of site where you come, look at the home page
and then leave." Log-file data, Praver adds, provides
instant feedback on how people are using these utilities.
An important metric for advertisers, visits--the number
of times a site is accessed--in combination with visitor
data determine how many users come back, and how often,
says ABCI's Bennett. "Where 'unique users' talks
to reach, 'visits' talks to frequency," he says.
Advertisers, however, also want to know definitively
how many of their messages reach the Web audience, meaning
that ad impressions and click-through rates are important
metrics as well.
But ad impressions, running in the millions per month,
can be mind-boggling to smaller, less sophisticated
advertisers--and a tough sell for publishers. Felix
Macgowan, president of Boulder, Colo.-based Inside Communications,
which publishes Inside Triathlon and cycling magazine
VeloNews, offers as an example a client who buys 150,000
impressions for a given month--a small share in VeloNews.com's
monthly cycle of some 6 million impressions--and then
wonders why he doesn't see his ad when he calls up the
site. The result: while VeloNews retains 90 percent
of its print advertisers from year to year, its Web
site holds on to only 10 to 15 percent. "They just
don't come back," Macgowan says.
So recently, Macgowan has begun pushing sponsorships--where
an advertiser's message is a fixture on the page--but
expects it will be several months before he sees any
Another small publisher, National Trade Publications,
of Latham, N.Y., also wrestles with a disconnect between
advertisers and the impressions concept. "They
[advertisers] just don't get the concept that when they
contract for 1,000 impressions, it may take two weeks
or it may take six months for those impressions to be
delivered," says National Trade president Humphrey