May 26, 1996
Banners: Bullying, Boosting and Blocking
Related Articles
- Hits, Views, Clicks and Visits: Web Advertisers Face Data Jungle
- Who's Doing All This Measuring?
By JAMIE MURPHY and ED FORREST
ncreasing click rates is becoming an abiding goal of advertisers and Web site owners,
especially since Proctor & Gamble recently rewrote the rules for buying advertising banners.
Not content with the cost-per-thousand, or CPM, model, P&G is playing hardball with Web sites,
forcing them to accept a different payment design. For example, it pays search engines through a
scheme that factors in the number of clicks on its banners. The broadcast industry has long used a
similar strategy, typically during early morning hours on low-traffic cable TV stations, where advertisers
pay per call to a 1-800 telephone number.
Yahoo was the first search engine to acquiesce to P&G's terms. Still, Yahoo and other sites
have argued that click rates depend on a well designed banner.
"How do we distinguish between a great site and a lousy ad?" I/Pro's Erin Gaffaney asked.
"Do we blame the ad or blame the site?"
Few sites so far have bowed to the P&G model.
A recent Nielsen I/Pro study found that click rates varied from 2.4 percent to 17.9 percent,
depending on the advertiser's banner. DoubleClick has shown that a few simple banner changes
can increase the click rate by 200 percent. DoubleClick's Test It! program allows an advertiser
to compare two similar banners and in 48 hours see which has a higher click rate.
C/Net claims that an animated Java-enhanced icon increased the click rate nearly four times over
than of a static design for the same ad.
Click rates also depend on the type of Web site hosting the ad banners. The Nielsen I/Pro study
found that the click rate varied by as much as seven times between different sites.
A study by Jumbo!, released to CyberTimes, suggested that click rates fluctuated from 2.2 percent
to nearly 6 percent, depending on the Web site hosting an ad banner. This translates to a cost per
click of 31 cents to 35 cents, far below other rates.
Will Margiloff of Jumbo! argues that the true cost of Web advertising lies somewhere between
the old and new models. "CPM is fine, but it should work in conjunction with cost per click,"
he said, "because both are needed to understand the true value and effectiveness of an ad buy."
If hits, visits, clicks, impressions and registration weren't enough to toss into this salad of Web
ad rates and measures, consider "smart" or "targeted" banners. These are aimed at increasing
click rates and impressions by taking advantage of software that recognizes such user characteristics
as time of day, computer type or domain name -- .edu or .com -- to deliver an ad banner tailored to
a specific user profile.
DoubleClick provides this service on a real-time basis between the user and the Web site.
User information is first sent to DoubleClick, which then transmits the ad banner to the site's page as
it's displayed on the user's computer. NetGravity offers a similar software package that resides on the
Web site's server. Search engine sites like Yahoo! work smart banners based on keywords.
A keyword search for "Florida," for example, would likely return to the user ad banners specifically related
to that state.
And those who mourn the Web's commercialization can take heart in a new category of software known as
ad blockers.
PrivNet's "Internet Fast Forward" is a sort of anti-ad radar that detects and blocks out all ad banners
appearing on a requested Web site. Basically, this means a free lunch on advertising-supported sites.
Ad blocking isn't a big concern yet, said DoubleClick's Kevin O'Connor, who sees less than 5 percent
of users rigging themselves with such anti-ad radar. But he added: "If it were to ever become a big problem,
we would simply defeat it. The consumer choices are 'take ads or pay the site.' There is no such thing as a
free lunch."
Copyright 1996 The New York Times Company