PHOTO: Andrei Tchernov/iStockphoto; logo: Google
|
31 October 2007—The mainstay of the US $14 billion
online-advertising industry—the “sponsored links” that
appear alongside Google
and Yahoo
search results—contains an inherent inefficiency that is
destined to give birth to third-party search-term
auctioneers and fuel the growth of so-called mediators,
like the Web sites Shopzilla
and Shopping.com.
This unorthodox conclusion emerges from a new study
by engineers at the University of California, Los
Angeles, in which a rigorous subset of mathematical
laws—first formulated by John Nash (the subject of the
2001 film A Beautiful Mind)—was applied to the volatile
pay-per-click advertising marketplace. Specifically,
Sudhir Kumar Singh and his coauthors at UCLA’s
department of electrical engineering employed the
game-theory concept of Nash equilibrium, in which one’s
strategy continually shifts
because one’s opponent’s strategy is continually
shifting.
Typically, searches in Google or Yahoo produce a
string of ads (sponsored links) along the side of the
screen and across the top. Those links are generated by
advertisers placing bids—sometimes $2 or more per
click—on phrases like “digital camera” or “hybrid car.”
Economists studying auction behavior, says Vwani
Roychowdhury, Singh’s advisor and study coauthor, have
used the Nash equilibrium concept to determine auctions’
outcomes. The group’s work, he says, “shows in a formal
way why mediators are necessary if this way of
advertising survives.”
The authors point out that most searches are not as
generic as the so-called “AdWords” that companies buy.
Though a company might purchase the phrase "digital
camera" a potential customer might search for “digital
camera Bluetooth accessories in St. Louis,” for example.
Often as a result, the sponsored links miss the point
altogether, and direct this user with a specific request
toward catchall sites, such as national online
electronics retailers, that are of no use to the person.
Roychowdhury uses the example of a specialized market,
such as cellphone headsets for the hearing impaired. A
hypothetical company that sells these might be able to
afford to appear in Google’s sponsored links section for
queries on “hearing impaired.” But, he says, very few of
the people searching on a term that generic would be in
the market for a gadget that specific.
Yet, if the same company put in a bid to appear in the
sponsored links section for any searches involving the
word “cellphone,” they’d be up against the likes of
Nokia, Amazon, and Verizon. They couldn’t afford to
compete with the ad budgets of such big companies, and
they shouldn’t have to, because the hypothetical headset
company is not in direct competition with these retail
giants.
Instead, Roychowdhury says, a third-party company,
called a lead generator or “leadgen,” ideally would
exist to aggregate, say, AdWords specifically for
cellphones and cellphone accessories. The
hearing-impaired headset company would then pay the
leadgen company. The leadgen company, in turn, would
place its bids on all conceivable forms of cellphones
and cell accessory–related queries with Google and
Yahoo.
“This leadgen company could take the traffic when you
search for ‘headset,’ ” Roychowdhury says. “And on their
page, they’d feature [headsets] for hearing-impaired
people.”
Paul Ryan, CEO of the home-improvement referral site
Done
Right! (and a consultant on a business venture
for Roychowdhury) says the UCLA group’s approach
rigorously shows that well-run third-party AdWord
auctioneers will always boost the bottom line for both
the search-engine company and for advertisers.
The study shows that Google AdWord auctions are only a
great deal for a certain limited set of advertisers,
says Ryan. “This is what the paper’s all about,” he
says. “It’s inevitable that the auction-based model that
AdWords and Yahoo Search Marketing uses will cause the
rise of mediators.”
At least some in the Internet business have already
intuited the importance of mediators. In 2005, to the
surprise of many analysts, eBay
bought Shopping.com for $634 million.
Roychowdhury’s research answers why.