Interesting article in the NY Times [nytimes.com] on online
advertising, data mining, etc.:
ON a recent Thursday, Darren Herman, the president of Varick
Media Management, was sequestered in his SoHo office. He wasn't
scrutinizing a television ad or images from a photo shoot. He
was combing through graphs and Excel spreadsheets.
Mr. Herman had run 27 ads on the Web for his client Vespa, the
scooter company. Some were rectangular, some square. And the
text varied: One tagline said, "Smart looks. Smarter purchase,"
and displayed a $0 down, 0 percent interest offer. Another read,
"Pure fun. And function," and promoted a free T-shirt.
Vespa's goal was to find out whether a financial offer would
attract customers, and Mr. Herman's data concluded that it did.
The $0 down offer attracted 71 percent more responses from one
group of Web surfers than the average of all the Vespa ads,
while the T-shirt offer drew 29 percent fewer. And Mr. Herman
didn't just compare the messages in the ads he also looked at
the sites where they ran, when they ran and what groups of
From the "Mad Men" era until now, advertising has been about a
catchy tagline, an arresting image, the Big Idea. But Mr. Herman
and his competitors are bringing some Wall Street-like analysis
to Madison Avenue, exploiting the huge amounts of data produced
by the Internet to adjust strategy almost instantly.
"It's putting numbers to an industry that never had numbers
before," says Mr. Herman, 27, who started and sold three media
and technology companies before founding Varick last summer.
"It's nice to be able to tell your brand manager or the chief
marketing officer which audience is interacting with the unit,
what time of day, what day of the week, and what the response is
on certain types of offers. Before, nobody could really tell you
This approach turns marketing "upside down," says Ron Proleika,
the vice president of marketing communications at Windstream
Communications, an Internet service provider and a client of Mr.
Herman's. "It forces marketers to stay on their toes and think
of thousands of small great ideas instead of one great big one."
Major advertising holding companies like WPP, the Publicis
Groupe, Havas, MDC Partners and the Interpublic Group are
starting data practices, hoping to latch onto what is expected
to be the fastest-growing category of online advertising in the
next five years.
Where the data guys were once an afterthought in a marketing
presentation, now they are at the core of the online strategy.
What's more, they can help advertisers save money in traditional
media by testing different phrases or images online to see what
works before producing an expensive television commercial or
magazine ad. Who attracts more clicks in a grape juice ad, for
example the blond girl or the brown-haired boy?
The shift to data-based campaigns is forcing marketers to learn
new skills and drawing a new breed of worker to Madison Avenue.
While most data executives now in the field came from media
backgrounds, they are recruiting Wall Street math geniuses
because the job requires hourly adjustments in strategy based on
Mr. Herman is trying to hire people from Citigroup and Bank of
America, and he hopes that the layoffs in the financial industry
will help him do it on the cheap.
"It mirrors the financial markets in many ways," he says, so
"that's where we go."
Still, getting advertising agency employees to rely on data is
difficult, agencies say. And as people trained on Wall Street
migrate to Madison Avenue, executives anticipate battles between
creative types and wonks.
Traditional ad agencies still don't have budgets that allow for
a lot of digital experimentation, Mr. Herman says. He notes that
most traditional agencies "make the bulk of their money in
print, radio and television."
So even as this area becomes increasingly technology-driven, old
ways of doing business and clients reluctant to embrace
radically new approaches mean that the advertising culture won't
"At the end of the day," Mr. Herman says, "the entire process
isn't digital because our clients aren't."
UNTIL the Internet, advertising required heavy research at the
front and back ends. Millions of dollars went into television
and print ads, so the advertisers had to get the idea right
before they produced one. Determining the effectiveness of those
ads was hard. It required follow-up surveys and interviews. And
once advertisers began a campaign, they were locked into it they
usually booked TV spots four months before the season began, for
instance, and even if a show tanked, they couldn't always abort
"In the olden days, the consequences of planning were great, so
we'd spend nine months before air date" doing research, says
Barry Lowenthal, the president of the Media Kitchen, a media
planning and buying company that, like Varick, is a unit of MDC
Partners. "Then, nine months after we'd been running the ad,
we'd finally figure out whether it was working or not."
Online, though, advertisers get instant measurements and can
make instant changes to a media plan.
Varick and its handful of competitors cement their strategies
around a system called exchanges, a mechanism that helps online
publishers like NBC.com or Yahoo.com sell ad space. While
publishers have some ad space no company would bid on in advance
few advertisers would book a random Yahoo mail page, for
instance publishers still want to show an ad when someone loads
that page. So the publishers let an ad exchange like Right
Media, from Yahoo, or DoubleClick Advertising Exchange, from
Google, sell that space instantly, through an electronic
auction, and get a cut of sales.
Such random, seemingly unwanted space could be virtually
worthless. But because ad agencies can now use multiple sources
to gather very specific demographic data about visitors, such
space gains value and can be brokered on an exchange.
Among the sources agencies rely on for data-mining is
gathered from other sites. Imagine that every time
someone entered a store while shopping, she received a stamp on
her hand. By the time she got to Macy's, the clerk could see she
had visited Williams-Sonoma and Home Depot and could direct her
to housewares. A similar principle is followed online.
When someone visits a site like Expedia or Autobytel.com, that
site captures valuable Someone is a first-class
traveler, for instance, or shopping for a hybrid car. Those
sites have deals with data companies, like, to place a so-called
cookie a small text file on that visitor's hard drive,
indicating those preferences. An advertiser like Varick bids on
those cookies, instructing an exchange that it will pay a
certain amount for an ad when a certain cookie is for sale.
Other companies, like Media6Degrees and 33Across, analyze the
world of social media, using cookies and interaction data to
find "lookalike" groups among friends on Facebook, Flickr or
other social sites. Their theory is that friends share values
and are likely to respond to similar marketing messages.
Finally, companies can add cookies for anyone who visits pages
on their sites if someone gets to the checkout page, then
abandons his shopping cart, the company will probably pay lots
of money to advertise to him again.
This combination real-time data and ad exchanges has monetized
what was once considered throwaway space online. ThinkEquity, a
research firm, estimated that advertising based around Web
publishers' extra space brought in $4.1 billion in 2008, up 32
percent from 2007, and it expects it to be the highest-growth
segment of the online advertising market between now and 2013,
outpacing even search. (It is still a small part, however, of
what ZenithOptimedia, a media agency, estimates to be the
overall $487 billion advertising market.)
All this tracking has raised privacy concerns. Some privacy
advocates have asked Congress and the Federal Trade Commission
to investigate the issue, seeking clear policies about sensitive
data, more on the way companies are tracking
consumers and options for consumers to avoid online tracking.
So far, the commission has recommended that the industry police
itself. But Jon Leibowitz, one of the commissioners, warned in
February that the industry needed to do a better job or face new
laws and regulations.
Without much regulation, says Michael Brunick, vice president
and media technology director for Cadreon, a competitor of
Varick's, "the data game right now is a little bit of the Wild
WITH so much to trade on, several advertising firms
are creating their own data-based practices.
"We have, over the last year or so, gotten more and more
interested in the ways that you can use data to make advertising
more effective online," says Matt Greitzer, the vice president
of search marketing and auction-based media at Razorfish, which
is building its exchange group.
In addition to what an ad should say, and where and when it
should run, advertisers have to figure out how much each ad, or
"impression," is worth. The data helps them do that. "You're
making, in some cases, real-time decisions about how much to pay
for a specific impression," Mr. Greitzer says.
In a simple example, if an advertiser knew that his ads
attracted more clicks on profiles.yahoo.com than on
movies.yahoo.com, he would pay more when space was auctioned for
the first site.
Edward Montes, the managing director for North America at Havas
Digital, who oversees its exchange group, says that his data
analysts are "basically looking for anything that affects
performance any time they find variance in the matter of how the
media performs, that's what they go in and exploit, and that's
what the exchanges are perfectly set up to do."
As data executives continue to build on their research, this
arena could resemble Wall Street even moRe: yield managers could
hedge their purchases, buy futures to lock in prices and use
other trading strategies. And this type of sophisticated testing
and trading will require changes in clients' attitudes.
Mediabrands, a unit of the Interpublic Group, has been quietly
running a data practice called Cadreon for nine months that it
soon plans to roll out more publicly. In addition to buying
standard Web site ad space, Cadreon also buys mobile
advertising, online video slots and, soon, spots on digital
billboards and other new media.
Traditionally, marketers allocate certain amounts of money for
each medium. Quentin George, the interim chief executive of
Cadreon and the chief digital officer of Mediabrands, says
Cadreon instead would base its strategy on the audience, not the
For a campaign it's now running for a technology client, Cadreon
bought data on visitors to Web sites of the client's
competitors. It divided them into groups that its client already
used to segment existing customers offline like new parents,
gamers or designers.
By examining clicks and other data, Cadreon determined the
demographic profile of groups that were most interested in the
ads. In this particular campaign, new parents responded at high
rates so Cadreon emphasized pitching ads online to that group.
As more ads are bought and sold through exchanges, it could
transform the ad marketplace. "It is foreseeable that you can go
into the system, select an audience and not know whether you are
ultimately buying" a cellphone ad or a video ad on a Web site
like Hulu, Mr. George says. "That's a very, very big change."