Target: Madison Ave.

She's been near the heart of the Web search business since before that business could plausibly be described as such. From 1996 to 2002, Ellen Siminoff was one of Yahoo's top executives--the woman behind the company's early investment in Google , as well as its relationship with paid-search pioneer Overture (which Yahoo eventually acquired). Today Siminoff no longer dwells inside the belly of the beast, but she's just as big a player in search as she was at the height of the bubble. You could even say she's emerged as the industry's undisputed Queen of Keywords.

Siminoff is president and CEO of Efficient Frontier, a three-year-old Silicon Valley startup that's leading the field of search-engine marketing. Though Efficient Frontier is replete with heavy-duty geeks (many of them holding doctorates in math or computer science) and serious technology, its impact is likely to be felt more profoundly in the media industry than in tech--for what Siminoff is building is nothing less than a new kind of advertising agency. Indeed, if Google is the CBS of the Internet era, Efficient Frontier may turn out to be its Ogilvy & Mather.

Not that Siminoff had anything so grandiose in mind when she first came into contact with Efficient Frontier in 2003. "I'd always been a believer in search," she says one afternoon at the company's Mountain View offices. "But one of my biggest frustrations at Yahoo was that I was running all these businesses, and I had absolutely no idea how to execute a marketing plan for them. There weren't the tools or the technologies available to optimize Web advertising."

So Siminoff was intrigued when she met Efficient Frontier's founder, Anil Kamath. A Stanford computer science Ph.D. and the founder of eBoodle, a comparison-shopping startup ultimately acquired by BizRate, Kamath was creating a system to help advertisers more effectively place their text ads on Google and Yahoo, both of which conduct online auctions of search keywords that determine where the ads appear. Employing vast computing power and fiendishly intricate algorithms, Kamath's system would indicate which keywords to bid on and the optimal amount to bid--based on countless variables, from past click-through performance to the number of rival bidders, while taking account of constant fluctuations in auction prices.

All this should sound familiar to anyone acquainted with the incursion of computer-driven quantitative methods that transformed high finance in the 1990s. As Siminoff puts it, "Anil's approach was based on the same sort of math they use on Wall Street." (Before starting eBoodle, Kamath was a vice president at D.E. Shaw, a hedge fund that specializes in applying quant-jock models to program trading.)

"Google and Yahoo are like exchanges, so the analysis is essentially similar," Siminoff says. "You're studying all of the potential scenarios and choosing the ones likely to yield the best ROI in aggregate. Then you just keep iterating and refining the analysis."

Duly impressed by what Kamath had devised, Siminoff invested an amount she'll only say was "less than $1 million" in 2003, and the next year she officially came onboard as Efficient Frontier's chief executive. Today the company boasts 70 major clients doling out more than $100,000 each month on keywords, including Edmunds.com, Knight Ridder , and LendingTree, which pay 5 to 15 percent commissions for its services.

With $150 million in annual ad spending under its control, Efficient Frontier is one of the biggest buyers of sponsored links from both Google and Yahoo, and it manages a database containing roughly 100 million keywords. "You need complex strings, synonyms, antonyms, plurals, and misspellings," Siminoff explains. "We're way beyond the old days of simply buying banner ads."

The increase in complexity is one reason Efficient Frontier is in such great demand. Another is that, with Microsoft and IAC/InterActiveCorp's Ask.com already building their own biddable ad marketplaces, search seems destined to grow only more powerful and pervasive as a platform for advertising. "It's the most targeted ad medium conceivable," Siminoff says. Hence a third reason for Efficient Frontier's takeoff: The payoff it typically delivers is irrefutable. For example, tableware retailer Replacements Ltd. saw its search-driven sales rise 30 percent in just six months. Meanwhile, at BabyCenter, site registrations tripled in the course of one year.

Even at this early stage, the success of Efficient Frontier has already left a mark on Silicon Valley. Most obvious is the sprouting of hundreds of search-engine marketing startups--though few are as sophisticated or scientific as Efficient Frontier. Subtler is the change being wrought on the very character of tech marketing. "Traditional brand-marketing people are no longer as valuable," says Excite co-founder and JotSpot CEO Joe Kraus. "What you're looking for now are people comfortable with numbers--what you're looking for are finance people."

Placed in historical context, these developments seem less surprising than inevitable. For 150 years, the rise of each successive new medium--newspapers, magazines, TV, direct mail--has spurred the invention of new forms of advertising. And each new form of advertising has birthed new forms of ad agencies, from print-space brokers such as Volney Palmer in the mid-1800s (the age of newspapers) to visually oriented shops such as Ogilvy & Mather in the mid-1900s (the age of television) to integrated global-marketing conglomerates such as WPP at the end of the 20th century (the age of satellite communications and borderless commerce).

Is Efficient Frontier really the next descendant in this lineage? At a glance, it might not seem so, but I think it is--and Siminoff doesn't disagree. "Technology and analytics are very important to what we do," she says, "but at the end of the day, we operate on a media model."

I mention to her that ad agencies perform three functions: account planning, media buying, and creative. "We certainly do the first two of those, but as of now, there's nothing very creative or sexy about writing a few lines of search-ad copy," she replies. "Now, will that change? Yes, over time. You'll be able to attach a video, say, to the keyword 'Honda Accord.' Search right now is successful more on the direct-marketing paradigm. The question is, will it evolve into branding? And if it does, will that lead to a better ROI?"

Siminoff hesitates to venture an answer to those questions, and so do I. What she does expect, however, is that search is only the first advertising platform to feature a biddable marketplace. "In the future, we expect to evolve our technology and apply it to all digital media," she says. "Eventually banners, shopping services, maybe even TV will be biddable. As Google extends into other media, that will be an opportunity for us to help companies with that."

What Siminoff is talking about is a radical reshaping of how advertising is bought and sold--the triumph of the auction model. Naturally, there are reasons to be skeptical of such a vision. But Wall Street was pretty skeptical when the quants first came to town. If I were an adman, I'd look at Ellen Siminoff--and what she represents--and be very, very afraid.

John Heilemann wrote "Pride Before the Fall." His next book is "The Valley." He lives in Brooklyn.