FORTUNE — Kamakshi Sivaramakrishnan never planned to go into advertising, much less run a startup. But when the 37-year-old Stanford graduate, with a Ph.D. in Information Theory, met AdMob founder Omar Hamoui, she turned her back on Wall Street and joined AdMob as a research scientist in 2007. There, she developed large, sweeping strings of math to help improve things like making ads more relevant to users.
“It was the era of feature phones,” Sivarakmakrishnan remembers. Though the iPhone launched that year, Sivaramakrishnan still owned an old-school cell with a small screen and skeleton web browser. “Who browsed the Internet on their phones then? We just used them to talk. It was a voice communication device — not a data communication device.”
Two years later, Google (GOOG), itself the long-reigning Internet ad king, scooped up AdMob for $750 million, one of the largest acquisitions the company ever made. Sivrakmakrishnan stayed with Google for a brief six-month stint before striking out on her own to finish what AdMob started.
MORE: 5 worst Internet acquisitions of all time
“If there is an unfinished legacy of AdMob that needs to be solved, it’s mobile advertising in this next generation,” she explains. Now the 36-strong San Mateo-based startup, with $20.5 million in funding from backers like Kleiner Perkins and Sequoia Capital, argues that today’s $7.29 billion U.S. mobile ad industry is broken. Many marketers continue to treat mobile and desktop behavior as separate sets of data, a flawed philosophy given how many consumers own multiple devices now. They may start surfing the web on the desktop, for instance, but continue doing so on their tablet or smartphone.
Last fall, Drawbridge debuted ad products that know better. Say a smartphone user searches for flat screen TVs, and they see a Best Buy (BBY) sales ad. They click on it, but realize they’d rather browse and buy on BestBuy.com from their desktop instead of scrolling on their phone’s comparatively smaller screen. “Best Buy doesn’t know that was the same person who clicked on their mobile app, and Google doesn’t either,” explains Eric Rosenblum, another Google alum and Drawbridge’s VP of product. All Best Buy knows is that it paid Google for an ad someone clicked on, but they did not buy a television, and that later, a different person came to BestBuy.com and bought a TV.”
Devices that all belong to the same person normally send off similar ad requests or requests made through the device by way of a digital exchange or marketplace. Drawbridge is able to sit on the sidelines of this digital exchange and observe these requests. Over several weeks, Drawbridge’s tech recognizes a user’s behavioral patterns across different devices and with 70%-plus accuracy, matches them up. In the case of Best Buy and Google, the value of Drawbridge is obvious: BestBuy (BBY) would know its ad directly contributed to a purchase, and Google could potentially charge more for that ad.
MORE: Watson will be taking your calls now
The startup has paired over 471 million devices so far, yielding complete desktop and mobile user profiles that number in the hundreds of millions. Travel booking site Expedia (EXPE) is already testing Drawbridge’s ad solutions. Drawbridge’s next step? Drilling deeper into user data and being able to tell marketers things like how often someone comes back to a web site and once they return, whether they’re viewing or clicking from mobile or desktop. Already, Sivaramakrishnan estimates their current ad solutions can offer marketing campaigns as much as 300% in boost.
Given its founder’s dedication, the Drawbridge crew should get there quickly. When she first started, the committed Sivaramakrishnan encouraged people to come into the office on weekends and even wondered whether it was OK to give employees Christmas Day off. Indeed, it was only after her beat-up 1990 Toyota Corolla got totaled in an accident did Sivaramakrishnan finally get a new car. “She was so busy she couldn’t buy one,” chuckles Matt Murphy, a Kleiner Perkins partner and Drawbridge investor, who offered to purchase one for her. (He ultimately didn’t.)
To Murphy however, that would have been a small price to pay so Sivaramakrishnan could continue her work. Explains Murphy: “She’s going after a big problem that frankly, is going to be extraordinarily hard for others replicate.”