CUPERTINO -- Ro Khanna has knocked on 2,200 doors in his quest to unseat veteran Democratic Congressman Mike Honda and there's a good chance Khanna's campaign knows more about the voters who answer than they know about the¿ candidate.
The former Obama administration official is bringing one of the president's most successful Silicon Valley-style campaign strategies home in what could be the first House campaign in the country using "big data" to target exactly whose doors are worth knocking on.
The results of the campaign's confidential analysis is on Khanna's smartphone, which recently lead him to the Cupertino doorstep of 76-year-old retiree Lee Crothers in Cupertino. After a brief handshake and some small talk, Khanna, 37, turned away and tapped on his phone, recording Crothers' reaction for instant upload to his campaign headquarters.
Call it what you want: "big data" analysis, "predictive modeling," "nanotargeting" or -- as some critics dub it -- "stalking." Whatever the name, it means gobbling up all the digital bread crumbs voters leave behind -- not just the party registrations and voting histories of old, but things like the Web pages they browse, the things they buy, the cars they drive and all they share about themselves on social media. The goal: to predict which voters are most likely to be swayed.
In an interview later, Crothers, a Democrat, joked that he hopes campaign operatives aren't digging through his garbage cans -- though that's practically what they're doing in the digital world.
"I was glad that he came out to see me -- I liked the cut of his jib when he did that," Crothers said, adding he's not put off by having his data mined for political ends.
"I'm not creeped out by it," Crothers said.
Michael Cornfield, a campaign-strategy expert at George Washington University, said the 72-year-old Honda's battle to be re-elected in San Jose's 17th Congressional District "may be a vanguard race" because it will test whether the tools used by President Obama to technologically pummel GOP challenger Mitt Romney in 2012 can be used to the same advantage in races involving far fewer voters.
"There are questions about how well all of this statistical modeling works on a local level," Cornfield said, "but if it's going to scale down anywhere, it's going to scale down in Silicon Valley."
Valley voters are among the nation's most tech-savvy, engaging in online activity more than most. And the more voters do online, the more campaigns can learn about them.
Jeremy Bird, Obama's national field director in 2012 and now Khanna's general campaign consultant, also hopes to prove that neighbor-to-neighbor field tactics that helped Obama win eight of nine battleground states will work on a local level.
Obama's campaign coupled its data analysis with a decentralized field campaign in which neighborhood team leaders recruited local volunteers who knew what issues resounded with local voters -- and sometimes even spoke their native languages. That was "the genius of Barack Obama," giving volunteers personal ownership of the campaign, Khanna said.
"You can't win a campaign on Facebook. You can't win a campaign on Twitter," Khanna said. "At the end of the day, the best way to get a vote is to shake someone's hand and have a conversation."
Though it surely helps to know exactly which hands to shake -- a targeting mission made more crucial by the entry of three Republican candidates into the race, diluting the "anybody-but-Honda" vote.
Honda campaign manager Doug Greven said his canvassing will start "very soon" and will also be based on data analytics, though he wouldn't give details.
"There's no serious campaign that is just going to talk to every voter with the same message," he said.
Mark Beatty, Obama's deputy battleground-states director in 2012 who now works with Bird at their Washington-and-Chicago-based 270 Strategies consulting firm, wouldn't specify what data they're mining for Khanna to put the pins on his smartphone app.
"I don't want to go too deep into our campaign strategy, " he said.
But data-crunching in 2012 found that high-voter-turnout Democrats are likelier to drive Subarus or Saabs, drink sauvignon blanc or gin, and eat at Red Lobster or Boston Market, while high-turnout Republicans are likelier to favor Volvos or Chryslers, cabernet sauvignon or Scotch, Outback Steakhouse or the Olive Garden.
Khanna doesn't know stuff like that, or even more personal stuff like Web browsing habits, when he meets the voters this data has predicted are swingable. All he has on his smartphone app is their name, age, gender, party, address and phone number.
Khanna's big bankroll -- his campaign had three times as much money as Honda's as of Dec. 31 -- will help pay for the highly trained personnel needed to analyze such data and turn it into campaign strategy, said Andrew Rasiej, whose Personal Democracy Media tracks the intersection of technology and politics.
"Technology is just a tool and there are many factors that affect a candidate's electability. Donald Trump could have all the data analytics he wants; he's not going to win an election," Rasiej said. "But all things being equal, the candidate with a smart data-analytics strategy and technologists working in his campaign will have an advantage."
Obama's campaign crunched data from tens of millions of battleground-state voters while the 17th District has only about 293,000, with still fewer considered likely voters.
"The larger the population you're looking at, the easier it is to do these techniques well," said David Nickerson, "director of experiments" in the analytics department of Obama's 2012 campaign and now a University of Notre Dame political science professor. "It works just like a survey -- if you survey 50 people, it's not as precise as if you survey 500."
Yet with "more knowledge out there and more data floating around" than in 2012, Nickerson believes many more House races will adopt these methods.
That bothers artist Kathleen Mitchell, 65, one of Crothers' Cupertino neighbors. The Democrat wasn't home when Khanna knocked, so he left a note on her door and his volunteers came another day. Until a reporter contacted her, she didn't realize she was among the campaign's data-targeted voters.
While she's "not going to lose sleep over it," it's food for thought, she said.
"There's this whole privacy-intrusion that's going on many levels, not just this one, and that concerns me," Mitchell said, noting she was among millions of victims in last year's credit-card data breach at Target stores.
"But on the other hand, I'm sure this kind of thing has been going on for a long time -- it just seems it's better organized now."