RALEIGH, N.C. -- In the days leading up to the 2008 Democratic presidential primary, most polls showed Barack Obama leading Hillary Clinton by five to 15 points in South Carolina. But two days before voting, a small polling firm in Raleigh, N.C., gave Obama a 20-point lead.
"People were like, 'Who are these people? They're crazy. There's no way he's going to win by that much,'" said Tom Jensen, 32, director of Public Policy Polling. Obama ended up winning by nearly 30 points.
Having been around for just seven years and known only for polling North Carolina races, PPP dipped its toe in the Palmetto State's primary after most pollsters missed Clinton's 2008 New Hampshire victory. Since then, the six-person Democratic firm has grown into one of the most prolific -- and controversial -- national polling companies with an impressively accurate record.
PPP correctly predicted the winner of every state in the 2012 presidential race, as well as the winner of every competitive Senate race it polled that year.
Jensen, a politics junkie and statistics nerd (he's been calculating baseball players' batting averages in his head since he the age of 6) came on board in 2007 and has had a driver's side view of the firm's evolution.
With a desk buried in papers pointing every which way, his office looks more the part of an old-school polling director. But housed in a sleek glass office building, PPP is an unapologetically millennial firm.
Jensen crowdsources ideas on Twitter and poses survey questions no one else will ask. A February poll, for example, found that 38 percent of Florida voters thought Texas Sen. Ted Cruz could be the Zodiac Killer, an elusive California serial killer from the 1960s and 1970s.
That the media sometimes pays more attention to the silly questions doesn't bother Jensen, and the attention is good for the firm. "It's the kind of thing that would have driven me crazy in 2008, when I was 24. And now it's just like, whatever people want to take from our polls, that's fine."
And some of those quirkier questions, he argues, have been instructive. A recent series asked Trump supporters whether they believed the North should have won the Civil War. "People were like, 'Oh PPP is trolling again.' But it really showed that these were the kind of people that Trump was appealing to, and I think it explains a lot of his staying power," said Jensen, who was raised in Ann Arbor, Mich., before going to UNC-Chapel Hill. His slight twang hints at his family's heritage in upstate South Carolina, where his great grandfather was close with former Sen. Strom Thurmond.
Jensen has also conducted horse race polls about the GOP presidential primary, and he doesn't buy the argument that a Trump nominee will bring inevitable success for Democrats in November -- a warning he tries to impart to Democratic audiences.
"I'm like, 'Listen y'all, you really need to be realistic about the fact that it's going to be a close race even if it's Trump. And Trump really might even be able to beat Clinton,' and I just see all these shocked faces across the room," Jensen said.
As much as PPP is waist-deep in the presidential race now, the firm didn't start out that way. Wealthy businessman Dean Debnam founded PPP in 2001, two years after his wife ran for mayor of Raleigh. "She really had sticker shock from the cost of polling," Jensen said. Along came PPP to fill the void for low-budget, local candidates.
Today, the firm is working for political powerhouses -- Correct the Record, a super PAC supporting Hillary Clinton, and Senate Majority PAC -- and lower-tier efforts alike. They've done polls for state legislatures and county commissioners across the country.
With 2,016 phone lines, PPP conducts 30 to 40 polls a week, calling 100,000 to 200,000 people per night. By September and October, PPP will be conducting 30 to 40 surveys a night. Ninety percent of that is private polling; the other 10 percent represents their own public polls.
One series conducted early this month suggested that Republicans' refusal to fill the Supreme Court vacancy could boost Democrats in competitive Senate races, where the majority of voters want to see the seat filled this year. PPP publicizes many of its horse race polls, which consultants tend to take with a grain of salt.
"I'd never replace a traditional pollster for PPP," North Carolina Democratic consultant Morgan Jackson said.
But Jensen doesn't pretend to be a traditional pollster. "We never claim that we can do the sort of 45-minute, 75-question really long, in-depth poll that helps you set your whole campaign strategy," Jensen said.
Instead, he said, PPP offers periodic check-ins for its bigger clients that take the temperature of the race at a certain point in time -- likely for $2,000 to $3,000 rather than the $15,000 a bigger firm would charge. And for smaller clients, Jensen argued, PPP offers the only affordable source of research.
Its specialty is Interactive Voice Response polling, also known as automated telephone polls -- the cheaper and less trusted younger sibling to live-caller surveys. In Democratic circles, PPP gets high marks in the IVR market. "As far as IVRs, I'd use them before I use anybody else," said Jackson.
PPP started including cell phones in its surveys in the beginning of 2014 -- an important methodological development that enhances accuracy. In Michigan, for example, Jensen suggested that many pollsters overstated Clinton's strength because they relied on landlines only, missing the many young Bernie Sanders voters who use cell phones exclusively.
Cell phone polls aren't perfect, though. For starters, plenty of people don't answer their phones if they don't know the number. "I mean, my sort of personal policy is that the only two people who are allowed to call me are my mom and my grandmother. If you're not one of those two people, do it in a text or an email, right?" Jensen joked. "I, even as a pollster, am not really reachable to any pollster who might call me on my cellphone."
Jensen thinks much more of national and state level polling will be done online by 2020. Polling purists still frown on online polling, in part because of its inherent selection bias, but Jensen sees it eventually opening up participation rates because of the convenience.
PPP has, at times, paid a price for doing things its own way. In 2013, FiveThirtyEight's Nate Silver and Nate Cohn, then of the New Republic, went after Jensen for the way PPP weighted its samples. Jensen admits the firm has since changed the way it weights variables, but says the controversy didn't stop them from having their biggest year in 2014.
Being in the polling business, Jensen stressed, is about adaptation -- and PPP's evolution shows that. This cycle, for instance, the large numbers of new voters turning out for Sanders and Trump prompted Jensen to change the way he screens participants.
"I really think a big part of what's helped us to be successful is we sort of just do things our own way and don't worry about what everyone else is doing," Jensen said.