Michael Ramlet couldn’t help but notice that the polling profession “still looked like the 1980s” when he was working at Purple Strategies, a communications firm in Washington, DC.
Ramlet wanted to take advantage of how little the field had been penetrated by more advanced technology. Along with a couple of co-founders, he took in $30,000 in angel investments to start a polling venture.
“None of us quit our day jobs,” he told Business Insider. “None of us were going to quit until we knew this could actually be a thing.”
Flash forward three years from Morning Consult’s meager start in 2013, and the polling outlet is partnering with publications such as Politico, Vox, and Bloomberg to deliver some of the fastest, most thorough polls of the 2016 cycle.
Ramlet, Morning Consult’s CEO, spoke of how the outlet was the first to show Sen. Bernie Sanders of Vermont above 30% in the New Hampshire primary. He also noted how their poll displayed Republican candidate Donald Trump’s support as solid enough to withstand his at-the-time seemingly campaign-crushing remarks about Sen. John McCain’s military service.
Morning Consult’s initial polls following each of the presidential debates, the leaked “Access Hollywood” tape, and FBI Director James Comey’s bombshell letter to congressional leaders came well before their competitors were able to go live.
“I’d be kidding you if I said we’d be where we are now three years ago,” Ramlet told Business Insider, adding that the company went from that initial $30,000 to now projecting more than $7 million in bookings this year.
The outlet more or less was born out of a healthcare email newsletter Ramlet, now 29, had started when he was in college at the University of Minnesota.
Morning Consult officially launched with a 2,000-person poll looking at whether young and uninsured Americans were going to sign up for the Affordable Care Act’s insurance exchanges right before they went live in 2013. The White House used that poll in its briefing later that day.
“One of the biggest problems in old-school polling is that they have really small sample sizes,” Ramlet said. “Because they have to pay people to sit in chairs and make phone calls, and that was easy to do in the 1950s when everyone answered their phone calls on their home telephones, but harder to do in 2013 where response rates were really low. So we said no one has any idea if the young and uninsured want to buy health insurance.”
After appearing in the White House briefing, Ramlet said he was contacted by a Fortune 500 healthcare company within two weeks to buy their polling technology. A week later, a DC media company made an offer to purchase the entire outlet.
“So we were like, ‘Oh, s—, we should probably quit our day jobs now and make this a thing,'” he said.
The company would soon start to grow. Initially operating out of a row house near Capitol Hill, the firm grew from 13 employees to roughly 40 and a larger office space.
The row house “might have been the most depressing office space we’ve ever seen,” Meghan McCarthy, Morning Consult’s chief content officer, told Business Insider. She added that the goal of the company when founded was to provide issue-based polling that the cofounders felt was lacking in the polling industry.
But by the time Morning Consult moved into its new office, the “horse-race polling” for the 2016 cycle “really took off,” McCarthy added.
“Because we were in the field every single week, we were one of the first places to see Bernie Sanders was picking up steam in New Hampshire, when he wasn’t really viewed as a serious candidate against Hillary Clinton,” she said. “And still early in the Republican field, we saw Donald Trump consistently polling numbers in the 30s, and that wasn’t something that polling places were out there with yet. Realised we were the only place doing this and we were doing it very well.”
On to 2016
Morning Consult’s initial polling on Trump and Sanders helped to catapult the group into the national discussion in the 2016 polling realm.
“I think my cofounder lost six pounds within 48 hours after we released that [Sanders] poll,” Ramlet said, noting the nerves that came with having Sanders above 30% when the prior New Hampshire polling had him garnering less than 10% of the vote.
During the primary season, a major story in the polls was how Trump was performing better in online surveys than in traditional polls, as shown in The Washington Post. But as the primary season wound down, it was clear Trump’s assessment in the online polls was not an inaccurate projection of where the race would end up, as he easily eclipsed the needed delegates to clinch the nomination. It was validation for outlets such as Morning Consult.
The rationale for this was simple, Ramlet said. During the primary season, there was still a “social desirability” bias at play — meaning, some people were not willing to say they supported Trump in a live interview but were willing to do so online.
“Not surprisingly, there was not the same level of comfort in saying you were a Trump supporter,” he said. “If you’re higher educated, higher income, you were less likely to admit that you were a Trump supporter in a live telephone interview than in the anonymous environment. When we got the results it was like, ‘Holy s—, this is exactly what’s happening. All the old-school pollsters are going to get it wrong again.'”
But as a study published last week by the firm found, that effect has been virtually wiped out in the general election.
Ramlet acknowledged, as a relatively new, online polling outlet, it took plenty of shots from the old-school pollsters for their validity. As statistician Nate Cohn proclaimed in The New York Times last year, “ready or not,” online polling had arrived in 2016. Public-opinion researchers predicted internet polling would overtake telephone polling, but they had hoped it would not happen before “online polling was ready.”
The most common attack against online pollsters is that there is no way to randomly and scientifically contact people on the internet, while telephone polling still provides the opportunity to conduct random, scientific samples.
Ramlet considers that excuse to be a weak excuse telephone pollsters use to stay in business.
“It’s a really convenient excuse for why they think people should still do live telephone polling, but it doesn’t pass the sniff test,” he said.
“Their first argument is that you can’t get a random sample, which is totally true if this is 1950. Like, anyone can possibly be interviewed because anyone can pick up their live telephone poll. The problem is, no one picks up their live telephone poll anymore. So they don’t have probability anymore either.”
“Many Americans don’t even have access to the telephone, so they can’t even claim what they’re claiming to have achieved. That was the easiest way to keep the media folks from reporting on the new technologies, was to kind of create this. The reason they did it is sort of a really legitimate business reason. They all bought call centres. So they have to defend the methodology until they’re dead.”
Though he said he wouldn’t defend every online polling outlet, praising both SurveyMonkey and YouGov in the online polling sphere, Ramlet said the “smart people are getting the hell out of the phone game.”
“It isn’t rocket science,” he said. “People aren’t answering their phones. Smaller samples than everyone else, and their claim to more accurate samples is just factually wrong.”
Here’s how Morning Consult’s online polling works: The firm uses a stratified sampling process and sends out a survey to multiple vendors, which it said gives it access to tens of millions of Americans. On average, the surveys are being taken by 1,000 people per day, and can include questions based on video and images.
The firm then weights the results when returned, which it said are able to be much lower than other online polls based on the number of people to which they have access. Then come the published results, with the entire process taking just a couple of days.
Telephone pollsters, having to contract out a phone bank to make calls, take about three to seven days to have results returned and published.
In August, the data-journalism outlet FiveThirtyEight found that Morning Consult had the lowest “house effect” among all national firms that conducted at least five national polls this cycle. That means the firm’s samples were the least pro-Clinton or pro-Trump of all outlets.
Telephone pollsters “correctly did see us as a threat, we were a threat,” Ramlet said. “Basically we’re saying, ‘You’re more like weather forecasters, whereas we’re meteorologists.'”
“This election is where the dam finally broke,” he added. “You just can’t get it wrong for a third time in this many elections.”
The firm has three divisions: A research and data science team, a technology team for its “Intelligence” platform, and a newsroom filled with 15 reporters and editors that report based on data created by the research team.
Once the dust from the 2016 election settles, Ramlet said, Morning Consult will make an increased effort to conduct market-research based polling and to beef up the Intelligence platform, a tool that allows a user to pull information from past questions asked in surveys and overlay them with other trends to create custom charts, graphs, and slides.
And, they will still be well-entrenched in politics.
“We wanted to make it really easy to search all those questions,” Ramlet said of the platform. “Even like [RealClearPolitics] and those platforms still feel like the 1990s. There’s more to polling and market research than just tracking the presidential race. There are a lot of questions that would be good to provide relevance on … I think this is really where we’re going to be going moving forward here after the election.”
The goal, the firm said, is to become the largest source of public-opinion data, regardless of the subject.
“We want to be the Bloomberg of public opinion,” McCarthy said.
In the next four years, the polling debate over online vs. telephone surveys will have likely reached its nexus — and Morning Consult believes they’re well positioned.
“If you think about the research and the survey side, eight years ago the debate was about cell phones and landlines and how much to include,” Kyle Dropp, Morning Consult’s chief research officer, told Business Insider. “Now the what we see in 2016 is that online polling is the right means of data-collection.”
He said that will be defining in the next election.
For Ramlet, the 2020 election cycle will be “dramatically different” from a polling aspect. He joked that “for all I know,” polling could be “Oculus Rift” driven.
“Candidly, I think what you’re going to see is that these traditional outlets are going to be further flanked by companies like ours and Survey Monkey and YouGov,” he said. “And the technology is passing them by.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.