Are social conversations more accurate predictors of how people vote? We think so. Remember Eric Cantor? His pollster showed him surging to a 34-point victory in last year’s Virginia primary; instead the former House Majority Leader got crushed, down 11 percentage points. The wild disparity got us thinking: Did Twitter conversations tell a different story? Yes. Months earlier our algorithms could spot when the tide was turning against Cantor, why it was turning, and who was turning it. Think how different the outcome might have been if Cantor had this real-time social intelligence at his fingertips – and acted on it.
These days, people share just about every detail of their lives with their friends on social networks, and frankly, few of us ever pick up the phone to talk to a pollster. If we do, we answer what they ask us, rather than talking freely about what matters most to us.
On social, millions of clues surface minute-by-minute about our circumstances, needs, desires and opinions. We are influenced, and we influence. Sentiment can change in a heartbeat, or with a slow, steady rhythm; new trends can break out fast, or patterns and correlations surface over time.
No politician or campaign can afford to ignore the predictive power of social intelligence.
Again and again at Verifeed, we’ve found that Twitter proves both prescient and a consistent bellwether of opinion.
Take the neck-and-neck U.S. Senate battle in North Carolina. Polls showed Democratic Sen. Kay Hagan in the lead up to Election Day, but Verifeed analysis of Twitter for client CQ-Roll Call told a different (and accurate) story of Republican dominance, viral amplification and disciplined message coordination by party activists and outside groups such as the Koch brothers’ Americans for Prosperity, in the final weeks of the campaign. In the end, Republican Thom Tillis triumphed 49 percent to Hagan’s 47 percent.
Verifeed filtered, parsed and patterned millions of Twitter conversations in North Carolina from August through Election Day to unearth insights about what trends were resonating and who was influencing sentiment and why. We found close race – the most expensive in 2014 at $81.6 billion – may well have been won – and lost – on Twitter.
While a Rasmussen Reports poll found Hagan ahead 45 percent to 39 percent, and another poll by SurveyUSA gave Hagan a 3-point lead (46 percent to 43 percent), Verifeed found the GOP and its surrogates in North Carolina engaged almost 14 times the number of people on Twitter as Democrats in the final week of the campaign.
Republican activists outperformed Democrats in sheer volume — and resonance — of tweets, with a veritable army of party activists faithfully re-tweeting and favoriting each other’s tweets regularly, if not hourly. The result calculated by Verifeed in the final seven days was direct engagement with 15,436,367 people by the top 20 GOP influencers — more than 14 times that of the top 20 Democratic influencers, who by contrast engaged just 1,746,178 people on Twitter.
The race began with Democrats clearly winning the Twitter battle for hearts and minds — engaging many more voters in issues traditionally important to Democrats such as public education, equal pay and job creation. Early on, Republicans were virtually absent from these conversations and slow to see the swell of support for Hagan around unpopular statehouse GOP cuts to education spending. The trend was clear from Verifeed’s parsing of Twitter data in August, but went unacknowledged by pundits, pollsters or media until later in September when it was reported as the key factor behind Hagan then bucking a strong national trend favoring Republicans.
The Twitter tide turned in mid October.
Republicans stepped up their social media conversations — in volume, coordinated repetition and consistency to dramatically boost their amplified influence. Their effort energized the party faithful to vote, and in a close race this proved significant to the outcome.
Tillis started to pull ahead on social media as Republicans bore down on the “fear” theme of tying Democrats to unease about Ebola and the Islamic State terror group, with consistent messaging and relentless repetition. The GOP, including Tillis, tweeted more about Ebola in the 23 days to Election Day than any other topic — with 997 tweets reaching 3.3 million people.
Republicans also kept up their focus on Hagan’s deciding vote for the Affordable Care Act; in particular tying Obamacare, already unpopular in the state, to economic uncertainty and job losses. The GOP engaged more people on this issue than any other — reaching 3.93 million in the final weeks compared to the Democrats who engaged only 152,496. Republicans also encroached on Democratic turf in later weeks to overtake Democrats on the economy and education. While Democrats had engaged 942,706 in education discussions from Aug. 15 through Sept. 15, compared with the GOP’s 76,877, the tide turned in late October, with Republicans engaging 343,314 compared to the Democrats down at just 207,171 as public interest in teachers pay and public school cuts dimmed.
Midterm elections are often won or lost on how effective the parties are at the get-out-the-vote ground game. Democratic influence and amplification on the topic of early voting, voter registration and GOP efforts to restrict both, outpaced the Republicans before early voting opened on Oct. 23. Twitter users Verifeed deemed Democratic influencers in this race engaged 3,159,499 with pleas to vote early, compared with the GOP’s 412,113. This Twitter trend played out at the polls when Democrats increased their share of the early vote, up 1.5 percent from 2010 according to the U.S. Election Project.
But again, Republicans stepped up in the last week and Democratic engagement on GOTV dropped 81 percent from the first half of October. In the final week they engaged 567,089 on Twitter compared with 840,266 for the Republicans.
So if you’re a politician – incumbent or newcomer – think carefully about stretching your campaign dollars more effectively and efficiently by adding social intelligence to your electoral toolkit.
Think again what happened to Eric Cantor and former U.S. Senator Kay Hagan. Pollsters can tell you only so much, limited as they are by a small sample size, long lead times, and the limitations of the questions asked. And of course we all want to be polite: If we can’t tell a pollster what they want to hear, well, we can always be “undecided” without offending anyone.
We’re not saying don’t use polls. But when you do, use social intelligence to improve the questions you ask, better target the voters you want to reach, and monitor in real time the impact of your statements and actions. See what trends are emerging, see who is influencing and changing opinions, and see even what phrases are resonating with whom. Monitor your opponent – what’s working and not working? The intelligence is powerful and can make all the difference in any race.
Once you know who you’re talking to, take some lessons from the Republicans in North Carolina:
- Tweet early, often, and get your followers to share with their friends around the clock.
- Influence is everything — the top 20 influencers actively engaged 33,538 active “unique amplifiers to reach 18,292,893 people in 23 days.
- Know your influencers and stay on message, even at the risk of deafening repetition.
- Listen, learn and leverage relationships: Twitter is a strategic platform both instructive and prescient. Republicans were listening and responding, and it worked.
Click here to view our final report for CQ.