Last week at Dreamforce, InsightSquared unveiled a new product — a custom report builder that complements our library of more than 400 pre-built sales, SaaS and demand generation reports. We’re calling it Slate by InsightSquared. By pairing Slate with our comprehensive report library, InsightSquared now provides customers with a full-stack, full-service sales analytics solution. Whether you want to track popular growth metrics or shine a light on a unique aspect of your business, InsightSquared has you covered.

As with most visual products, showing always “trumps” telling. Why the quotation marks? Because we figured if Slate could visualize patterns in the massive amount of poll data and social media activity surrounding the US presidential election, then its primary use case — advanced sales reporting — will look like child’s play. Plus, it’s more fun to read about.

While we believe our findings are provocative, they aren’t necessarily groundbreaking. Depending on how closely you’re following election news, you may have read similar observations in publications with heavy data journalism resources. What is groundbreaking, however, is the ease with which we — a sales analytics company — were able to process and visualize massive amounts of election data, thanks to Slate.


Whereas political news sites have entire teams of data journalists digging into the numbers, we were able to slice into millions of data points and uncover the same findings in our analyst’s spare time. It’s further proof that you don’t need a big team to make sense of data.

With that said, read on to find out what we uncovered, and to learn how Slate enabled us to dig, visualize and interpret.

Summary of Poll Data

We conducted our analysis on publicly available poll data from the Huffington post and CNN — refer to the bottom of the article for direct links to references.

In total, we analyzed 741,843 responses to some variation of the question “If the 2016 presidential election were held today and the candidates were Democrat Hillary Clinton and Republican Donald Trump, for whom would you vote?” The polls included in the analysis were conducted between July 2015 and September 2016.

One interesting note about the polls we analyzed is the majority were conducted online. In fact, three quarters of the total responses came from online polls.

While the number of people polled through other methods (automated phone calls, live calls, etc.) stayed fairly consistent over time, there was a drastic increase in respondents to internet polls when the election season hit high gear in March.

Now, on to the meat of the analysis.

Section 1

Section 2

Section 3

Section 4

The Convention Bumps

There was lots of talk in the media about the positive bump each candidate enjoyed following their respective conventions.

Much of the convention-cycle reporting emphasized polling bumps, but few reports looked back in the subsequent weeks and months to see if either bump turned into lasting support. We also had seen little compelling visualizations demonstrating trends in poll data during that time, so that seemed like a promising spot to start our analysis. Here’s what we turned up:

There’s no question each candidate enjoyed a boost during their convention. The RNC provided Trump with around a 3.5 point boost, which put the two candidates on even ground immediately following the convention. That changed the following week, however, when Clinton saw a similar boost following the DNC and Trump’s gains were effectively wiped out. Clinton’s lift, on the other hand, had a more lasting impact. Her post-convention bump decayed slowly for about a month before recovering slightly in mid-September.

Polling by Party

This election features historically unpopular candidates, and both candidates have suffered from a lack of trust with voters. As a result, the election is shaping up to be extremely partisan, with the core voting blocs on each side sticking to their party’s nominee. But what about voters who fall in the middle?

To get a better sense of what to expect from the independent vote in November, we broke down poll responses by self-reported party identification.

As expected, the vast majority of self-identified Republicans plan to vote for Trump, and a similar percentage of self-identified Democrats say they will vote for Clinton.

Trump holds a narrow advantage among self-declared independent voters and the number of self-identified Democrats (7.7%) who say they’ll vote for Trump is slightly larger than the number of self-declared Republicans who intend to vote for Clinton (7.2%). Nevertheless, these numbers suggest that party polarization coupled with divided independents will prevent either candidate from running away with the popular vote.

Demonstrating Intraparty Divides: Massachusetts Primary Data

Both candidates slogged their way through contentious primaries to earn their respective party’s nomination. To get a better sense for how that divide showed up during the primaries (and consequently, what kind of impact intraparty rifts may have on the general election), we decided to dive into data from a single state.

Given that our home base is in Boston, we chose Massachusetts.

Massachusetts has long been a Democratic stronghold, and this cycle is no different. Here are primary results as total votes broken down by party:

However, the story gets more interesting when you take it a step further and dig into the number of votes for each candidate:

Total Primary Votes By Candidate

A Deep Dive Into Presidential Election Data

And total primary votes broken out by party and candidate:

Massachusetts Primary Votes By Party and Candidate

A Deep Dive Into Presidential Election Data

Even though Democrats secured a majority of the total primary votes within the state, there was an almost even split within the party between Clinton and Sanders. On the other hand, Trump earned a significant majority of Republican votes within the state.

Percent of Votes by Party

A Deep Dive Into Presidential Election Data

The Massachusetts primary data demonstrates the importance of unity for Democrats in the general election. Republican voters appear to be more prepared to rally around Trump, but it remains to be seen whether Sanders supporters will ultimately support Clinton.

Much of today’s election reporting centers on the importance of party unification. The need for unity is most acute on the Democratic side of the aisle, at least as far as Massachusetts is concerned.

Examining The Social Presence of Each Candidate

Another critical aspect of this election is the impact of social media. Social media is, obviously, playing a major role in the campaign strategy of both candidates, and there have been several tweets from both sides that ended up being the center of passionate political discussion.

Given the oversized influence social media is having in this election cycle, we decided to close out our analysis by digging into publicly available twitter data. The findings below are the result of analysis conducted on over 24 million tweets. Thanks to Slate, we were able to ETL and visualize that amount of data with less than 200 lines of code.

Much has been made of Trump’s outsized influence on Twitter, but as it turns out, there have actually been more tweets (1,712 more, in fact) from Clinton’s official account than from her opponent’s over the past six months:

The candidates’ level of activity is only one half of the story. The other half, which is arguably more important, is public response to their tweets. As a proxy for public response, we examined the number of retweets each candidate received over the past six months.

We found that Trump’s tweets triggered more retweets than Clinton’s did, even during months when Clinton was more active on Twitter.

This trend could support the Republican candidate’s claims that a “movement” is coalescing around him, or it could speak to an opposing perspective that it’s the controversial nature of some of his tweets that have sparked such a strong reaction.

Number of Retweets for Each Candidate

A Deep Dive Into Presidential Election Data

The last step we took in our analysis was to go back and dig further into the types of tweets candidates were posting.

We performed sentiment analysis that categorized words as positive or negative and then assigned tweets a “polarity” score of -1 to 1 (for more details on the intricacies of sentiment analysis, read here and experiment with a demonstration here). Included in the sentiment analysis was a sentiment intensity analysis, which yields a score from 0 to 100 which represents the ratio of positive sentiment to negative sentiment in each tweet.

Once we’d scored all the tweets, we took the intensity scores and plotted their trajectory month-over-month. The results are shown below (the Y axis is sentiment: a lower score indicates the candidate’s tweets are negative, higher scores indicate the tweets are more positive).

Ratio of Positive to Negative Tweets Sent by Candidates

A Deep Dive Into Presidential Election Data

Despite Trump’s generally negative rhetoric during speeches and on the campaign trail, the words he uses in his tweets appear to be consistently more positive than Clinton’s. Missing from sentiment, of course, is the context of those words.

This analysis was a fresh way for us to test out our new product, but, of course, politics aren’t really what we’re all about. We’re all about sales performance analytics. If you’re interested in applying this type of analysis the varied data sources that inform your go-to-market strategy, we should talk. Slate takes the technical overhead out of data analysis and visualization, and lets you dive straight into uncovering new insights.

It’s the perfect tool for enterprising sales ops leaders, analysts, and business intelligence teams. Head on over to to learn more.

Recent Posts