Hillary Clinton’s emails have been a constant source of news throughout this election cycle. And while attention on Donald Trump has been focused on secret servers that allegedly talk to the Russians recently, it turns out both candidates have a lot in common when it comes to sending emails.

Today, a new study from Persado explains the similarities, differences, and mistakes both camps are making when it comes to the emails they send supporters, and the results are both surprising and a lesson for all businesses, since email is still the most efficient marketing channel in existence.

Persado, a cognitive content platform, analyzed more than 6,000 subject lines using a neural network classifier that identifies and categorizes emotional elements found in the text. The emails were sent from January 1 through August 31 of 2016. At a high level, there are some distinct differences in approach between Clinton and Trump.

Clinton uses savvy marketing techniques that are more akin to the types of emails seen from major retailers and leading marketers. There is evidence of A/B split testing, a standard conversion rate optimization technique, and a third of her emails include an explicit request for action or response.

The most used phrase in her emails is “thank you,” suggesting a personal touch is important in her messaging. Trump’s most used phrase, by comparison, is “approval rating.” Her camp sends a high number of emails that feature an inquisitive, inviting tone and demeanor to help build personal connections with supporters.

How high a number of emails? Clinton sends 900 percent more emails than Trump. That’s how high.

But volume isn’t everything. While nearly half (45 percent) of email subject lines sent from the Clinton camp use first-person (“I”) and second-person (“you”) pronouns, true personalization is lacking.

Clinton doesn’t appear to understand the value of merchandise for her supporters. Only 4 percent of her emails leverage merchandising, but these are precisely the emails that seem to be driving interest from her followers, generating an average response rate of 21 percent. That’s high considering the average for her emails is 16 percent.

Trump, on the other hand, uses a smaller number of bold, short messages that emphasize his personal brand through positive, emphatic language. His emails are designed to fire up his supporters. But the intention of many of these emails is, on the whole, much more self-serving. While that may not be surprising, the results of his promotional emails might be.

Trump frequently plugs signed editions of his book, Trump: The Art of the Deal. His supporters, on the other hand, do not seem to respond to the book promos, and mentions in the subject line of merchandise sales dropped the average response rate by approximately 14 percent. With his supporters shunning business book opportunities, the suggestion that Trump would be better off building a right-wing media empire from his email list isn’t such a bad idea.

In fact, Trump’s open rates typically improve whenever his camp capitalize entire words or phrases (“FW: WE. WANT. TRUMP.”) These subject lines show up in 8 percent of Trump’s emails and drive “yuge” attention, generating an average 22 percent open rate. Exclamations work well for Trump, too. Subject lines like “Disgraceful!” and emotionally charged introductory clauses such as “BREAKING: Big Endorsement” were efficient, with open rates jumping from an average 18 percent to 29 percent.

We know from our research and webinars at VB Insight, and studies from the likes of Experian Marketing Services, that something as simple as including the recipient’s name in the subject line provides a serious uplift in click rates. However, both candidates seem to be lacking when it comes to real personalization. Why is that?

“Ultimately, real personalization transcends using someone’s name,” Greg Dale, COO at Persado, told me. “To truly inspire a person to act, the message needs to resonate on an emotional level. The science behind the thousands of marketing campaigns we’ve been involved with proves this, again and again. That’s why we were curious to see whether Trump and Clinton, who certainly elicit an emotional response in all of us, used emotional language to get people to open their emails, and if those emotions were effective.”

And both camps aren’t leveraging emotional triggers with any real effectiveness, either.

“Turns out, they usually made an emotional appeal, but there’s definitely room for improvement when it comes to really understanding what would motivate their audiences,” Dale said.

What were the most surprising results from the findings?

“Two surprises,” Dale said. “Trump was positive and Hillary less so; and that the emotions each candidate utilized were almost identical.”

That’s right. Trump, it turns out, is marginally more positive than Clinton when it comes to tone.

“Subject lines coming from the Trump camp with a positive emotional tone edged out the neutral and negative, 51.6 percent versus 46.7 percent,” Dale said. “Hillary’s was flipped, more often employing a neutral or negative tone (51.36 percent) than positive (46.6 percent).”

And while Trump is seemingly more associated with fear, uncertainty, and doubt — based on what we read in the press and on social media — it turns out that both candidates use the same combination of emotions in the emails they send.

“You would expect differences in their individual posturing and style, but it was eye-opening that Trump and Clinton used a nearly identical lineup of emotions as their top 5 (in order: anticipation, joy, fear, trust, pride for Clinton and joy, anticipation, fear, trust, pride for Trump),” Dale said. “Their messages might have been different, but the emotion they’re trying to tap into were almost exactly the same.”

The problem with these emotional triggers? As an aggregate across all U.S. emails analyzed by Persado, “pride” and “trust” are the most effective motivators. Unfortunately for both Clinton and Trump, those two emotions are in fourth and fifth place when it comes to usage throughout the analyzed email messages.

With a lack of any real personalization — possibly due to a lack of accurate data beyond an email address — a failure to understand how merchandising affects both groups of supporters, and a misunderstanding of which emotions motivate people to action, both camps are making email marketing mistakes that, with less than a week to go until the vote, could move the needle.

The full report, which includes 30 in-depth points of comparison, is available from Persado today.