How to Use Artificial Intelligence in Email

I originally authored this article for Marketing Artificial Intelligence Institute. You can see and subscribe to that content here.

How often do you listen to your intuition when making a decision?

That “gut feeling” is an essential human function. It’s the sensation from deep within our lizard brains that tells us something feels off, and we might be in danger. It’s helpful at worst, and life-saving at best. 

But when it comes to marketing?

Your gut feeling is probably wrong.

The sad truth is, too many marketers still rely on gut feeling to choose the email subject line that will most intrigue recipients. To select the send time that will guarantee an open. To structure a newsletter for optimal CTA clicks.

This could be because, although there is data out there to guide us toward the best decisions with our emails, many marketers are uncomfortable assessing that data. But more likely, there’s just so much of it that it’s impossible to review it all. 

You have to run A/B tests on ten different subject lines. Assess the data and compare it against tons of other factors. Ultimately, select the best one. And then replicate it for the next million emails you send, into eternity—and do it all in your brand’s voice—it’s just too much for a human to replicate accurately and at scale. 

This is where artificial intelligence can help.

AI is an email marketer’s intuition, if that intuition could read millions of data points in a matter of seconds and make the correct decision every single time. And this technology is not out of reach for the average marketer. 

In fact, AI-powered tools are incredible attainable, and could be the key to perfecting email marketing. 

Before we get into the myriad use cases for AI in email, let’s start by defining it.

A 2-Minute Definition of AI

AI sounds futuristic and overwhelming, but it doesn’t have to be. Here’s our favorite definition:

AI is the “science of making machines smart.”

Demis Hassabis, the CEO of DeepMind, is the one who said that. 

Machines are already pretty smart, you might be thinking. You’re right. We have programs that house email workflows, sending emails to certain people based on complicated branching logic. We have software that schedules social media shares.

But in these examples, it’s a human making the rules for the machines to follow. In an email workflow, we tell the machine which if/then logic to consider, what to do when a factor meets the criteria, and when to send a message.

When a machine uses AI, it gets smarter on its own. (This specific kind of AI is called “machine learning.”) It uses massive amounts of data to make recommendations and decisions that improve as it learns. Once the humans initially train the machine, it can then use this training to discover its own solutions and pathways. 

Conversely, regular machines require human interference to fix errors and ensure the machine doesn’t make them again. For example, if Facebook’s photo-tagging recommendation tool incorrectly identified your face in a photo, it learns immediately (with your interference) that it isn’t actually you in the picture. And whatever caused it to identify you likely won’t trip the machine up again. If it wasn’t using AI, it would keep incorrectly identifying you in photos until some poor programmer at Facebook manually tinkered with the program to tell it not to make that error again. 

With more than 1.5 billion active users on Facebook and hundreds of millions of photo uploads daily, there’s just no way a standard machine or human could keep up. 

But machine learning algorithms can. And when they’re trained to complete narrow tasks, they do so with incredible speed and accuracy. 

That's why AI is able to give marketers superpowers—especially when it comes to email. 

How AI is Used in Email Marketing

Gone are the days of making instinctive, gut decisions about your email marketing. 

AI can outperform humans, make accurate predictions, and give stellar recommendations. And it goes way beyond Smart Compose or Smart Reply in your Gmail inbox.

Here are the top use cases for AI in email marketing. 

Write Better Subject Lines

One subset of AI, natural language generation (NLG), can actually write language. It must be trained on an extremely large, structured data set, but then it’s capable of producing narratives for a multitude of purposes—press releases, social shares, and even email subject lines. 

You see, AI can assess the essential data to create an optimal subject line. This includes things like word choice, sentiment, emoji use, and more. And it does this at scale, so it’s accurate and fast. 

Phrasee has AI technology that writes subject lines so well, it beats humans in 98% of head-to-head tests. That’s because it assesses existing copy to learn a company’s brand voice, then uses NLG to produce tons of email subject lines guaranteed to perform well. It uses natural language processing (“NLP,” another subset of AI) to do a sophisticated sentiment analysis, and then predicts with amazing accuracy how the subject line will perform in practice. 

With AI, marketers can produce far better subject lines, generating more opens than a human could.

And in the end, more opens means higher revenue. 

Optimize Email Send Times

One aspect of AI that makes it invaluable to marketers is its ability to personalize at scale. Today’s consumers expect highly personalized recommendations and experiences, but it’s nearly impossible for marketers to deliver that to each individual consumer. And as an audience grows, it becomes even more overwhelming. 

Such is the problem with email send times. We marketers can debate all day whether our audience is more likely to open an email at 10:00 a.m. or 3:00 p.m. And the sad part is, we’re probably wrong no matter what—because for optimal email open rates, we need to send emails according to each audience member’s preference. 

That’s impossible for humans to do at scale, which is where AI comes in. 

Tools like Seventh Sense use artificial intelligence to assess mountains of data from your HubSpot portal, Marketo instance and / or corporate email system. It then uses the data to build a predictive model for each contact. 

That means each individual person in your list receives an email at the time they’re most likely to open it, based on data. 

Segment Email Lists

Smart email marketers know the value of solid lists. They segment their audience before sending custom emails—and segmented email campaigns have been shown to increase revenue by 760%. But that segmentation can be based on infinite criteria. How do you know you’re spending valuable time segmenting for the most impact?

Smart segmentation with AI is a great place to start, because it’ll help you develop lists that increase opens and clicks. 

For example, according to Salesforce, AI can analyze your current subscribers’ behavior—everything from purchase patterns to website forms—and identify patterns that help you segment your audience in new ways. And it does this way faster than any human could. 

Like we mentioned earlier, customization is the key to capturing today’s audiences. And with smarter segmentation, you’re left with optimized email lists that help your emails perform better.