AI-Based Email Marketing: What You Need to Know

AI-Based Email Marketing: What You Need to Know
05 June, 2023 • ... • 1497 views
Daria Zhuravleva
by Daria Zhuravleva

What do you call a blogger who uses machine learning to optimize content and promotional campaigns? An AIfluencer!

And now that we got your attention with an awful dad joke, let’s explore the question of using AI in email marketing. In this article, we’ll discuss what it is, the possible applications, and — the one y’all are waiting for! — whether a neural network can replace a marketing specialist.

What is AI in email marketing?

Since technologies involving AI like Midjourney provoke heated discussions, the first thing you’re thinking of is probably machines creating content for you. And, although it’s one of the use cases and we’ll come back to it later, the question is a little more complex. Let’s ask ChatGPT 4.0 for funsies.

Prompt: What is AI in email marketing?
Output: AI (Artificial Intelligence) in email marketing involves the use of machine learning algorithms to analyze and make predictions from large amounts of data about how email subscribers interact with the emails they receive. This data can then be used to create more targeted and personalized email campaigns that are tailored to individual subscribers’ preferences and behavior.

The overly verbose writing style hasn’t improved since we tried writing a blog article with ChatGPT — but the answer is not incorrect. But before we explore all the use cases, let’s clarify one thing: not all neural networks are the same.

Different types of AI

A neural network is a mathematical function that is designed to recognize patterns in data. What does it do with these patterns later? There are many types of algorithms but we’re not diving into details about the technology here. If we were explaining it to a preschooler, we’d put it like this:

  • “Detectors” simply give you information that the patterns are here. Think of these algorithms as smoke detectors — they only go off when someone lights a cigarette in the room. Solid examples are speech and image recognition algorithms.
  • “Predictors” extract patterns from the historical data and give you information about the future. For example, Netflix algorithms see that you liked “Black Mirror”, “Stranger Things”, and “Doctor Who” — and recommend you more science fiction shows to watch.
  • “Creators” use the patterns they recognized earlier and recombine them to create something new. These are called generative algorithms — they generate the output based on the input. ChatGPT and Midjourney are the most popular examples but there are many more.

The thing is, all three types of algorithms can be used in email marketing. Now, let’s take a look at some of the use cases.

Ways to use AI in email marketing

We already know that not all neural networks are generative — but there’s a use case for every type of algorithm in email marketing. Buckle up for a ride!

Drafting email copies

Will AI tools replace human copywriters? Not now — sending raw AI output to your entire contact list may not be a good idea. However, if you’re uninspired and short on time, drafting a copy with an AI can be a solid starting point. Drafting is a key word though — for example, this one from ChatGPT 4.0 needs some editing:

Prompt: Write a welcome email for a meditation mobile app called ZenMedit that features funny guided meditations recorded by celebrities. The app targets people with anxiety. Invite them to read a guide on new users. Style: friendly, informal
Output: Subject: Welcome to ZenMedit, the meditation app that will leave you laughing!

Dear [User],

Welcome to ZenMedit, the app that promises to turn your meditation practice into a fun and enjoyable experience. If you suffer from anxiety, you’ve come to the right place.

At ZenMedit, we believe that laughter is the best medicine, which is why we’ve teamed up with celebrities and comics to produce guided meditations that are both hilarious and effective. Our expert teachers will guide you through breathing exercises, visualization techniques, and introspective meditations, all while cracking jokes and keeping you entertained.

We understand that starting a new app can be a little daunting, but don’t worry, we’ve got you covered. We have put together a guide for new users that will give you all the information you need to get started and make the most of your ZenMedit experience. The guide is user-friendly and straightforward and covers everything from account set-up to meditation sessions. You can access the guided tour anytime from your profile.

We are excited to have you here and can’t wait for you to try out our meditation sessions. Remember to keep an open mind, take a deep breath, and get ready to laugh your way to a calmer you.

Thank you for choosing ZenMedit. Let’s meditate together and spread positivity.

Peace and serenity,

The ZenMedit team.

This newsletter copy is pretty solid for an AI. However, it’s too long for a welcome email and it still emits important details on how to use the app. Generative algorithms can’t write about a new product because their learning data only has information about stuff that already exists.

But ChatGPT is not the only generative AI on this planet. Even better, there are algorithms designed specifically for email marketing like Hoppy Copy. This AI-powered service has different prompt formats for each type of marketing email — here’s what it looks like:

Welcome email prompt on Hoppy Copy
Source: Hoppy Copy

Once you fill in the form, you get five slightly different (and disturbingly good) copies and bonus subject lines for A/B testing in the future. Here’s what we got for the prompt above:

Hoppy Copy welcome email example
Source: Hoppy Copy

This copy might need some editing like going more specific about the blog content and adding links — but, overall, it’s a pretty decent welcome email for an AI. The takeaway is, no matter which tool you use, an AI-generated draft will speed up your writing process.

Writing catchy subject lines

As we mentioned before, Hoppy Copy also drafts several versions of email subject lines. Here’s what we got from the algorithm.

Bonus subject lines for a welcome email example
Source: Hoppy Copy

Honestly, these variants look cringey and outdated. They are too long for a subject line, the title case and too many emojis look like a throwback to the 2010s, and the copies themselves are generic. But you don’t have to use them as they are! If you’re out of ideas, even underwhelming AI output is a good place to start brainstorming. And drafting is not the only way algorithms can help you with email subject lines.

Phrasee and similar AI-powered tools don’t just write content for you. They learn on your brand guidelines and identify if your content fits in the tone-of-voice, study your subscribers’ behavior, and analyze the subject lines you created — like this:

AI subject line analysis example
Source: LXAHub

All these features can help you stay on brand and create the most effective subject lines that will get you opens and clicks. And if you’re skeptical, Phrasee runs A/B tests for AI-generated vs. human subject lines so you can see the difference.

Generating visuals

Midjourney is the best AI “painter” for now even though it still has trouble with text on images and human hands. Despite its flaws, it can be a savior for marketing teams without an in-house illustrator. The algorithm can help you with email header pictures and illustrations to spice up your copy.

But it’s not the only way to use AI for email design. Sarbacane allows you to generate email templates based on your website design. Such templates don’t just save time compared to manually designing each email — they also ensure your brand is consistent across all channels.


List segmentation is dividing the entire subscriber base into groups and sending them different newsletters based on these groups’ characteristics. You can segment the contact list based on location and time zones, gender, age, behavior, and other criteria depending on your goals. It’s possible to do manually but it has certain drawbacks:

  • It’s time-consuming, especially with a large subscriber base and convoluted behavioral criteria.
  • It’s prone to the human factor — you can send the wrong emails to the wrong people.
  • It’s less efficient — without big data analysis, you can only create large subscriber groups using superficial criteria, which is better than nothing but has room for improvement.

Neural networks are great at quickly recognizing patterns and extracting valuable insights from humongous datasets. When they see your huge subscriber base with all the behavioral data, they create smaller groups based on more complex criteria in seconds. For example, this AI tool by Emarsys allows you to create prediction-based segments with customers who are likely to move down the sales funnel or drop over a time period of your choice.

AI predictive segmentation example
Source: Emarsys

Technologies like this bring lead nurturing to the next level. With predictive segmentation, you can send reactivation emails to customers you’re about to lose with higher chances to win them back. And fine tune your nurturing campaigns so warm customers get one last gentle push to purchase at the right time, which will skyrocket your conversions. 

Optimizing send times

There’s no consensus about the perfect send times. We did an overview and our own research on the best time to send an email — none of these numbers count as a foolproof success recipe. Manual or automated, send times optimization without AI has a lot of problems:

  • The time zone data is yielded automatically from IP addresses — but what if your subscribers use VPNs? In this case, this data has no information on where your customers really live and what their daily schedule looks like.
  • Once a customer subscribes to your newsletter, the time zone can’t be changed. So, if they were abroad at that time, they’ll get your emails with send times adjusted to the wrong location.
  • Even if you create time zone-based subscriber groups, they’re not checking their inboxes at the exact same time.

That’s where AI comes at handy. Machines are better at detecting data patterns than humans — they’ll optimize send times specifically to your subscribers’ email browsing habits. A great example is the Smart Send feature created for Hubspot by Seventh Sense. Here’s a little sneak peek at what it looks like:

Send time optimization example
Source: Seventh Sense

Cleaning up email lists

List hygiene is vital for email deliverability. If you keep unengaged subscribers in your list for long enough, it might trigger spam filters and ruin your sender IP’s reputation. Some ESPs have a list-cleaning feature that allows you to automatically remove inactive contacts.

However, automated non-AI filters are rather coarse — like a mesh sieve that only allows you to strain pasta water but grains like rice will pass through it. AI filters are better at detecting more subtle patterns of subscriber activity — it will be a fine mesh sieve that allows you to sift flour. Some ESPs already have the AI-powered list cleaning feature. But if yours doesn’t, you can use separate apps for it — for example, Listclean. This tool can remove temporary emails, spam traps, role-based addresses, and other contact list infestants. It also provides AI-based verification to make sure you only send promotional newsletters to real people.

Listclean list dirtiness analytics example
Source: Capterra

Email retargeting

Targeting ads or emails is showing promotional content to the people who don’t know a lot about you but might be interested in your offer. And retargeting is showing data-powered content to those who are already interested but haven’t bought stuff from you yet.

Retargeted emails include content based on the customers’ browsing history — here’s an example:

Browse abandonment email from REBEL8
Source: Drip

In this case, AI tracks user behavior and tweaks the email content for each user — both the browsed items and recommendations. You can achieve that both by integrating your ESPs with third-party AI tools and in-build ESP features. For example, Mailchimp has a tool that predicts the potential customer lifetime value of all your subscribers and suggests sending them special promotions based on these predictions. 

Recommending certain products

We already mentioned Netflix as an example of AI-powered product recommendations. This AI use case in marketing is probably one of the oldest in our list. Many retailers use it to recommend products that are:

  • Bought together often — for example, fitting batteries or lenses for a camera
  • Bought by people with similar purchase history
  • Bought by people who liked the same product
  • Better alternatives to the product
  • Similar to a certain product

Here’s a great example from Amazon:

Amazon recommendation email example
Source: Rejoiner

AI-powered recommendations bring email personalization to the next level — without it, you can do as much as generic selections of the most popular items from your store. Machine learning will help you provide more relevant content for your subscribers and upsell more efficiently.

Benefits of AI-Based email marketing

Now that we know all the use cases of AI in email marketing, let’s discuss how investing in machine learning can do good for your business.

Higher ROI

According to McKinsey’s survey, AI implementation in sales and marketing leads to cost reduction by up to 20%. Even better, using AI in marketing also leads to increasing revenue by 5% and above, which sums up a higher marketing ROI.

Here’s how it works:

  • AI optimizes campaigns so they get more customer engagement and sales.
  • AI optimization also helps you send less irrelevant and ineffective email content, so you burn less marketing budget without earning it back.

And here’s how you get a higher ROI with the help of machine learning!

Easy personalization opportunities

Customers don’t like being treated like walking ATMs — email marketing statistics prove it. According to Wunderkind, 73% of US consumers expect personalized communication — and 90% of them are loyal to brands that are good at adjusting messages to their needs. In 2023, personalization is not just a marketing trick — it’s a must.

Email personalization is still possible without AI — for example, you can use merge tags to address your recipients by their first names. But AI digs deeper and allows you to include or refer to a lot of customer data in your emails automatically. This type of personalization leads to an incremental increase in revenue. For example, an eyewear retailer GlassesUSA increased their revenue up to 88% just by implementing AI-powered personalized recommendations. And you can go beyond emails! Personalizing push notifications and home page content with AI will bring you more money and happier customers.


Imagine starting every morning with a coffee brewed in a cezve. To follow this routine, you need to wake up at least 20 minutes earlier to pour coffee and water into a cezve, put it on a stovetop, and keep watching it so it doesn’t overspill. But what if you had a coffee machine that automatically makes you a perfect cup of Joe minutes after you wake up? You could spend these 20 minutes sleeping or doing more pleasant things instead of watching a cezve on a stove.

Now imagine that you have to manually set up each bulk email you’re sending. In this case, creating automated sequences that will work for as long as you wish to will save you hundreds of hours. And using AI will save you even more time — especially when it comes to designing complex, behavior-based sequences. And, instead of doing boring but important work, you can refine your email marketing strategy, take Selzy’s free course on email deliverability, or just chill. It’s a win-win situation!

So, can AI replace email marketers?

Long story short, not really. AI can perform marketing data analysis more efficiently than humans and help you save time and money on mundane tasks. However, even the most advanced neural network can’t replace an email marketer today. Here’s why:

  • Someone needs to push the button. Automated writing content? Someone has to write prompts for ChatGPT or whatever GAN you’re using. Hired a neural network for data analysis? Someone has to come up with a task that an AI can solve, prepare a dataset, and interpret the results. Neural networks are not autonomous entities — these are apps that need turning on and off, and they still need human input to give output.
  • AIs can’t “think” strategically. Artificial intelligence can only do narrow and specific tasks — but marketing is more than that. Defining a UVP, building a brand, coming up with a long-term strategy — all that is on human email marketers for now.
  • AIs don’t have soft skills. Even if you outsource most of the tasks to AI including content creation, managing a marketing team is still a job for humans.
  • AI content generation is not that good yet. Yes, you can use generative algorithms to speed up the writing process or create illustrations — however, all the AI output still needs human supervision and editing.

So, yeah, even though AI is a helpful instrument for campaign optimization and saving time for more labor-intensive tasks instead of getting buried in the routine, none of the existing algorithms can replace the entire email marketing department. Good news for the anxious marketers scared of losing their jobs!

Final thoughts

AI can’t replace email marketers — but it can ease their jobs with automation and boost the efficiency of marketing efforts. There are many types of machine learning algorithms. Some simply search and detect patterns, others learn to create something new. All of them can be used for email marketing — here’s how:

  • Drafting email copies — they still need editing but it’s a great starting point.
  • Writing catchy subject lines — both generating and optimizing your ideas for maximum efficiency.
  • Restructuring email campaigns — finding the best structures for automated sequences using the historical data.
  • Segmentation — looking deeper into your subscribers’ behavior.
  • Cleaning up email lists — eliminating all the inactive subscribers for the sake of your domain reputation.
  • Email retargeting — creating personalized messages to boost sales.
  • Recommending certain products — sending data-powered recommendations your subscribers will love.
05 June, 2023
Article by
Daria Zhuravleva
I'm a writer with 3 years of experience, knowledge and interest in all things IT and marketing, and a passion for the English language. As a staff author at Selzy, I see my mission as an educator who makes your life easier by explaining complex topics in a digestible and somewhat entertaining way. Hobbies include birdwatching, all things music and art, writing freeform poetry, and hiding in the woods.
Visit Daria's

Latest Articles

Selzy Selzy Selzy Selzy