AI in email copywriting: What it actually should be used for

Email marketing remains one of the most profitable channels and is widely reported as one of the most lucrative in terms of return on investment. The problem, of course, is that getting people to open your email in a spam-filled world is a real challenge. Data from Selzy's email marketing platform shows the proportion of emails written with AI went from 10% in 2023 to 43% in 2026. By 2025, the figure was just over 50%.

But what if you could use advanced technology to go further than just optimizing your email marketing and actually transform the way you do it? That's where AI comes in. AI does more than just automate tasks; it also helps businesses engage with their audiences in a completely different way, which leads to more opens, more clicks, and ultimately more sales. Let's look at how AI can transform your email marketing and give your business an edge.

Where AI is actually useful in email copywriting

First drafts and brainstorming

Writing good email copy takes time. GPT-powered tools can generate compelling copy in seconds. They review successful email campaigns, identify patterns, and generate new copy that appeals to your audience.

A fairly typical workflow would involve providing the AI tool with a goal (e.g., re-engaging inactive subscribers), describing your offer and its key benefits, and then getting back a variety of copy versions. The marketer would then pick the best copy, add relevant supporting arguments, and tweak it to fit the brand voice.

Example prompt for generating an email:

Subject lines and copy variations

The first thing the recipient sees is the subject line. The AI tools use NLP to analyze emotional triggers and keywords that help increase open rates. Instead of having a copywriter come up with one subject line and then spend another 20 minutes hunting for alternatives, the marketer can generate 10 or 20 options in a few seconds.

The same applies to:

• subject lines;
• preheaders;
• CTAs;
• opening hooks;
• shorter and longer versions of the same message;
• variations for A/B tests.

Rewriting and editing

It's often easier to hand an editing job off to AI than to do it yourself. You can give it a very specific instruction to improve your draft: "Cut this by 30%." "Remove the repeated bits." "Make this easier to scan on mobile." "Rewrite this in simpler language." "Turn that paragraph into three bullets." "Make the CTA more specific."

These are tasks with very clear inputs and very clear expected outputs. There's little scope for the model to invent information or stray from the campaign's point. The model can also rework the copy to a different tone or format. A long product announcement can be turned into a short promotional email. A formal draft can be made more conversational. A long paragraph can be condensed into a headline, two sentences, and a CTA.

The final piece will still need a human touch. The model can produce awkward phrasing, get the translation wrong, or remove details that at first glance seem unimportant but are, in fact, vital.

Personalization and segmentation

It's no longer enough to just put a recipient's name in an email. AI can look at subscriber behavior, purchase history, product views, and even how long they spend on various pages of the website to build out relevant, personalized communication journeys.

For example, AI can automatically:

• recommend products that are similar to those that have been viewed or left in the shopping cart;
• suggest further services or accessories that go with the previous purchases;
• tailor the email content to the interests of your subscribers (e.g., educational articles, reviews, curated product picks);
• adjust the email frequency and format in line with recipient engagement.

Subscribers receive messages tailored to their actual interests rather than a generic email campaign. Selzy's wider AI marketing advice is that machine learning can find patterns in customer data and enable a far more granular level of segmentation than you can achieve with just demographic segmentation.

Email personalization is still achievable without AI (you can use merge tags to address your recipient by their first name, for instance). AI goes further and automatically enables you to include or reference a huge amount of customer data in your emails. This level of personalization and email experience generates an incremental uplift in revenue. One eyewear retailer, GlassesUSA, saw a revenue increase of up to 88% simply by implementing AI-driven personalization recommendations.

Repurposing content

Most email teams already have more source material than they can possibly reuse efficiently.

A webinar can be the basis of a whole series of newsletter ideas. A product page can be the basis of a launch email. A case study can be the basis of a customer story. A blog post can be the basis of a short educational sequence. AI is great at taking the relevant bits and reconfiguring them into an email.

For example, a marketer can provide a 2,000-word article and ask for:

• a short newsletter introduction;
• three subject lines;
• a five-email educational sequence;
• a product-focused version for existing customers;
• a CTA to take the reader back to the full article.

Conversion: AI vs. human

The fast uptake of AI-generated copy doesn't mean that AI-written emails are consistently better than human-written ones. Selzy's data is actually quite controversial in this respect.

In 2023, AI-written emails had a marginally higher open rate (20.36%) than human-written emails (19.69%). The reverse was true in 2024: human-written emails had a 14.64% open rate, compared with 11.89% for AI-written emails. The gap grew in 2025 when human-written emails had a 20.09% open rate and AI-written emails had a 16.8% rate.

There's no year-on-year advantage for either group. Open rates are influenced by so many other things besides the body copy itself (subject line, sender reputation, audience quality, timing, sector, campaign type, etc.) that the data doesn't support the idea of "AI-written" or "human-written" being a standalone conversion tactic.

The open-to-click figures add another dimension. AI-written emails had an open-to-click rate of 9.51%, compared to 7.42% for human-written emails. That's an interesting difference, but it must be read with caution. It doesn't prove that AI copy results in more clicks. Campaign type, offer, audience, and how the AI is used could influence the outcome.

What the numbers do show is that AI-generated copy isn't necessarily worse at driving action. The open-rate data provide no evidence of a clear overall performance advantage.

Challenges to consider

AI in email marketing is still very much a new technology, so there are a few things to bear in mind if you're using it:

1. Tone of voice: Early AI-generated emails tended to be a bit too formal. Most AI tools can be tuned to your brand's voice, but you'll still need to edit the final copy.
2. Quality: The quality of AI output depends largely on how well you write your prompt. Even with a very detailed prompt, you'll probably still need to do some editing and polishing.
3. Accuracy: AI can make mistakes, so you absolutely must fact-check.

Remember that one AI tool is unlikely to do it all. Some will be superb at writing the copy and designing the campaigns; others will be brilliant at analyzing the audience and working out when and how often to send the emails.

How to create an email campaign using AI

Creating an AI-powered email campaign is not simply a matter of typing in a single prompt. It's a multi-stage process: defining your objectives, creating the campaign, and analyzing the results. There is a human element at every stage. The AI tool will generate the content and propose various structures based on your input, but it won't make the final decisions or take ownership of the outcome.

Step 1. Define your idea and structure

Before using AI, answer these important questions:

1. What's the objective of your email? To drive sales, share information, re-engage a customer, or welcome a new subscriber?
2. Who are we going to target? New customers, existing customers, or people who haven't bought from us in a while?
3. What are you offering? A product? A service? A discount?
4. What do you want the recipient to do? Go to your site, complete a form, or reply to the email?

If your campaign isn't straightforward, you can develop a few hypotheses and have the AI tool suggest the best structure or tone. You'll then need to have a clear idea of what you're aiming for and work out the structure of the email.

The AI tool can suggest various structures and content sequences, but you will, of course, have to approve the final one. A standard email structure includes a greeting, a description of the customer's problem or need, your offer, the benefits of that offer, and a call to action. Without a clear plan, the AI tool is likely to produce content that's all over the place and won't achieve your business objectives.

Step 2. Generate content

Give the AI tool your campaign objective, your target audience, the structure you'd like to use, the tone of voice you're after, and some examples of previous emails. It will then generate a draft email.

What AI can generate:

• subject lines and preheaders (the text recipients see before opening an email);
• email body copy;
• multiple variations for A/B testing;
• ideas for visuals and graphics.

The more detailed your instructions, the more relevant the results will be. For example, simply asking AI to "write an email about a discount" will likely produce generic content. However, specifying your audience, campaign objectives, and offer will help AI generate copy that better meets your needs.

What you should review:

• facts and figures: AI may invent information or misrepresent data;
• tone and style: AI-generated content may sound overly formal or too conversational;
• legal considerations: Check terms and conditions, consent requirements, and mandatory disclosures;
• logical consistency: AI can produce well-written content that doesn't actually support your business objectives.

Step 3. Analyze the results

AI can also assist with campaign analytics. Built-in email marketing tools automatically collect and display campaign performance data. The system can compare results with previous campaigns, identify which subject lines and email variations performed best, and suggest improvements for future emails.

However, some aspects still require human evaluation. Make sure to check whether:

• the content met your audience's expectations;
• the email contained any factual or logical errors;
• the campaign achieved its intended goals, such as generating sales, form submissions, or consultation bookings.

Wrapping up

AI can be a useful tool in email marketing, but don't forget it always needs a human check. Use it to speed up brainstorming, content creation, and optimization while keeping control over your brand voice, messaging, and accuracy.