The short version: AI tools can help you plan, draft, and test marketing emails in a fraction of the time it used to take, but only if you know how to prompt them correctly and edit what comes back. Done right, you can go from blank page to send-ready email in under 20 minutes. Done wrong, you end up with something that sounds like every other brand in your inbox and gets ignored.
Why most small businesses are still doing email the slow way
I speak to a lot of small business owners and solo consultants who tell me the same thing: email is the channel they know matters, and also the one they keep putting off. They open a blank document, stare at it, write three sentences, delete two of them, check Instagram, and eventually send something mediocre at 11pm on a Thursday.
That was me too, at various points. And I knew better. I have been doing this for years.
The problem is not lack of effort. It is a process problem. Writing a good marketing email from scratch is hard because you are making dozens of micro-decisions at once: the angle, the subject line, the opening hook, the offer, the call to action, the tone. Your brain stalls because it is trying to do all of that simultaneously.
AI does not solve the writing problem. It solves the blank-page problem. That distinction matters, and I will come back to it.
What AI does well in email marketing
Before we get into the how, let us be clear about what AI is good at here and what it is not.
AI is good at:
- Generating multiple subject line options quickly so you can test or choose
- Drafting a first version you can edit rather than create from scratch
- Reformatting the same core message for different audience segments
- Suggesting structures (problem, agitation, solution; story, lesson, offer) when you are stuck
- Writing plain-language summaries of complex offers or announcements
- Repurposing existing content (a blog post, a social caption, a client case study) into an email
AI is not good at:
- Knowing your specific audience the way you do
- Generating the kind of personal anecdote that builds real trust
- Staying on brand without very precise guidance
- Making strategic decisions about timing, segmentation, or what your list needs to hear right now
If you understand that AI is your first-draft engine and not your strategist, you will use it correctly. If you treat it as a finished-copy machine, you will send emails that feel flat and wonder why your click rates have dropped.
The exact process I use: from brief to send in under 20 minutes
This is the actual workflow I have built and refined over the past year and a half. It is not theoretical. I use it for my own list and I have walked consulting clients through it.
Step 1: Write a one-paragraph brief before you open any AI tool (3 minutes)
This is the step most people skip, and it is the reason their AI output is rubbish.
Before you type a single word into ChatGPT or Claude or whatever tool you are using, write a paragraph that answers these four questions:
- Who is this email going to? (Be specific. Not "my list." Freelancers who have been on my list for over six months and have not bought anything yet.)
- What do I want them to do when they finish reading?
- What is the one thing I want them to feel or believe by the end?
- What is the tone? (My tone is direct, warm, a bit blunt, British.)
That paragraph becomes part of your prompt. It transforms generic AI output into something that is at least in the right ballpark.
Step 2: Write the prompt (2 minutes)
Here is a real example of the kind of prompt that works for me:
"You are helping me write a marketing email for my newsletter. My audience is small business owners and freelancers in the UK and US who are interested in using AI to grow their business. They are smart and slightly sceptical of hype. My tone is warm, direct, and a little blunt. I never use corporate jargon. The goal of this email is to get people to book a free 20-minute call with me. The one thing I want them to feel is that I understand their specific frustration with email marketing being time-consuming. Please write a 300-word email with three subject line options. Use a personal story hook. Do not use phrases like 'I hope this email finds you well' or 'exciting news.' Start directly with the story."
That prompt takes about two minutes to write and it produces something usable about 70% of the time. The other 30% of the time I tweak one element of the prompt and run it again.
Step 3: Edit for your voice (7 minutes)
Read the output out loud. I mean this literally. If a sentence sounds weird coming out of your mouth, it will sound weird to your reader too. I rewrite any sentence that I would not say to someone in a conversation.
Things I almost always change in AI-generated email drafts:
- The opening line (AI tends to open with a statement; I usually prefer to open with a question or a scene)
- Any sentence that contains the word "journey" or "solutions" or "elevate"
- The transition between the story and the offer (AI handles this clumsily nine times out of ten)
- The call to action (AI defaults to "click here" or "learn more," which are the weakest possible CTAs)
This edit pass takes me about seven minutes for a 300-word email. Not twenty. Not an hour. Seven minutes.
Step 4: Generate and test subject lines (3 minutes)
Ask the AI to give you ten subject lines across different styles: curiosity, direct benefit, question, controversy, personal. Then pick two or three and test them if your email platform allows A/B testing. If it does not, pick the one that you would open yourself.
I have found that subject lines generated with a "write ten options in these five categories" prompt are consistently better than the three lines that come bundled with the email draft. More variety means more chance of finding a good one.
Step 5: Preview and send (5 minutes)
Paste into your email platform, check the preview on mobile (over 60% of emails are opened on mobile, and this number is climbing), send a test to yourself, and read it one more time. Then send it.
Total time: under 20 minutes for a solid, on-brand marketing email.
A real story: the email that changed how I work
About eighteen months ago I had a client consultation call booked in for 9am and I realised at 8:15 that I had promised to send a re-engagement email to her dormant list that morning before the call. She had not sent anything to these people in four months, the list had around 2,400 subscribers, and we needed something warm, honest, and not defensive about the gap.
Old process: I would have panicked, sent something half-baked, and spent the call explaining why the email was not quite right.
New process: I spent three minutes writing a brief (dormant list, four-month gap, tone is honest and slightly self-deprecating, goal is a single click to a blog post, no promotion, just reconnection). I prompted Claude. The first draft was about 65% there. I edited it in six minutes. We had ten subject line options in another two minutes. She approved it at 8:47. It went out at 8:55. Open rate on that email was 31%, which for a four-month-cold list is a good result. Three people replied just to say they were glad to be hearing from the brand again.
That experience showed me what the workflow can do when you have the brief sharp and the edit eye trained.
The honest point most articles will not make
Here it is: AI will make mediocre email writers faster at writing mediocre emails.
If you do not already understand what makes an email compelling (a clear angle, a specific audience, a single CTA, a reason to care right now), AI will accelerate your confusion, not solve it. You will produce more emails that do not work, more quickly.
This is not a criticism of AI. It is a criticism of the idea that tools replace thinking. The people I have seen get the best results from AI-assisted email are people who already had some instinct for good copy and used AI to remove the friction from getting that copy onto the page.
If you are at the beginning of your email marketing journey, the most valuable thing you can do alongside learning AI prompting is to study why emails work. Read the emails you open and ask yourself why you opened them. Read the marketing metrics that matter for small businesses so you know what you are optimising for. A 45% open rate on a list of 200 engaged people is worth more than a 12% open rate on a list of 10,000 who barely remember signing up.
How to prompt AI for different types of marketing emails
Not all emails are the same. Here is how I adjust the approach for the most common types.
Welcome emails
Brief the AI on exactly what you promised when they signed up and what they will get going forward. Welcome emails have the highest open rates of any email type (often 50 to 80% in smaller niches), so the quality bar matters more here than anywhere else. Tell the AI to make the email about the reader, not about you.
Promotional emails
Be ruthlessly specific in the brief. Tell the AI the exact price, the exact deadline, the exact audience segment, and the exact objection you are trying to overcome. Vague briefs produce vague promotional emails, which produce low conversion rates.
Re-engagement emails
Tell the AI the gap (how long since last email), the tone (honest, not grovelling), and the one thing you want them to do. Do not let AI be defensive or over-apologetic. A clean, confident re-engagement email outperforms an apologetic one almost every time.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Nurture sequences
Give the AI the full arc first. Write a brief for the whole sequence (five emails, moving from problem awareness to solution consideration to offer), then prompt each email individually within that arc. Trying to prompt a sequence email without context produces emails that do not connect to each other logically.
Newsletter editions
For regular newsletters, I brief the AI on the three or four things I want to include, write my personal section myself, and use AI to draft the supporting sections. The personal section is always mine because that is the bit readers are coming for. Everything else is scaffolding.
The prompting mistakes I see constantly
After working with clients on this across dozens of sessions, the same errors come up again and again.
Too vague on audience. "Write an email for my business" is not a brief. It is a starting gun for generic content. Always name a specific type of person.
No tone guidance. Without tone guidance, AI defaults to something that sounds like a mid-sized company's HR department. Friendly, corporate, inoffensive, forgettable.
Asking for too much in one prompt. "Write me a five-email welcome sequence, a monthly newsletter, and three re-engagement emails" produces something for everything and nothing good for anything. One email per prompt.
Accepting the first draft. The first draft is a starting point, not a finished product. Anyone who sends AI first drafts without editing them is handing their brand voice to a statistical pattern-matcher. Edit it.
Ignoring the subject line entirely. I have seen people spend fifteen minutes perfecting an email body and three seconds on the subject line. The subject line determines whether anyone reads the body. Treat it as a separate creative task.
Tools worth knowing about (and what each is best for)
I am not going to tell you one tool is the answer because the landscape changes constantly and what works best depends on your workflow. What I will say is this:
ChatGPT is good for iterative drafting where you want to have a conversation and refine. Claude tends to produce longer, more nuanced first drafts and handles tone instructions well. Gemini integrates neatly into Google Workspace if your email platform lives there. Specialist email tools exist (I have tested a few) but most of them are built on the same underlying models with a prettier interface and a monthly fee. The underlying model matters more than the wrapper.
If you want to see how AI fits into a broader marketing workflow beyond just email, the work I have done on AI automation for marketing teams with real examples goes much deeper into that territory.
Using AI to repurpose content into emails
One of the highest-ROI uses of AI in email is repurposing. You already have content. A blog post. A LinkedIn post that landed well. A podcast transcript. A client case study. AI can turn any of these into a solid email in minutes.
The prompt structure I use for repurposing: "Here is a blog post I wrote [paste the post]. Please turn this into a 250-word email for [specific audience]. The tone should be [my tone description]. The goal is to get them to [specific action]. Lead with the most interesting insight from the post, not a summary of the post."
That last instruction matters enormously. Without it, AI summarises the content, which gives readers no reason to click through to the original. Leading with the best insight makes the email itself valuable and makes the click-through feel like a bonus rather than the entire point.
This approach means your content is working harder. One blog post becomes an email. The email gets traffic back to the post. The post does its SEO work. If you are thinking about creating content that can work across multiple channels, understanding which income-generating activities compound over time helps you prioritise where to put your energy.
What to measure so you know if it is working
When you change your email process, give yourself at least eight sends before drawing conclusions. Email performance fluctuates for reasons that have nothing to do with copy quality: send time, day of week, what else is happening in your reader's life, whether your email landed in promotions.
The numbers I watch:
- Open rate: a baseline benchmark for your list specifically, not industry averages
- Click-to-open rate (CTOR): how many of the people who opened clicked. This is the copy quality metric. Industry average sits around 10 to 15%; anything above that for a small engaged list is a signal you are doing something right
- Reply rate: harder to track but incredibly valuable. Replies mean your email felt personal and worth responding to. AI-heavy emails with no editing tend to get low reply rates because they do not feel like a human wrote them
- Unsubscribe rate per send: a spike usually means something in the email felt off-brand or irrelevant. Worth noting.
Building a bank of prompts that work for you
Once you find a prompt structure that produces good output for your voice and audience, save it. Keep a simple document with your best-performing prompt templates. Over time you will build a library of maybe ten to fifteen prompts that cover 90% of the emails you need to write.
This is the compounding benefit of investing time in prompting from the start. The first email takes twenty minutes. The tenth email, once you have your prompt library, takes closer to twelve.
If you work with other writers or are building a content operation, the same logic applies to the broader team. I have seen what the freelance writing market looks like right now and the writers who are thriving are overwhelmingly those who have integrated AI into their workflow rather than resisted it. Email is one of the most transferable skills there is.
One more thing before you start
The best time to set up this process is before you are under pressure to send something. Build it, test it, iterate on it when you have breathing room. Because the moment you really need it is when a client wants something in forty-five minutes, or your campaign is going out tomorrow and you have not written the follow-up email yet.
That is not the moment to be figuring out how to prompt an AI tool. That is the moment you want a process you have already tested sitting quietly in your workflow, ready to go.
And if you are thinking about the broader picture of how AI fits into your consulting or business model, understanding what an AI marketing consultant does and when it makes sense to hire one is worth your time. Sometimes the fastest path is building the skill yourself. Sometimes it is getting someone to build it with you. Either way, the work still gets done.
Frequently asked questions
How much time can AI realistically save on email writing?
For most small business owners, the saving is somewhere between 50% and 70% per email once the process is set up. A 300-word marketing email that used to take 45 minutes from blank page to send typically takes 15 to 20 minutes using a strong brief, AI drafting, and a focused edit pass. The saving grows over time as your prompt library gets better.
Will my emails sound like AI if I use these tools?
They will if you skip the edit step. AI output without editing sounds like AI output: competent, structured, and slightly hollow. The editing pass, especially reading the email out loud, is what makes it sound like you. Spend at least a third of your total writing time on the edit, not just on reviewing the first draft.
What is the single most common prompting mistake for email marketing?
Not specifying the audience precisely enough. Writing "my customers" or "small businesses" gives the AI almost nothing to work with. The more specific you are about who is reading, what they already know, what they are worried about, and what they want, the better the first draft will be. Specificity in the brief is the single biggest driver of output quality.
Should I use AI for every email I send?
Not necessarily. Short, personal emails to individual clients or one-to-one follow-ups are often better written by hand because the personal texture matters more than the speed. Use AI for broadcast emails to your list, for sequences, for promotional campaigns, for newsletters. Reserve your unassisted writing for the moments where the human connection is the entire point.
This builds on my main AI marketing guide, my main guide on the topic.
Related reading: How to Set Your Rates as a New Virtual Assistant and The Virtual Assistant Niches That Pay the Most in 2026.
