Generic prompts result in generic emails. In this article you’ll discover five AI email marketing prompts for e-commerce that provide the AI with a specific audience, a concrete offer and a proven copywriting framework to follow. All three steps are essential for AI email marketing emails that sound like a real person wrote them, which is now vital with AI written emails now found to cause mistrust and unsubscribes.
The problem with most AI email prompts isn’t that the AI is too lazy to follow your instructions properly. It’s because AI email prompts are often too generic.
Ask Claude to “Write a promotional email for my skincare brand” and Claude will write something. It may even read well on first glance: grammatically correct, a logical subject line, a concisely worded body and an appropriate call to action.
The problem?
The content AI writes falls into generalities.
Its points come from the average of what it’s found published on the internet. It doesn’t know how to be original. And while it may appear sentient at times, AI still can’t think for itself.
Instead, AI hedges its language. It repeats itself, makes stuff up and writes about customer pain points in a dull way that’s emotionally flat.
More critically, AI doesn’t know your brand.
AI only has a surface level grasp of your customers’ pains and frustrations.
Even the latest voice apps, using an amalgamation of past emails, still write in a way that slips into the same “cheerful corporate” tone of voice that screams AI.
Most AI Email Marketing Prompts for E-commerce Produce Generic Outputs
AI writing lacks the specific details, visceral language or against the grain views that make emails interesting to read.
Instead, you get wishy washy copy that gets lost in the glut of AI emails clogging up people’s inboxes.
The good news is that improving the quality of AI written emails is simpler than you may think.
Not fast or easy.
Human judgement and editing is still needed.
But once you understand the process, you can create AI email marketing prompts for e-commerce that produce decent copy, rather than a pile of AI slop.
Generic Emails Cause Distrust and Drive Unsubscribes
For the last few years, businesses have been churning out cheaply written AI emails, blogs and social media posts. And people are starting to notice.
A Litmus study found that as many as 55% of people will lose trust in a brand and unsubscribe if it sends obvious AI written emails.
This is why it’s essential to use prompts that guide AI in the specific details to include. That way you can avoid the emails slipping into the generalities and robotic language that defines AI slop.
You need to educate the AI on the specific details of the problem the reader is dealing with.
Before you even prompt the AI, you need to create a document outlining the specific details to include in your emails around three key areas:
- Who are you writing to?
Whenever people read your emails they’re thinking “what’s in it for me?” Your emails must be laser focused on their specific pains, frustrations and dreams. You also need to train the AI in what solutions your readers have likely tried before, the exact words and phrases the reader uses when discussing their problem and what differentiates your brand from the competition.
- What is your product and what differentiates it from the competition?
Train the AI ln your product like you would a new member of your sales team. What is the origin story behind its creation? How does it address the customer’s problem in a radical new way? What proof do you have that it works?
- What tone of voice and framework should the email follow?
Sharing 5-10 emails that best capture your tone of voice is a good start. You should also include clear instructions on your viewpoints, the language you use and what to avoid in your emails.
There are also many ways to write emails. You can send direct promotional emails that feature the product, storytelling emails that teach readers a lesson that will motivate them to try your product or they could be thought leadership to build trust in your brand. Every type of email has its own framework. Feed the AI examples of each type of email you send so it knows what framework to follow.
When you attach all these specific details as a .txt file to your prompts it will naturally improve the specificity of your emails and reduce the amount of editing you have to do afterwards.
The 5 Best AI Email Marketing Prompts for E-commerce
Here are five prompts for five different types of emails you can adapt for your own brand:
Prompt 1: The Welcome Email
Your welcome email is the highest-opened email you’ll ever send, and a peak opportunity to build a bond with your brand. Most ecommerce companies waste this opportunity with a brief “Thanks for signing up!” message and a discount.
Here’s an AI email marketing prompt for an e-commerce welcome email that builds trust in your products and expertise:
Write a welcome email for [BRAND NAME] about [product]. You’ll find all the details in the attached file.
Subscriber context: Just signed up via [lead magnet / pop-up / checkout]. Has not yet purchased. May be comparing us to competitors.
Goal: Build trust, briefly introduce the brand story, and move them toward their first purchase — without being pushy.
Voice reference: Replicate the tone of voice shared in the email examples in the attachment.
Framework: Open with a specific problem the subscriber recognises. Introduce the brand as the solution through a brief moment the writer experienced that the reader would relate to (2–3 sentences). Close with a single low-friction CTA — not “shop now”, but something that moves them forward (e.g. “See our bestsellers” or “Read why X works”).
Constraints: 150–200 words. No exclamation marks. Do not open with “Welcome to the family”, “We’re excited”, or “Thanks for signing up.” Subject line under 45 characters.
Prompt 2: The Abandoned Cart Email
Abandoned cart emails can add as much as 10% to your bottom line from recovered purchases. Rather than offer a discount, this AI e-commerce email aims to increase cart completions by addressing a key objection:
Write an abandoned cart email for about [product]. You’ll find all the details in the attached file.
Subscriber context: Added [PRODUCT NAME] to cart [X hours] ago. Did not complete purchase. This is a [first-time visitor / returning customer who didn’t complete their purchase].
Goal: Recover the purchase by addressing the most likely reason they didn’t complete it — not by offering a discount.
Most common objection for this product: [insert your most common objection]
Voice reference: Replicate the tone of voice shared in the email examples in the attachment.
Framework: Open by naming the specific product they left in their cart. Address the stated objection directly (one sentence of social proof or guarantee). One CTA linking back to their cart. Close with a brief, honest PS that restates the guarantee or a relevant review.
Constraints: 120–160 words. One CTA only. Do not open with “Oops, you forgot something.” Subject line should name the product, not the behaviour.
Prompt 3: The Promotional Email
Promotional emails are the bread and butter for all ecommerce companies. Here’s a prompt for a promotional email that uses a customer story to build interest before making an offer:
Write a promotional email announcing [OFFER — e.g. 20% off sitewide / buy-2-get-1 / limited restock] for [product]. You’ll find all the product details in the attached file.
Subscriber context: [e.g. customers who haven’t purchased in 60–90 days / engaged subscribers who’ve never bought a bundle]
Offer: [Specific details — discount, bundle, limited quantity]. Valid until [DATE / TIME].
Goal: Drive purchases from subscribers who know the brand but haven’t acted recently.
Voice reference: Replicate the tone of voice shared in the email examples in the attachment.
Framework: Open with a brief customer result or story taken from one of the example emails in the attached file — not the offer. Introduce the offer in paragraph two. Use real, specific scarcity [insert exact number of units or hard deadline — not “limited time”]. One CTA. PS line repeating the deadline in plain language.
Constraints: 180–220 words. Do not open with the discount. Do not use “Don’t miss out” or “Act now.” Subject line under 50 characters.
Prompt 4: The Re-engagement Email
Subscribers who haven’t opened in 90 days are costing you money. Why? Because they’re damaging your deliverability every time you send to them and they don’t open. You should suppress them. Before you do, make one last attempt to reactivate them with this winback email for ecommerce prompt:
Write a re-engagement email [brand]. You’ll find all the brand details in the attached file.
Subscriber context: Has not opened an email in [X days]. Last purchase was [X months ago / never purchased].
Goal: Earn one more open and restart the engagement. This is not a hard sell — it’s an honest check-in.
Voice reference: Replicate the tone of voice shared in the email examples in the attachment.
Framework: Open with a direct, honest acknowledgement — no pretending this is a normal email. One genuine reason to re-engage (new product / something has meaningfully changed / useful piece of content). One low-friction CTA. Make opting out easy — include a clear option to unsubscribe. A subscriber who wants to leave should leave.
Constraints: Under 130 words. No “We miss you” subject lines. No discount as the primary hook. Subject line should feel personal, not promotional — ideally under 40 characters.
Prompt 5: The Post-Purchase Cross-Sell
Immediately after a purchase is when trust is highest and buying intent is still warm. This is a golden time to offer them a product that pairs with their original purchase and, ideally, enhances its results.
Try using this AI prompt for a post-purchase cross-sell email for ecommerce:
Write a post-purchase email for [BRAND NAME] recommending [PRODUCT B] to customers who bought [PRODUCT A]. You’ll find all the details in the attached file.
Subscriber context: Purchased [PRODUCT A] [X days] ago. This is their [first / second / third] purchase. They have not yet viewed [PRODUCT B] on the website.
Goal: Introduce [PRODUCT B] as the natural next step — not as an upsell, but as something that makes [PRODUCT A] work better.
Voice reference: Replicate the tone of voice shared in the email examples in the attachment.
Framework: Open with a specific, helpful tip for getting the most from [PRODUCT A] — something genuinely useful, not a prompt to leave a review. Bridge naturally to [PRODUCT B] as a complement. One customer review or result referencing both products if available. Single CTA.
Constraints: 160–200 words. Do not open with “Since you bought X, you might also like Y.” Subject line should reference [PRODUCT A] use — not the product you’re selling.
What AI Email Marketing Prompts for E-commerce Can’t Do on Their Own
These AI email marketing prompts for ecommerce will get you significantly better output than a generic instructions. The emails will be more specific, more structured and better aligned with your brand. But there are still limits to what any AI prompt can do.
Whichever AI model you use, even with a voice reference, dozens of email examples and specifics about your customers and products, AI still slips into generalities.
Due to how AI works as a prediction engine rather than based on real thought, it still repeats itself, uses AI slop signalling words and phrases and sounds emotionally flat.
If your subscribers notice before you do, you risk plunging engagement rates and soaring unsubscribes.
Most AI Models Can’t Improve Because they Can’t Learn from Your Data
Most AI models need training on your data every time you prompt them.
Unless you attach your data in a .txt file, AI models don’t know that your welcome sequence converts at 1.8% and that the third email is where people are dropping off.
They don’t know that your abandoned cart email is being sent too quickly after the abandonment and is landing in an inbox that isn’t ready for it yet.
And they don’t know which subject line patterns have been killing your open rates across the last three months.
Without this data, most AI models can only write emails in response to your prompts. They don’t know where an email fits into your entire email strategy, or how an email can be improved to boost results.
Smart Ecommerce Email Marketers Use Claude Code to Train the AI On Their Data
Smart email marketers are going beyond mere prompting to building Email Marketing Machines with Claude Code.
Claude Code can store all your data, learn from it and continuously improve the performance of your campaigns.
This includes continually refining its tone of voice based on your most successful emails, recommend segments of hot buyers to send to and run monthly reports with actionable recommendations each month.
Best of all, Claude Code can build apps for you, such as an app for creating your monthly calendar based on past results and creating entire sequences based on a simple template.
Claude Code is also email service provider (ESP) agnostic. It connects and sits above all your ESP AI tools and allows you to switch providers without losing all your data.
If you’re interested in harnessing Claude Code as the world’s best email marketing assistant, who’s always learning from your data and helps you continuously improve performance, I offer this as part of my AI email marketing strategist service.
If you want to find out how Claude Code can improve your readership, retention and revenue, get in touch to request a free performance report based on your data. Your report will also include actionable ways Claude Code can help improve your performance month after month.
Apply for a free email performance report
Frequently Asked Questions
Do these prompts work with ChatGPT as well as Claude Code?
The prompts shared in this article will produce usable output in ChatGPT, but Claude Code tends to produce more natural-sounding email copy. Claude Code’s instruction-following is also more reliable for the structural constraints (word counts, “do not open with X” instructions). If you’re using ChatGPT and getting generic output, the prompt architecture is the same — but Claude is worth trying for the copy quality difference alone.
How long should the voice reference section be?
5-10 of your best-performing emails is usually enough to give Claude Code a meaningful grasp of your tone of voice. Short, representative examples work better than long documents. The goal is to show Claude what your writing sounds like at its best.
How do I know which objection to use in the abandoned cart prompt?
You could check your customer service inbox for the most common pre-purchase questions about the product; your product reviews for what people say they were unsure about before buying; or your site search data for what people search for after landing on the product page. Any of those sources will surface the real objection faster than guessing. If you have post-purchase survey data, that’s the most reliable source of all.
Should I use a different AI email marketing prompt for each e-commerce segment?
The subscriber context section of each AI email marketing prompt for e-commerce is where most of the segmentation value is created. A welcome email prompt for a first-time subscriber looks very different from one for someone who signed up through a referral programme and already knows the brand. Swapping the context section is the easiest way to adapt a single e-commerce email prompt template to multiple segments without rebuilding it from scratch each time.
Discover more from The Copywriter's Crucible
Subscribe to get the latest posts sent to your email.
