The short answer: Use AI for the repetitive work: drafting campaigns and automations in your brand voice, planning a calendar around your sales moments, and suggesting subject lines and segments. Keep a person on facts, offers, timing and deliverability. Brief the AI on your brand first, check every email before it goes live, and judge results on clicks and orders.
That is how to use AI for email marketing without sending generic copy or getting a discount code wrong. This guide is written for Shopify merchants and works with any tool, including Shopify's own. It covers what to hand to AI first, a step-by-step workflow, the checks to run before anything sends, and the deliverability basics AI won't handle for you.
What can AI do in email marketing?
Two kinds of AI show up in email tools. Generative AI writes: subject lines, body copy, whole sequences. Predictive AI looks at your data to suggest who to email, when, and what they might buy. Most Shopify apps now mix both. Here is where each task sits:
| Task | What AI does well | What you still check |
|---|---|---|
| Campaign copy | First drafts in a set tone, fast | Facts, offers, prices, product names |
| Subject lines and preview text | Several variants to choose from | That the subject matches the email |
| Campaign calendar | Mapping campaigns to seasons and retail moments | Your real launches, stock and sale dates |
| Automations (welcome, cart, post-purchase, win-back) | Writing the sequence once, ready to switch on | Triggers, delays and exit rules |
| Segments | Suggesting groups from order and engagement data | Who is excluded, and why |
| Design | Laying out an email in your colours and logo | Mobile view, alt text, contrast |
| Reporting | Summarising what sold and what didn't | How orders are attributed |
The pattern: AI is good at producing a lot of reasonable work quickly. It is not good at knowing what's true about your store today. Your job moves from writing to briefing and checking.
What should you automate first on a Shopify store?
Start with automations, not one-off campaigns. An automation is written once and then sends every day to every shopper who triggers it, so the time you spend checking it pays back the longest. In order:
- Abandoned cart or checkout email. It reaches shoppers who were minutes from buying. See our abandoned cart email examples for what to put in it.
- Welcome email or series. Every new subscriber gets it while their interest is highest.
- Thank-you and review request after purchase. Easy to write, and it starts the next order.
- Win-back for lapsed customers. A reason to come back, sent when they've gone quiet.
- Then your campaign calendar: weekly or seasonal campaigns around the moments that matter to your category.
For each one, AI can write the emails, but you set the rules:
| Automation | Usually starts when | Stops when | You decide |
|---|---|---|---|
| Abandoned cart or checkout | A shopper starts checkout and doesn't pay | They place an order | The delay, and whether email two has an offer |
| Welcome | Someone subscribes through a form | The series ends | Whether there's a first-order discount |
| Thank you and review request | An order is placed or fulfilled | The series ends | How long after delivery to ask for a review |
| Win-back | A customer hasn't ordered for a set time | They order again | How long counts as lapsed for your products |
With those five running, AI is doing the work that recurs. One-off ideas can wait.
How to use AI for email marketing, step by step
1. Brief the AI on your brand before it writes anything
Generic AI copy usually comes from a generic brief. Before generating a single email, write down, or set in your tool:
- Tone of voice: two or three words (for example "friendly, plain, a little dry") and one sentence you would never send.
- Who buys from you: one line on your typical customer.
- What you sell: categories, best sellers and what makes them different, in your own words.
- Your policies: delivery costs and times, returns window, any free-delivery threshold.
- Words to use and avoid: product names spelled correctly, terms customers use, phrases that aren't you.
- Language: the language your customers read, and British or American spelling.
Keep this brief in one place and reuse it. Every tool that writes for you works better with it.
2. Use real store data, not guesses
Ask the AI to work from what is in Shopify: your product titles, prices, collections and discount codes. Never ask it to invent a statistic, a review or a "customers love" claim. If a figure isn't in your store or your own reports, it shouldn't be in the email.
3. Plan the calendar, then write
Map the year before drafting: your launches, restocks, sales, and the retail moments your customers notice. Let AI suggest the rest, then delete anything that doesn't fit your shop, stock or region. A calendar keeps sending steady rather than bursty, which helps both sales and inbox placement. Our Shopify email marketing guide covers how often to send and what a first month looks like.
4. Draft fast, then edit like an editor
Treat the AI's draft as a first draft. Read it aloud. Cut the opening sentence if it's throat-clearing. Swap vague words ("elevate", "curated", "must-have") for specifics: the fabric, the size range, the delivery date. If a draft is close, edit it; if it's wrong in tone or structure, ask for a fresh version rather than patching every line.
5. Set up automations with triggers you understand
For each automation, you should be able to say in one line when it starts, how long it waits, and when it stops. On Shopify, triggers usually come from store events such as a started checkout, a placed order or a fulfilled order. Check two things: that a shopper who buys leaves the flow, and that you don't have the same automation running in two tools. Shopify's own help centre warns that automations add up: two welcome automations mean customers get twice as many emails (Shopify Help Center).
6. Test before anything goes live
Send a test to yourself. Open it on a phone. Click every link and the main button. Check what happens to personalisation when data is missing: "Hi [CUSTOMER_FIRST_NAME]" should never reach a customer as a raw tag, and "Hi ," looks broken. If your tool has a preview for desktop and mobile, use both.
7. Measure clicks and orders, not opens
Open rates are a weak signal. Google says it doesn't track open rates and that low open rates aren't necessarily a sign of deliverability problems (Gmail sender guidelines). Judge AI-written emails on clicks, orders and revenue, and know how your tool attributes an order to an email, because a longer attribution window credits more orders to email.
8. Regenerate, rewrite or retire
Once a few weeks of results are in, keep what earns clicks and orders, ask the AI for new versions of what doesn't, and switch off anything that only adds volume. AI makes emails cheap to produce, so the discipline is sending fewer, better ones.
AI prompts for Shopify email marketing
If you write with a general AI assistant rather than an email app, the prompt is your brief. Paste your brand brief from step 1 first, then use prompts like these. Swap anything in [brackets] for your own details.
Campaign draft: Using the brand brief above, write a campaign email for [collection or product] aimed at [customer type]. Include a subject line under 60 characters, preview text, a headline, 80 to 120 words of body copy and one button label. Use only the product details I give you: [paste titles, prices, materials]. Do not invent reviews, statistics or urgency.
Subject lines: Give me ten subject lines for this email, in five different styles (plain, curious, benefit-led, playful, personal). Mark which ones would be misleading if the email only contains [what it actually contains].
Welcome series: Plan a three-email welcome series for new subscribers to [store]. For each email give the send timing, the job it does, a subject line and a short outline. Email one includes the [offer] code [CODE]; the others don't mention a discount.
Calendar: Suggest one campaign a week for the next three months for a [category] store selling in [country]. Include public holidays and retail moments my customers would notice, and leave gaps for these dates I already have planned: [dates].
Editing pass: Edit this draft to sound like the brand brief. Cut filler, replace vague adjectives with specifics from the product details, keep it under [number] words, and list any claim in the draft that I need to check is true.
That last prompt is worth running on every draft, including ones your email app wrote. Asking the AI to list its own claims makes the fact check quicker. For more on what AI writing tools can and can't do, see can AI write marketing emails?
What to check before an AI-written email goes out
This is the part most guides skip. Run every AI-written email through this list before you activate or schedule it:
- Prices, product names and stock match your Shopify store today.
- Discount codes exist, work at checkout, and have the end date and exclusions the email states.
- Dates are right for your region, delivery cut-offs and time zone.
- No invented social proof. No made-up reviews, ratings, customer quotes or "bestseller" claims. In the US, the FTC's rule on fake reviews and testimonials covers reviews by someone who doesn't exist, such as AI-generated fake reviews (FTC).
- The subject line matches the content. The CAN-SPAM Act bans deceptive subject lines and requires your postal address and a working opt-out.
- Personalisation has fallbacks for missing names and empty carts.
- The audience is right: subscribers only for marketing, recent buyers excluded from "come back" emails.
- It sounds like you. If you wouldn't say it to a customer in your shop, change it.
- Images have alt text and the button has enough contrast to read on a phone.
- Another language? If the AI writes in a language you don't read, have a fluent speaker check the first few emails.
Deliverability checks AI won't do for you
Great copy is wasted in the spam folder, and AI can't fix sender reputation. Gmail's sender guidelines set the baseline (Google):
- Authentication: all senders need SPF or DKIM. Senders of more than 5,000 messages a day to Gmail accounts need SPF, DKIM and DMARC.
- One-click unsubscribe: required for marketing messages from senders over 5,000 a day.
- Spam rate: keep it below 0.10% in Google Postmaster Tools and never let it reach 0.30%.
- Consent: send only to people who asked to hear from you. Spam reports lower your domain's reputation over time.
AI makes it tempting to send more. Don't jump from one email a month to five a week; raise frequency gradually and watch unsubscribes and spam complaints. For the full set of rules, see our guide to Gmail and Yahoo sender requirements for Shopify.
Where AI email marketing goes wrong
- Everything sounds the same. Without a brand brief, AI copy drifts to the same polished, empty phrases. Your customers notice.
- Made-up details. A countdown that doesn't exist, a "last chance" that isn't, a product feature the item doesn't have.
- Too many emails. Content is cheap now; attention isn't.
- Two tools, one shopper. An app's automation and Shopify's both switched on, so customers get doubles.
- Nobody owns the review. "The AI sends it" isn't a process. Someone checks before it's switched on.
- Judging on opens. Opens are an unreliable measure; clicks and orders tell you what worked.
How tinyAlbert handles this on Shopify
tinyAlbert is a Shopify app that writes and sends your email marketing for you, built around the workflow above: a brand brief first, everything written, then you check and switch on. Using what the app shows:
- The brief: at setup it asks four things: layout (Colourful, Minimalist or Professional), tone of voice (seven options), content language (ten options) and up to 1,000 characters about your company and products. It syncs your Shopify store data, and a brand kit holds your logo, colours, social links and up to five hero images.
- What it writes: automations (Welcome Series, Welcome Series 10% Discount, Abandoned Cart with the shopper's cart products, Thank You For Purchase, Product Review and Win Back Flow), a campaign calendar built for your store's category and retail moments, and four signup forms.
- How much it sends: about one campaign a week by default. You choose how many campaigns a week, the send time, and whether campaigns go to subscribers only or to subscribers plus Shopify contacts.
- The checks: every automation and campaign has Preview (desktop and mobile), Edit, Send test and Regenerate (an AI rewrite), and nothing sends until you switch on its Activate toggle. The campaign builder adds subject-line best-practice checks and a preflight checklist.
- Deliverability: sending is authenticated from day one, with DKIM, SPF and DMARC verified on a managed sender, and there are 28 AI-recommended segments, 12 of them for deliverability and inbox recovery.
- Measurement: an order counts towards an email when it's placed within 10 days of the shopper's latest open or click.
Plans start at $3 a month for up to 100 contacts, after a 7-day free trial. You can see how it works step by step, or read about done-for-you email marketing if you'd rather not write emails at all.
It isn't the only route. Shopify's own Campaign Autopilot, free on paid Shopify plans apart from channel costs, generates branded emails and automations such as abandoned checkout and welcome series, with the merchant approving tactics within guardrails (Shopify Help Center). For a side-by-side look at the options, see our roundup of the best AI email marketing tools for Shopify.
Frequently asked questions
Is AI-written email bad for deliverability?
Not by itself. Inbox providers judge your sending: authentication, spam complaints and whether people asked for your email. Gmail asks every sender to keep spam rates under 0.30%, and bulk senders to use SPF, DKIM and DMARC. Where AI hurts deliverability is volume: sending far more because content is cheap, to people who didn't sign up, raises complaints.
What should I never let AI do without checking?
Anything a customer could hold you to: prices, discount codes and their end dates, delivery promises, stock claims, reviews and ratings, and legal footers. Also check who an email goes to. AI is good at drafting and suggesting; a person should approve facts, offers and audiences before an email is switched on or scheduled.
How much does AI email marketing cost on Shopify?
It ranges from free to a monthly subscription. Shopify's Campaign Autopilot is free on paid Shopify plans apart from channel costs, and Shopify Messaging includes 10,000 free emails a month. Email apps charge by contacts or sends; tinyAlbert starts at $3 a month for up to 100 contacts, priced by list size, after a 7-day free trial.
Will customers know an email was written by AI?
Only if it reads like it. Customers notice generic phrasing, wrong details and too many emails, not the tool behind them. Give the AI a clear brand brief, edit out filler, and keep specifics such as real product names, delivery dates and your own sign-off. An edited AI draft should read like any other email from your shop.
Does AI email marketing work for small Shopify stores?
Small stores often gain the most, because the owner is usually the only person writing emails. Automations written once and a planned calendar replace hours of writing. The trade-off is review time: someone still needs to check each email before it goes live, and keep the brand brief and offers up to date.
Can I use ChatGPT or another AI assistant for my Shopify emails?
Yes, for drafting. Paste your brand brief and real product details, ask for a draft, then copy it into your email tool. What a general assistant can't do is send, trigger automations, pull each shopper's cart or track orders. You still need an email app or Shopify Messaging for that, and the same checks before anything goes out.
tinyAlbert is an AI email marketing app for Shopify stores, made by tinyCo, the company behind tinyEmail. About tinyAlbert.