Black Friday Email Marketing for Small Stores: Segments With AI

Plan Black Friday emails backwards over three weeks. In week one, export your order history and have AI sort customers into segments by recency, spend and what they bought. In week two, warm up your best customers with early access. In week three, send each segment its own offer, keeping spam complaints under 0.10% so Gmail keeps delivering you.

The segments matter more than the copy. A store that sends one email to everyone gets the unsubscribes of a mass mailing and the sales of a generic one. A store that writes four short emails for four kinds of customer needs the same hours and gets more orders and fewer complaints. AI is good at the sorting and the first drafts; you supply the counts, the offer and the check.

Follow me on Instagram@sagnikteaches

Why the whole-list blast is the expensive option

Every email platform will let you send once to everyone. The cost shows up later, in two places: customers who are annoyed by a discount on something they paid full price for last week, and mailbox providers who see a rise in spam reports and start sending you to junk. Google tells senders to keep the spam rate reported in Postmaster Tools below 0.10% and never to reach 0.30%. That sounds small, and it is. On a send to 3,000 Gmail addresses, 0.10% is three complaints. A single irritated segment can use the whole allowance.

Connect on LinkedInSagnik Bhattacharya

Segments let you protect the people who might complain. Your best customers rarely mark you as spam, but your lapsed and never-bought contacts do. Treat the second group with care: fewer emails, a gentler message, and no send at all to people who have not opened anything for a year.

Subscribe on YouTube@sagnikteaches

The same thinking runs through segmenting a small email list with AI, which covers the general method. This tutorial applies it to one seasonal week and your own order history.

Export the order history and strip it to what segmenting needs

The data you need is already in your store. From Shopify, export your customers and orders as CSV files. Shopify's order export downloads directly for up to 50 orders and emails larger exports to you, so for a real store expect an email with a link. In your own spreadsheet, build one row per customer with these columns:

• Customer ID (a row number, not a name or email)
• Number of orders, ever
• Total spend, ever
• Date of first order and of last order
• Top product category by spend
• Share of orders that used a discount code
• Number of refunds or returns
• Email status: subscribed, or not subscribed (never send to the second)

Leave out names, emails and addresses at this stage. The assistant only needs numbers to propose rules, and you re-attach the emails inside your email platform, where the segments will actually live. If your export is big, ask the assistant for the rules, then build the filters in the platform rather than pasting thousands of rows.

Before moving on, run three checks: duplicates (the same person in two rows), test orders from your own staff, and guests who bought without an account. A clean file is half the job, because every wrong row becomes a wrong email.

A prompt that proposes segments and shows its arithmetic

Paste a sample of 100 rows (or the summary counts) and ask for segment rules in plain language, with the order of priority, so each customer lands in exactly one segment. Here is a prompt that works well.

I run a small online pet shop with about 4,800 contacts. Below are columns and a sample of rows
from my order history (customer ID, orders, total spend, first order date, last order date,
top category, share of orders using a discount code, refunds).
Today's date is 9 November. Black Friday is 27 November.
Propose between 5 and 7 segments for Black Friday emails. For each segment give:
1. A name
2. The exact rule (use only the columns above)
3. The priority order, so a customer falls into the first rule they meet
4. What the email should do (early access, win-back, welcome, and so on)
5. A caution about that segment
Then tell me what to check in the counts before I trust the rules.
[paste columns and sample rows]

An illustrative reply, shortened:

1. VIP: 3+ orders AND last order within 180 days. Early access, no discount first.
2. Repeat: 2 orders AND last order within 365 days. Standard offer, add product they bought.
3. New: first order within 90 days. Welcome-style, one favourite product.
4. Discount-led: 70%+ of orders used a code AND last order within 365 days. Lead with the price.
5. Lapsed 6-12 months: last order 181-365 days ago. Win-back with a reason to return.
6. Lapsed over 12 months: last order over 365 days ago. One email, then rest.
7. Subscribers who never ordered. Explain the range, softer offer.
Check: segment counts add up to your list; no customer is in two segments; refunds over 2 move to a review list.

Two corrections you will usually make. The assistant will often miss the exclusion of customers with several refunds, who should not get an aggressive offer, so add that rule yourself. And it will sometimes write rules that overlap, such as "VIP" and "Repeat" both capturing a customer with three orders; the priority order is the fix, and the closing check is there to confirm it works.

A pet shop's list, segment by segment

Take an illustrative pet shop selling food, toys and accessories online, with 4,800 contacts. After applying the rules, the counts come out like this.

The counts add to 4,800, which is the first check. The second is whether the plan makes sense. A single send to the full list on the Friday would reach everyone once. The segmented plan sends 380 x 4 + 720 x 3 + 610 x 3 + 540 x 3 + 900 x 2 + 1,350 + 300 x 2 emails, which is 1,520 + 2,160 + 1,830 + 1,620 + 1,800 + 1,350 + 600 = 10,880 sends. That is more than twice the sends, but only the smaller segments get the long sequences.

A quick sum for expected orders. These conversion rates are assumptions to replace with your own past figures, not industry numbers: suppose VIP buyers order at 8% across the campaign, repeat at 5%, new at 4%, discount-led at 6%, lapsed 6 to 12 months at 2%, lapsed over 12 months at 0.5% and never-ordered at 1%. That gives 30 + 36 + 24 + 32 + 18 + 7 + 3 = about 150 orders. At an average order of $48, that is $7,200. If your own previous campaign converted at half those rates, the same plan brings $3,600, which is still worth the hours. Replace each assumption with last year's numbers as soon as you have them, and the plan becomes a forecast.

If you want to reach back to customers who have gone quiet, finding lapsed customers and sending a win-back shows the platform side of the lapsed segments.

The three-week schedule, day by day

Black Friday is the Friday after the fourth Thursday in November, which makes it 27 November in 2026, with Cyber Monday on 30 November. Counting back three weeks gives a start on Monday 9 November. Shift the dates if your year differs; the order is what matters.

Keep Thursday of week two free. It is the day you find that a discount code stacks with another one, or that the free-shipping rule applies to the wrong region. Every year, someone discovers this on Friday at 9 a.m. with 4,000 emails already delivered.

Subject lines per segment, tested before they are sent

One subject line for the whole list is the cheapest way to look generic. Ask the assistant for several per segment, then pick and test.

Write 8 subject lines for each of these segments for a Black Friday email from a small online pet shop.
Segments: (1) VIP customers, 3+ orders, getting early access, no discount; (2) lapsed 6-12 months, win-back;
(3) new customers, first order in the last 90 days.
Rules: under 45 characters, no all capitals, no more than one exclamation mark, no claims we cannot back up,
no "last chance" in the first email. Include the offer only where I have said there is one.
Our offer: 20% off sitewide from Friday for 4 days; VIP get it from Wednesday.
Return as a table: segment | subject line | preheader (under 70 characters).

Illustrative output for one row each, and the edits you would make:

The edits did three things: removed a claim ("We have missed you" is fine for people with a relationship, awkward for a stranger), cut an exclamation mark, and added a specific. If your platform supports it, split-test two subject lines on 10 per cent of a large segment, then send the winner to the rest. Segments under 500 people are too small for a split test to tell you anything; pick with your own judgement and learn from the results afterwards.

Avoid inventing urgency. "Only 3 left" on a product you restock tomorrow earns an annoyed reply and, in time, a complaint. If stock really is limited, say what is left and say when it ends.

Writing the emails: one before and after

Draft each email with the assistant, then rewrite the opening line and the offer yourself. Here is an illustrative before and after for the VIP early-access email.

Before, from a generic first draft: "Dear Valued Customer, we are excited to announce our Black Friday sale! Don't miss out on amazing deals across our whole store!"

After, with the details the assistant did not have: "Hi [first name], you have ordered from us four times this year, so you get the sale a day early. From today until Thursday night, everything is 20% off, including the grain-free range you bought in July. It opens to everyone on Friday."

The second names a fact about the customer, states exactly what the offer is and says when it ends. Feed the assistant these details through merge fields from your platform, such as first name and last category bought, and make sure each field has a fallback for rows with missing data. A blank merge field that sends "Hi ," to 300 people is the commonest small embarrassment of the weekend. Send yourself a test email using a customer record with missing fields before the real send.

Check the footer too. Marketing emails need a working unsubscribe link and your business details; a checklist for AI-written marketing emails covers what to include. AI will happily write a lovely email and forget the footer entirely.

Gmail's sender rules and a send plan that respects them

Gmail's rules apply even to small stores, in three ways. Google requires authentication, which means SPF or DKIM, from every sender; for senders of more than 5,000 messages a day to Gmail accounts it requires SPF, DKIM and DMARC. At that volume, marketing mail must also support one-click unsubscribe and show a clearly visible unsubscribe link in the body. And the spam rate you can see in Google Postmaster Tools should stay below 0.10%, and must never reach 0.30%.

A small store may never reach 5,000 messages a day to Gmail, but a send to a 4,800-contact list with 55 per cent Gmail addresses is 2,640 messages, and two full sends in one day would pass 5,000. Treat the larger-sender rules as your standard anyway; they cost nothing and the platforms mostly handle them for you. Google's page does not explain exactly how Postmaster Tools calculates the rate, and a small sender may see sparse data, so watch the complaint figures in your email platform as well.

Practical steps:

1. In week one, ask your email platform or domain manager to confirm that SPF and DKIM pass for the address you send from. Send a test email to a Gmail address and look at the message's "Show original" details, where the checks show as pass or fail.
2. Remove contacts who have not opened or clicked for 12 months from the main sends. Leave the "lapsed over 12 months" test email as the one deliberate exception, and send it to a small slice first.
3. Stagger the large sends. Send the VIP and Repeat segments first, wait an hour, check complaints and unsubscribes, and only then release the next segments.
4. Do not change your sending domain or address in the fortnight before the sale. A new domain has no reputation.
5. Stop a segment's next send if complaints rise. A segment that annoys is a segment to rest.

Free plans have limits that a big week will hit. Mailchimp's Free plan allows 250 contacts and 500 sends a month, no more than 250 a day, and no email scheduling, so a staggered plan like the one above needs a paid tier; Mailchimp's page shows its Standard plan from $20 a month for 500 contacts, with the cheaper Essentials tier beside it. Klaviyo's free plan covers 250 profiles and 500 emails a month. Those figures move, so confirm them on the vendor's page. Whatever your plan, count your planned sends (10,880 in the example above) and compare them with the monthly allowance, because a plan priced by contacts may still cap sends. For the platform choice itself, Klaviyo or Shopify Email for a small store sets out the trade-offs.

The 8 a.m. blast that triggered a spike

This is the mistake that gets repeated every year, in an illustrative case. A gift shop with 6,000 contacts, mostly casual buyers, sends its Black Friday launch to the entire list at 8 a.m. The offer is a flat 25% off. By lunchtime the shop has taken a record morning of orders, and by Monday its email platform shows complaint rates at four times the usual. Older contacts who bought once, two years ago, had not heard from the shop since. Their first email in two years was a discount shout. A few marked it as spam and a mailbox provider noticed.

The repair took until the next quarter: a cleaned list, a reduced send frequency, and a re-introduction sequence that rebuilt trust with the older contacts. The fix before the sale would have taken an afternoon: suppress the cold contacts, send the first wave to the people who knew the shop, and keep the old contacts to one gentle email a few days later.

The lesson is not "send less". It is "send in the order of trust". Best customers first, middle customers next, and the coldest contacts last, if at all.

What to measure over the weekend and afterwards

Pick four numbers per segment and write them in a simple table you update on Friday evening, Sunday night and Tuesday.

• Orders and revenue from that segment (use a campaign tag or discount code per segment so the store reports it).
• Unsubscribes and spam complaints per send.
• Click rate on the main button, which says more than opens, since mail apps can inflate open counts.
• Refunds and returns in the two weeks after, because a deep discount sometimes brings buyers who return the goods.

Once the weekend has settled, paste the table into your assistant and ask for three observations and three things to try next year. Treat the output as a prompt for your own thinking; it cannot know why a segment behaved as it did, only what the numbers show. Save the lesson list somewhere you will find it in October.

Before you start, one more link is worth following: abandoned cart emails for a small online store deals with the shoppers who start an order and leave, who are often the cheapest sales to recover during a busy weekend.

Black Friday email questions from small stores

How many Black Friday emails is too many?

There is no safe fixed number, so let the segment decide. Your best customers can take four emails across the fortnight if each has a clear reason, such as early access or a last chance. Lapsed customers should get two at most. Watch unsubscribes and spam complaints after each send, and stop the next send to any segment that reacts badly.

Can I upload customer data to an AI assistant to build segments?

Upload only what the job needs. Segmenting needs order counts, dates, totals and categories, not names or email addresses, so replace customer identifiers with row numbers first. Use a business plan if you can, and check the vendor's data terms for your account before pasting any export into a chat.

Should I send to subscribers who never bought?

Yes, but separately and gently. They joined for a reason, so a welcome-style email that explains your best products and the offer often fits better than a hard discount. Keep them out of your largest send if they have never opened anything, because unopened mail is more likely to be marked as spam.

Is a free email plan enough for Black Friday?

Only for a very small list. Mailchimp's free plan, for example, allows 250 contacts, 500 sends a month and 250 a day, with no email scheduling, which a multi-email sequence will outgrow quickly. Compare your contact count and planned sends with the plan limits a week before you start, not on the day.

Further reads

• Klaviyo Back in Stock Emails for Shopify: Setup and AI-Drafted Copy — Add a back-in-stock email for sold-out deals.
• Restaurant Email Marketing: A Weekly Send Written by AI — Another view of segments and subject lines, from a weekly send.
• AI Marketing Automation for Small Businesses: Chocolate Gift Seasons — How a gift business handles a seasonal rush.
• Online Shop AI Mistakes That Hurt Trust and Conversions — AI slips that cost a shop its customers' trust.
• How to Write Shopify Product Descriptions With AI That Convert — Tighten product pages before the traffic arrives.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Google's email sender guidelines and Postmaster Tools help; Shopify help pages on exporting orders and customers; Mailchimp plan pages.