Email segmentation for ecommerce: the segments a small store needs

What email segmentation means for a store#

A segment is a group of contacts chosen by a rule, such as “ordered in the last 30 days” or “bought outerwear but never knitwear”. Segmenting means you decide who an email is for before you write it, rather than sending every message to everyone.

Segmentation helps in three ways. The message is more relevant, so more people click and order. Fewer people receive email they do not want, so unsubscribes and spam complaints stay low. And you can send more often overall without any one customer feeling flooded. Gmail asks senders to email only people who want their messages and to keep spam complaints below 0.10% ^S1, and relevance is the most practical way to stay there.

Segments that matter for a small store#

Take Ninefold, a small apparel brand selling knitwear, outerwear and basics. These are the segments worth defining, written as rules you could say out loud:

Location matters more for apparel than it first seems. A Ninefold customer in Sydney and one in Toronto need opposite products in July, and a single “autumn collection” email is wrong for one of them.

Choosing lapsed windows: 60, 90 or 180 days#

How long counts as lapsed depends on how often your customers normally buy. Look at the typical gap between a customer’s orders in your own data, then set the windows around it. For a seasonal apparel brand that sees most repeat orders once or twice a year, a practical split is:

• 60 days: still warm. Show what is new; no discount needed.
• 90 days: cooling. Remind them what they liked, with a light, time-limited offer.
• 180 days: at risk. A clear win-back offer, then a last message asking whether they still want to hear from you. See the re-engagement campaign guide.

Subscribers who have not clicked or ordered in a long time are a different case from lapsed buyers: if a re-engagement email does not bring them back, removing them protects your sender reputation. Email list cleaning covers how.

RFM in plain terms#

RFM stands for recency, frequency and monetary value: how recently a customer ordered, how often, and how much they have spent. The full method scores each customer from 1 to 5 on all three and end up with dozens of cells. A small store does not need that much. Three buckets on two of the dimensions give you most of the value:

Add monetary value when you have enough orders for it to separate people, for example by moving your top spenders into the VIP row regardless of order count.

How many segments a small list needs#

Fewer than you think. With a list of a few thousand contacts, four to six segments you actually use each month beat twenty you defined once. A useful test: if a segment is so small that one or two orders swing its results, merge it with a neighbor until you can tell what worked.

• Under about 1,000 contacts: start with buyers and non-buyers, plus one lapsed segment.
• 1,000 to 5,000 contacts: add VIP, the three lapsed windows and one category split.
• Above 5,000: add location and finer category interest, and compare results per segment.

Define “engaged” by clicks and orders rather than opens. Apple’s Mail Privacy Protection loads remote content in the background when a message arrives ^S2, so an open is often recorded without anyone reading the email, and an “opened in the last 90 days” segment can include people who never saw your messages.

Building segments in Claude or ChatGPT with Tellnod#

With Tellnod connected to Claude or ChatGPT, you describe a segment in plain words instead of building it from filters. The assistant uses estimate_audience to count the people who match before anything is saved, and Tellnod removes unsubscribed and bounced contacts from the count. Requests a shop owner might type:

• “How many customers have ordered twice or more but not in the last 90 days?”
• “Save that as a segment called Slipping repeat buyers.”
• “Email people who bought outerwear last winter but no knitwear, about the new merino sweaters.”
• “Tag everyone in this CSV as VIP.”

Order-based rules need your order history: import it from your store, or with the Shopify connector in Claude, ask for the customers you want and Tellnod turns them into an audience. When you are ready to send, the confirmation card shows the recipient count, who is excluded and why, and the send time, and only your Confirm schedules it. The Free plan covers up to 500 contacts; see pricing for larger lists, or connect Tellnod to try it on your own list.

Common segmentation mistakes#

• Segments too small to learn from. A dozen recipients cannot tell you whether a subject line worked. Merge until the numbers mean something.
• Overlapping sends. A VIP who is also lapsed and also interested in knitwear can receive three emails in one week. Decide which segment wins and exclude the others, and check how often to send.
• Discounting your best customers. Repeat buyers and VIPs often buy at full price. Give them access and attention before a discount.
• Rules that never update. “Bought in spring” goes stale. Use rolling windows such as “in the last 90 days” so the segment refreshes itself.
• Forgetting consent. A segment can only contain people who agreed to marketing email. Unsubscribes and complaints stay on your suppression list whatever the rule says.

For how segments fit into a full month of email, see ecommerce email marketing; for turning a segmented campaign into orders, see how to get more orders from email.

FAQ#