How to Build Customer Segments in Shopify With AI

Customer segments are one of the most valuable features in Shopify, and one of the least used. The reason is the editor. Building a segment means writing a filter query, and "customers who ordered at least twice but not in the last 90 days" isn't obvious to express if you've never done it.

Now you can just say it. Shopify's segment guide lists Sidekick as one of the ways to create a segment: you describe the group in plain language and it generates the segment. Here's how to use it well, and how to make sure the segment holds the people you meant.

How it works

1. In your Shopify admin, go to Customers, then Segments.
2. Open Sidekick and describe the customers you want.
3. Sidekick writes the segment's filter query into the editor.
4. Run it to see the matching customers and how many there are.
5. Review the query and the description, then Save segment.

Shopify generates a short description of the segment. Its guide says to review that description and change it if it doesn't match the segment you want. Treat that as the first check: if the plain-language description isn't what you asked for, the query isn't either.

Prompts that work

A segment prompt describes people by what they did, where they are, or how they relate to your marketing. Shopify's own examples include:

• "Canadian customers who subscribe to my marketing emails"
• "Customers whose last order was more than 90 days ago"

Building on that pattern:

By purchase behaviour

• "Customers who have placed more than three orders."
• "Customers who have spent more than 500 dollars in total."
• "Customers who placed their first order in the last 30 days."
• "Customers who have ordered exactly once."

By recency

• "Customers who ordered at least twice, but not in the last 120 days."
• "Customers who haven't ordered in the last year."

By location and subscription

• "Email subscribers in the United Kingdom."
• "Customers in California who aren't subscribed to email."

By product

• "Customers who bought anything from the Skincare collection."
• "Customers who bought [product name] in the last six months."

Use exact names for collections, products, and tags, and real numbers for thresholds. "Good customers" isn't a segment. "More than three orders and over 300 dollars spent" is.

Check the segment before you use it

The danger with a segment isn't that it fails. It's that it works, returns a list, and the list is slightly wrong. An email to the wrong 2,000 people can't be unsent.

Look at the count. Does it pass a sanity check? If you have 10,000 customers and "repeat buyers" comes back as 9,400, something's off.

Open three customers. Pick three from the list and look at their order history. Do they obviously belong?

Find someone who should be in it. Think of a customer you know fits. Are they there? And someone who shouldn't be?

Read the query. You don't have to write the filter language to read it. The conditions are usually legible: a field, a comparison, a value. Check the numbers and whether conditions are joined with AND or OR. Those two little words change everything. "Bought twice AND not recently" is a win-back list. "Bought twice OR not recently" is nearly everybody.

Watch the date logic. "In the last 90 days" and "more than 90 days ago" are opposites, and they're the most common mix-up.

If it's wrong, tell Sidekick what's wrong in plain words: "This includes people who ordered last week. I only want customers whose most recent order is older than 120 days."

Five segments worth building first

1. Win-back. Ordered at least twice, nothing in the last 120 days. People who liked you and drifted.
2. One-and-done. Exactly one order, more than 60 days ago. The second purchase is the hard one, so target it.
3. VIPs. Top spenders or most frequent buyers. For early access and thank-yous, not discounts.
4. New customers. First order in the last 30 days. For onboarding and how-to content.
5. Subscribed but never bought. On your email list, zero orders. They need a reason to start.

Segments stay current. A customer who places a third order joins the repeat-buyer segment without you doing anything.

What to do with a segment

A saved segment appears in your Segments list, ready to use. The two common uses:

• Email. Send a campaign to a segment instead of your whole list.
• Discounts. Limit a discount code so only customers in a segment can use it.

Those two combine well: a win-back segment, a code only that segment can redeem, and an email announcing it. For the code itself, see creating discount codes with AI.

Where Claude and ChatGPT fit

If you use a connected Claude or ChatGPT, know the boundary. Shopify's documentation for the official integrations lists reading customers (name, email, phone, order count, total spend) and says customer records are read-only. It doesn't list an action for creating segments.

So the split is:

• Sidekick or the segment editor to build and save the segment.
• A connected assistant to think about strategy first: "Here's my customer breakdown from last quarter. Which three segments would you build, and what would you send each?"

Remember that asking an outside assistant about customers shares personal data with that provider. See is it safe to connect AI to your Shopify store?

Other ways to make a segment

Sidekick isn't the only shortcut. Shopify's guide also covers:

• Templates. On the Customers page, the Templates option offers pre-built segments by category. Pick one and its filters fill the editor.
• Copying an existing segment. Open one, change it, and save it as new.
• Writing the query yourself, once you've read a few and the pattern makes sense.

A good way to learn: have Sidekick build a segment, read the query it wrote, and adjust one number by hand.

The short version

1. Go to Customers, then Segments, and describe the group to Sidekick.
2. Use exact names and real numbers.
3. Run it. Check the count, three customers, the description, and the AND/OR logic.
4. Save it, then use it for a targeted email or a restricted discount.

For more on what Sidekick can do in the admin, see how to use Shopify Sidekick. For the full set of AI options, see every option compared.