A customer who bought an expensive dining table six months ago may have no reason to buy another one. Sending a “we miss you” discount treats a durable purchase like a replenishment subscription. For high-average-order-value businesses, email segmentation needs a defensible answer to a narrower question: what evidence makes this particular message appropriate now?
This guide turns that question into a reviewable workflow in Migma, with audience evidence supplied by a commerce or lifecycle platform such as Klaviyo. It is published by Migma. The proposed rules and examples are editorial recommendations, not measured performance results or universal timing benchmarks.
Write the message decision before building the segment
Start in Migma with a short creative brief for one message job: help an existing customer evaluate a relevant accessory, invite an interested prospect to a consultation, or explain how to care for a purchase. Do not start with a vague audience label such as “high value.”
Each brief should answer four questions:
- What useful action can the reader take?
- What observed fact makes the message relevant?
- What fact would make it inappropriate?
- Who approves the evidence and the final email?
For a hypothetical furniture brand, an accessory message might require an eligible past table purchase and a currently available matching care kit. An unresolved delivery complaint would exclude the customer. Those conditions are proposed business rules; the team must check that its data and audience platform can actually implement them.
Use a decision table that includes silence
A segment is useful only if it changes a decision. Build a small table before producing variants. Keep “no promotional message” as a valid outcome rather than forcing every person into a campaign.
| Evidence available | Suggested message job | Check before release |
|---|---|---|
| Recent purchase and confirmed delivery | Help the customer use or care for the product | Product match and support exclusions |
| Older purchase plus explicit interest in an accessory | Explain compatibility and answer practical questions | Interest freshness and accessory availability |
| Older purchase with no recent evidence | Consider a low-frequency editorial update or no send | Permission, contact history, and relevance |
| Unresolved service issue | Route to the responsible team | Prevent a conflicting promotional promise |
This is a planning example, not a ready-made Klaviyo segment. Your event names, integrations, consent fields, and suppression logic determine which rows can be implemented reliably.
Check whether the data can support the decision
Ask the data owner for a few representative records rather than a dashboard screenshot alone. Can the team distinguish a completed order from a cancelled order? Does a product identifier map to the right category? Are event timestamps recorded consistently? Could a delayed event make someone appear inactive?
Look at both eligible and excluded records. A segment that includes the intended buyer but fails to exclude a person with a refund is unfinished. Keep a written explanation of why each sample should or should not qualify.
Purchase gaps also require care. If you calculate time to a second purchase only among repeat buyers, you exclude customers who have not returned yet. That can make the apparent cycle shorter than the experience of the full customer group. Label the population used and involve an analyst before turning that estimate into an automatic win-back threshold.
Klaviyo's September 4 luxury segmentation article provides a useful example of why a brand's purchase cadence can invalidate a generic inactivity rule. Its reported experience belongs to that business. Use it as a reason to investigate your own records, not as a timing prescription.
Turn approved evidence into an email brief
Migma's persona setting supplies context for drafting. It does not select the delivery audience. Keep those two decisions separate in the handoff: the creative reviewer approves the wording; the audience owner approves who qualifies.
A workable brief for the furniture example could read:
Create a calm care-guide email for existing table owners. Explain the approved maintenance steps and link to the matching guide. Do not imply we know how the customer uses the table. Do not add a discount, a replacement deadline, or a personal recommendation unsupported by the supplied facts.
The example is intentionally specific about forbidden assumptions. A customer history can support a product reference without supporting claims about the customer's home, plans, or preferences.
Prepare a generic fallback if the product field is missing. Review the fallback as a complete email; removing one personalized sentence should not leave a broken transition or an unexplained button.
Review the delivered version, including the wrong-recipient case
Run Email Preflight after the copy and layout are stable. It checks inbox previews, links, writing, and common spam or delivery risks. Fix the findings and rerun the checks. Preflight cannot guarantee inbox placement or establish that an audience rule is correct.
If the campaign runs in Klaviyo, export the reviewed email and inspect the destination template. Migma documents HTML export and an optional beta adaptation for the Klaviyo editor. The final platform can alter wrappers or variables, so send a destination-platform test as well.
Use a compact release test set: a matching product, a missing product field, a recently refunded customer, an opted-out contact, and a customer who bought after the segment was initially assembled. The last three should test the audience rules rather than receive the promotion. Verify the final recipient selection at release time.
Measure usefulness without declaring every purchase a win
Choose the outcome before launch. A care guide may aim to increase use of support content; a consultation invitation may aim to generate qualified bookings. Sales can matter without being the only useful action.
Where volume permits, compare eligible recipients with a suitably assigned holdout and agree the observation window in advance. Keep offer, audience, and timing changes separate enough to interpret. With a small high-value audience, a few orders can move a rate sharply; report counts and uncertainty alongside percentages.
Also review unsubscribes, complaints, and service feedback. If the campaign produces confusion, revisit the decision table before generating more variants. Migma can help prepare the next reviewed message once the team has established what the evidence actually supports.
Frequently Asked Questions
When should a high-AOV brand send a win-back email?
When its own customer evidence supports a relevant reason to return. A long gap alone is insufficient for durable or infrequently purchased products. Define eligibility, exclusions, and a useful message job before choosing a schedule.
Does Migma choose the customers in a Klaviyo segment?
The workflow here uses Migma for drafting, review, Preflight, and export. The audience owner implements and verifies the segment in the sending platform. Selecting a Migma persona shapes the draft; it does not establish delivery eligibility.
What if there are too few repeat purchases to estimate a cycle?
Do not treat a small average as a reliable automatic trigger. Start with explainable product or preference evidence, inspect individual cases, and use a limited pilot. Record what remains unknown before expanding the audience.
Should every high-value segment get a different discount?
No. First test whether it needs a different message at all. Care instructions, compatibility information, or a consultation may be more relevant. Any offer should have approved terms and should not be introduced automatically from an audience label.
Why test again after exporting the email?
The destination platform can add wrappers, interpret variables differently, or apply its own sending rules. Open the exported template, check fallback values and links, and send a real test from the final platform before release.
