How many email segments do we actually need?

Start with five: an engaged core, VIP buyers, one-time buyers inside their reorder window, recently lapsed profiles, and a suppression pool. That handful covers most revenue decisions. Add segments only when each one changes what you send; a segment that gets the same email as its neighbor is a label, not a segment.

Can we segment on open rates?

Not safely on their own. Apple Mail Privacy Protection, rolled out with iOS 15 in September 2021, preloads tracking pixels and inflates opens for Apple Mail users. Weight clicks, site sessions, and purchases instead, and treat opens as a weak positive signal rather than proof a person read anything.

What is a sunset policy in email marketing?

A written rule for retiring silent profiles: for example, no click or site session in 120 days triggers one three-email re-permission series, and continued silence moves the profile out of active sending. It protects deliverability, since Gmail and Yahoo read sustained sends to unengaged recipients as a spam signal.

Resources · Strategy deep-dive

Email segmentation strategy: score the list, then believe it

Q: When do Klaviyo's predictive metrics become available?

Klaviyo unlocks predicted customer lifetime value, churn risk, and expected date of next order once an account has at least 500 customers with orders, 180 days of order history, orders in the last 30 days, and 3+ orders per qualifying customer. Under that bar, hand-built RFM and engagement scores do the same job.

By Maren Holt, Founder & Principal Strategist · Published July 2, 2026 · Updated July 26, 2026

Email segmentation strategy fails in one of two ways: no segments, so everyone gets everything, or thirty segments nobody maintains. The working version sits in between. Score every profile on behavior, split the list into a handful of tiers that each change what you send, and retire the silent. This guide walks through the model we build for clients, with the benchmarks that justify each step.

Why does segmentation matter more than volume?

Because the revenue distribution across a list is brutally uneven. Klaviyo's cross-account data shows automated flows, which are targeting at its narrowest, producing about 41% of email revenue from 5.3% of sends, with a placed-order rate 13 times higher than campaigns, per Klaviyo's benchmark reports. Behavior-matched mail simply performs at a different level than broadcast.

The other side of the ledger is cost. In a recent audit of ours, 64% of a meal-kit brand's sends went to profiles silent for 120+ days. That volume produced complaints, not orders, and complaint rates carry hard limits now: Gmail enforces at 0.3% and recommends staying under 0.1%, per Google's sender guidelines. The wrong audience is not neutral. It is expensive.

What signals should the score use?

An engagement score ranks every profile on recent behavior. Since Apple's Mail Privacy Protection (iOS 15, September 2021) began preloading tracking pixels, opens overcount badly on Apple Mail audiences, so the score leans on harder signals:

Clicks, weighted by recency: a click this month outweighs five from last year.
Site sessions tied to the profile through Klaviyo or Omnisend onsite tracking.
Purchases: recency, frequency, and value, the classic RFM triad.
Opens as a weak positive only, never as proof of attention.

Klaviyo accounts past a threshold get predictive metrics on top: predicted lifetime value, churn risk, and expected date of next order. Eligibility requires 500+ customers with orders, 180 days of order history, orders in the last 30 days, and 3+ orders per qualifying customer, per Klaviyo's documentation. Below that bar, the hand-built score does the same work.

Which segments earn their keep?

Segment	Definition	What changes
Engaged core	Click or session in the last 30–60 days	Full campaign cadence; the only default audience
VIPs	Top RFM tier, typically 5–10% of buyers	Early access, no discounts, higher frequency tolerated
One-time buyers in window	Single order, inside the reorder cycle	Reorder-timed sequence, not the weekly blast
Recently lapsed	Engagement decay past threshold	Win-back triggered by the score, not a fixed 90-day timer
Suppression pool	Silent 120+ days after re-permission	No regular sends; protects complaint rate

The test for any additional segment: does it change the send? A segment receiving the same email as its neighbor is bookkeeping. Five maintained segments beat twenty abandoned ones in every program we have audited.

How does a sunset policy work in practice?

Pick a silence threshold, ours is usually no click or session in 120 days, and write down what happens at it: profiles get a three-email re-permission series, and continued silence moves them out of active sending. The meal-kit brand that ran this cut send volume 34% with flat revenue, while complaints fell from 0.31% to 0.07%. The full arc is in the audience rebuild write-up.

Two cautions from that engagement. First, suppress before you rebuild: reputation recovers faster when the silent mass leaves early. Second, expect the list-size conversation with stakeholders; the honest answer is that the suppressed profiles were producing complaints and nothing else.

Where does the strategy go from here?

Segmentation is the foundation the rest of a program stands on. Flow timing draws on the same scores, covered in the Lifecycle Flows service, and new-subscriber quality by acquisition source uses the same measurement, covered in the list growth benchmarks guide. If you want the scoring built for you, that is the first two weeks of every MailScale engagement.