The Klaviyo Segmentation Playbook

Why isn't one Klaviyo segment enough?

One segment is a starting filter, not a strategy. A single engaged audience can mix recent intent with profiles that are drifting away. The canonical ZHS model uses four mutually exclusive operating tiers so reliable recent engagers can receive relevant opportunities while dormant profiles enter reactivation instead of every newsletter.

Start an audit with a simple question: does every campaign default to the same engaged segment? If it does, you cannot tell whether reduced response is a creative issue, a cadence issue, or an audience-quality issue.

Mailbox placement is not a simple penalty applied to every unopened email; providers use many signals and do not publish a complete scoring formula. The defensible operating move is to use engaged audiences, watch audience quality and complaint trends, and investigate changes before widening reach. Klaviyo's guidance on engaged segments and reputation repair supports that direction.

Mailing too wide can create both cost and deliverability risk. That is why the audience has to be intentional, not just large. See how a big list can create an engagement trap.

The ZHS model gives your most engaged subscribers the first relevant opportunity, reserves broader sends for an intentional decision, and routes cold subscribers to a win-back sequence instead of the regular newsletter.

Below is the exact structure we build.

Here is the video version.

---

What engagement tiers should you build in Klaviyo?

Build four mutually exclusive tiers from rolling engagement windows: highly engaged, warm, cool, and dormant. A thirty-, sixty-, and ninety-day model is a ZHS starting default, not universal law. Each tier gets a different cadence and the segments update as profiles move in and out.

Every send decision starts here.

Build these as saved segments in Klaviyo using opened or clicked within a rolling window. The segments update themselves as people move in and out. If you want the deeper logic behind these, see how engagement tiers protect deliverability.

Segment	Definition	Who gets what
Highly engaged	Reliable engagement in the last 30 days	Up to every relevant campaign
Warm	Reliable engagement 31 to 60 days ago; excluded from highly engaged	Selected strong campaigns
Cool	Reliable engagement 61 to 90 days ago; excluded from higher tiers	Highest-value moments only
Dormant	No reliable engagement in 90+ days	Reactivation, then suppress if still inactive

Think of these windows as a dial, not a fixed setting.

When response quality softens, test a tighter window for comparable sends. When it stays healthy and you need more reach, test a modest expansion. Keep the audience definition, offer, and campaign purpose visible in the comparison; a bigger audience may change the response mix, but it does not guarantee more revenue.

Build a small set of candidate engaged segments aligned with the purchase cycle, then compare their delivery, complaint, click, conversion, and revenue-per-recipient trends. Do not assume the same window works for every catalog or acquisition source.

The numbers to use. Treat open rate as a directional trend because privacy features can inflate it. Use the canonical click-rate bands for diagnosis, then judge placed-order rate and total attributed revenue against comparable sends from your own program. Product category, price, repeat cycle, offer, audience, attribution window, and campaign purpose make a universal vertical conversion table misleading.

The goal is steady improvement across comparable cohorts, not beating a blended ecommerce average whose denominator and audience you cannot inspect.

Watch the negative metrics too. Use the canonical unsubscribe and complaint bands as directional guardrails, then investigate bounce patterns in their sending and acquisition context. Review the trend on a regular cadence and before any material expansion.

---

What buyer segments should you layer on top?

Split by purchase history into four practical groups: never purchased, one-time buyers, repeat customers, and VIPs. Layer these on top of the engagement tiers so one subscriber sits in both. Treat discount eligibility as a deliberate economics decision, not a universal rule based on purchase count alone.

Engagement tells you who is paying attention.

Purchase history tells you how to talk to them. We layer these on top of the engagement tiers so a single subscriber sits in both.

The buyer split changes the decision you are asking the customer to make. In abandoned-cart and checkout messages, make discount eligibility a documented test: compare margin, conversion, repeat behavior, and customer treatment before applying a rule across the program.

---

What behavioral and predictive segments matter most?

Three useful layers are AOV tiers that split high from low spenders, category affinity based on what people click or buy, and churn risk for buyers past their reorder window. These sit on top of engagement and purchase history so each campaign can start from a more relevant audience.

Engagement and purchase count are the base.

These behavioral segments reduce manual audience rebuilding, provided the events and properties feeding them are accurate.

1. 01
   AOV tiersHigh vs low spenders

   Split by average order value. Premium buyers get premium products and bundles. Value shoppers get entry offers.

2. 02
   Category affinityClicked or bought a category

   If someone only clicks or buys your body care, show them more of that. The email reads as personal because it is.

3. 03
   Churn riskDue to reorder but gone quiet

   Past their typical reorder window with no recent purchase. This is your win-back and replenishment audience.

Structured from the canonical article steps for responsive reading and presenter mode.

Category affinity is where personalization jumps a level.

Segment on click and buy behavior. Then send that subscriber more of what they already told you they want.

---

Should you segment on site behavior, not just email?

Yes. Opens and clicks miss the buyers who ignore email but still browse the site. Build a second dial, a site-engaged segment for anyone who viewed a product, added to cart, or started checkout in the last X days. Now you have two dials, email engaged and site engaged, and can tune each on its own.

Opens and clicks are not the only signal.

Plenty of buyers ignore your emails but still visit the site. If you only segment on opens and clicks, you miss them.

So we build a second dial: the site-engaged segment. The person can receive marketing because they subscribed, AND they did something on site, viewed a product, added to cart, or started checkout, in the last X days. That captures the whole funnel from homepage to checkout.

Now you have two dials instead of one. Email engaged and site engaged. If one is dragging your metrics, tighten that window on its own and leave the other alone. Run the engagement report on each segment to see open rate, click rate, AOV, and profile age. That is how you stop guessing and start operating.

Go one level deeper with product-engaged segments. Same idea, scoped to a product or category. Someone who viewed, added to cart, or started checkout with a specific product in the last X days. Match the creative to exactly what they browsed.

Then cross-purchase patterns. Look for the next products that customers actually buy together in your own order data. A skincare brand might compare people who bought serum but skipped toner with people who bought the starter kit and did not return for a refill. Treat the pattern as a testable audience hypothesis, not a personalization claim you have already proved.

---

What is zero party data and how do you segment on it?

Zero-party data is information people volunteer about themselves, often through a popup: what they are shopping for, their skin type, or whether they are wholesale or retail. Build segments only for answers that will change the follow-up, and make the value exchange and consent clear.

Zero party data is information people volunteer about themselves.

It often comes from your popup. What are you shopping for? What is your skin type? Wholesale or retail? Build a segment per answer choice only when it maps to a useful content or product decision. A quiz funnel that captures zero-party data can feed these segments at signup.

Use the canonical popup ranges as a directional operating comparison. Match the denominator, completed signups divided by eligible visitors shown the form, and compare by device, traffic source, offer, and consent step before calling a format better.

---

How do dynamic segments drive every send?

A dynamic segment updates its membership as its qualifying conditions change, so no campaign has to start as "send to everyone." A new product drop might start with highly engaged profiles plus category affinity; a sitewide sale may use a broader audience; a restock can use purchase history and product interest. The exact routing should follow the offer, margin, and consent context.

The point of building these once is that they never go stale.

A dynamic segment recalculates its members as qualifying behavior changes. Someone who shows a reliable engagement signal can move into a higher tier; someone who falls beyond the tier threshold moves toward dormant treatment.

So a campaign is never "send to everyone."

A new-product drop goes to highly engaged plus category affinity. A sitewide sale goes to fading and dormant, where the discount does the heavy lifting. A full-price restock skips the discount and goes to repeat and VIP.

You pick the segment, and the segment is already accurate.

---

Who should you exclude from every campaign?

Exclude categories based on the product cycle, consent, engagement policy, and bounce context. Common candidates are recent purchasers, persistently unengaged profiles, and profiles with repeated delivery failures. Segmentation is about who you leave out as much as who you include.

Segmentation is not only about who you include. It is also about who you leave out.

Common campaign exclusions are:

1. Recent purchasers. Set the window to the actual product and replenishment cycle; a recent buyer should not receive a message that conflicts with the order experience.
2. Persistently unengaged profiles. Define this with reliable engagement signals, a documented reactivation attempt, and your consent policy, not a copied static threshold.
3. Repeated delivery failures. Follow the platform's suppression guidance and investigate the acquisition source or sending context before treating every bounce alike.

1. 01
   Start with the includeEngagement tier + buyer layer

   Pick the engagement tier the campaign is for, then layer the buyer segment on top. That is your starting audience.

2. 02
   Check the product cycleRecent purchasers

   Exclude people who just ordered when the campaign would conflict with their product or order experience. Set the timing from the actual purchase cycle.

3. 03
   Use documented policyPersistently unengaged

   Route profiles through reactivation and consent-aware policy before excluding them from regular sends.

4. 04
   Investigate delivery contextRepeated delivery failures

   Follow platform suppression guidance and inspect the acquisition source and sending context before the next campaign.

Structured from the canonical article steps for responsive reading and presenter mode.

One more gray area deserves a compliance review: explicit versus implied consent. Use the consent status that your platform and jurisdiction actually allow, document the basis for each audience, and do not turn a promotional deadline into an exception to that policy.

---

How often should you email each segment?

There is no defensible universal send count. Match cadence to the tier, the purchase cycle, the offer calendar, and evidence from your own comparable sends. The most engaged audience may be eligible for more relevant opportunities; fading and dormant audiences should receive a more selective treatment.

Once the segments exist, decide how often each one hears from you with an explicit operating rule:

• Highly engaged. Start with the campaigns that are genuinely relevant to their recent behavior. Watch conversion, complaints, unsubscribes, and total attributed revenue as volume changes.
• Warm and cool. Reserve the strongest messages, offers, or category-specific content. Test an expansion only against a documented comparable baseline.
• Dormant. Use reactivation or a deliberate high-value moment, not the default campaign calendar. Suppress or reduce exposure when the policy says the profile is no longer safely marketable.

For total campaign volume, start from the real work you can produce and approve at a high standard. A larger calendar is not a strategy if the audience definition, creative quality, or QA cannot keep up.

Use a controlled cadence test rather than a copied volume target: hold the audience definition and offer class as steady as possible, add or remove a planned send, and compare downstream business metrics with complaint and unsubscribe trends. The right outcome may be more sends, fewer sends, or a different audience split.

More volume should not mean mailing the same people indiscriminately. Use segmentation to create room for narrower, tailored sends while keeping recipient exposure and negative signals visible.

Useful campaign treatments to test deliberately. A resend to a non-opener can be appropriate for a major moment when the message, audience, and deliverability context support it. A follow-up to people who clicked but did not purchase can be useful when the next message adds real information or a warranted offer. Neither is automatic incremental revenue; evaluate each treatment against a comparable baseline.

---

How often should you clean your Klaviyo list?

Review list hygiene on a regular operating cadence and before a material billing or sending decision. Suppressing profiles can reduce active-profile costs depending on the plan and can protect the usable audience, but it should follow a documented reactivation and consent policy rather than a copied countdown.

Suppressing subscribers can feel like throwing away revenue. The appropriate question is whether a profile has a compliant, evidence-based path back to engagement. If not, keeping it on every campaign can dilute reporting, increase cost, and create deliverability risk.

Define the audiences that need a review in your own policy. Common examples include:

1. Profiles with no reliable engagement after an appropriate reactivation attempt.
2. Profiles inactive beyond a purchase-cycle-aware window.
3. Profiles that have not shown a reliable click, purchase, or qualifying site signal under the documented policy.

For each audience, document the eligibility rule, the reactivation treatment, the consent basis, and the suppression decision. Review the process regularly and adjust it if the acquisition source, purchase cycle, or delivery evidence changes.

---

What are the most common Klaviyo segmentation mistakes?

Common mistakes include sending every campaign to one broad engaged segment, applying blanket discounts, never defining a cold-profile policy, ignoring category behavior, relying on stale static lists, and treating first-time and repeat buyers as the same audience. Each is an avoidable loss of audience context.

1. Using one broad engaged segment for every campaign. Split into tiers and match exposure to reliable behavior.
2. Applying blanket discounts. Decide eligibility from margin, customer history, and a documented hypothesis.
3. Never defining a cold-profile policy. Route dormant profiles through reactivation and sunset rules, then suppress when the documented policy requires it.
4. Ignoring category behavior. Use click and purchase data to test more relevant content, not just more offers.
5. Building static lists. Use dynamic segments when the events and properties behind them are trustworthy.
6. Treating first-time and repeat buyers the same. The next action may be different, so the audience treatment should be deliberate.

---

Get Expert Help

Our team builds and maintains segment structures inside Klaviyo for DTC teams. If your account relies on one broad segment for every send, we can map engagement tiers and buyer splits that fit your catalog, consent model, and reporting.

We do not promise a percentage lift from segmentation alone. Results depend on traffic quality, offer, creative, purchase cycle, data quality, attribution, and the rest of the retention program. The same audience-design principles can inform other platforms, but their data models and controls need to be verified before the logic is copied.

See our pricing | Apply to work with us