E-commerce Segmentation: The Household-Level Playbook

Why most e-commerce segmentation under-performs

Open any D2C Klaviyo account and you'll find the same dozen segments: engaged 30 days, lapsed 90 days, VIP by spend, abandoned cart, browsed but didn't buy. They're useful. They're also generic - every brand running Klaviyo has them, and they describe what someone did with you, not who that person is.

The gap between "what they did" and "who they are" is where most segmentation work plateaus. A 90-day lapser who's an affluent owner-occupier mid-move is a completely different commercial bet from a 90-day lapser who was a one-off discount shopper. The behavioural data treats them identically.

The five types of e-commerce segmentation

1. RFM segmentation. Recency, frequency, monetary value. The classical view. Strong for prioritising retention spend, weak for predicting why behaviour changes.
2. Behavioural segmentation. Browsed, abandoned, bought, opened, clicked. Powers triggered flows. Doesn't tell you context.
3. Lifecycle segmentation. New, active, lapsed, VIP, win-back. Useful for campaign cadence and creative tone.
4. Predictive segmentation. Klaviyo's own churn score, predicted CLV, predicted next-order date. Better than nothing; brittle when context changes.
5. Household-level segmentation. Tenure, property type, family stage, purchasing power, move stage. The layer most brands don't have, and the one that turns segmentation from sorting into prediction.

Where household-level segmentation changes the picture

1. Suppression is finally accurate

Most "VIP" segments quietly include the same household twice (work email + personal email). Household-level suppression deduplicates at the address, not the inbox. Recovers 5-15% of wasted prospecting spend on day one.

2. AOV bands match wallet, not behaviour

A customer who placed two £40 orders looks identical to one who placed two £400 orders, by RFM. Their Purchasing power band doesn't. Premium categories see disproportionate lift here.

3. Replenishment timing matches life stage

Bigger Families replenish food, cleaning and consumables on a different rhythm to single-occupant households. Family-stage-aware flows beat purchase-recency-only flows in any category with a replenishment cycle.

4. Move stage is the highest-intent signal in home categories

Furniture, sleep, garden, kitchen, appliances - any home-anchored category sees a step-change in conversion from a Just Moved or Predicted Move segment. (Background here.)

5. Tenure unlocks owner-only categories

Solar, EV charging, kitchens, large garden products only make sense to owners. Renting-out as a suppression rule on these campaigns saves the impressions for buyers who can actually convert.

How many segments should an e-commerce brand have?

Most well-run D2C accounts land at 8-15 active segments. Past 15, you're usually duplicating segments instead of learning anything new. The right test isn't count, it's: does removing this segment change anything in flows or campaign send? If no, retire it.

How to segment by household income or property in Klaviyo

Klaviyo doesn't capture this natively. The pattern is:

1. Enrich Klaviyo profiles with household attributes (Family stage, Property type, Bedrooms, Purchasing power band, Move stage) via an external enrichment layer.
2. Attributes appear as standard custom properties on each profile.
3. Build segments against the custom properties using Klaviyo's standard segment builder - no schema changes, no developer.
4. Use the new segments in flows, campaigns, suppression, and as the source for Meta/Google Custom Audiences.

(Step-by-step setup here.)

Three high-leverage segments to build first

1. Affluent recent movers. Purchasing power band ≥ band 4 AND Move stage in (Just moved, Recently moved). Lowest-CPA prospecting target for any home-anchored brand.
2. Bigger families in 3+ bed properties. Family stage = Bigger Families AND Bedrooms ≥ 3. The right seed for replenishment and family-product flows.
3. VIP buyers in top purchasing power band. Existing top-decile customers AND Purchasing power band = highest. The seed worth pushing to Meta as the Custom Audience for prospecting.

These three on their own usually justify the enrichment investment in the first quarter.

How to measure whether household segmentation is working

• Revenue per recipient on flows that branch on household context vs the unbranched version of the same flow.
• CPA on prospecting Custom Audiences seeded from household segments vs lookalikes off the same source list.
• Suppression recovery: wasted CPMs against existing customers before and after household-level suppression.

One full purchase cycle for the category. Matched control where you can. Don't read day-three CPA.

The bottom line

RFM and behavioural segments tell you what people did. Household-level segments tell you who they are - and that's the variable doing most of the work in flow revenue and prospecting CPA. The brands getting outsized returns from Klaviyo and Meta in 2026 aren't running more segments. They're running the same segments, plus a layer of context the stack didn't capture.

Read next

• The Klaviyo segmentation guide for D2C brands
• Five Klaviyo flows you can't build without household data
• The customer data enrichment pillar guide