Household Segments: When Three Customers in Your File Share One Kitchen
Your file says three customers. The house says one family: a partner who orders, a partner who browses on the shared tablet, and a teenager with their own account and the family card. Individual level segments send the household three versions of the same campaign, count one basket as three people's intent, and report reach that geography would not recognise.
The signals a household leaves
Shared shipping addresses are the strongest public signal, followed by shared payment instruments, shared devices in analytics, and order patterns that interlock: sizes and categories that never overlap, gift orders between the accounts, support contacts referencing each other's order numbers. Any one signal can mislead, students share addresses, offices share everything, so householding is a confidence exercise, not a merge button.
The value is clearest in frequency and fatigue. Three accounts each under the frequency cap can still put nine messages a week on one kitchen table. The cap the household experiences is the sum, and it is the sum that drives the unsubscribe from the account you most wanted to keep.
What to household, and what to leave alone
Frequency capping, suppression, and reach reporting should work at household level: one winback conversation per home, one catalogue per address, honest deduplicated reach. Personalisation should mostly stay individual: the browsing partner and the ordering partner have different intents, and collapsing them produces recommendations for a person who does not exist, an average of two real people.
Identity and consent never household. An opt out is personal and legal: one member unsubscribing says nothing about another's choices, and merging consent across people is not a data technique, it is a breach with extra steps. Keep consent, account security, and communications preferences strictly per person, whatever the address match says.
Building it without a merge disaster
Start read only: flag probable households with the address and payment signals, report campaign reach and frequency both ways for a quarter, and let the difference argue for the operational changes. Most teams find the deduplicated numbers embarrassing enough to fund the work properly.
When you do act, act with flags, not merges. A household identifier layered over intact individual records can be revised when the student moves out or the couple separates. A hard merge of accounts cannot be unpicked, and life changes faster than databases expect.
The segment nobody builds
Households also explain puzzles elsewhere: the customer who never opens email but buys reliably is often reading the partner's copy; the gift buyer with impossible knowledge of sizes lives with the recipient. Reading data at household level turns some churn and attribution mysteries into ordinary family logistics, which is a relief, because family logistics needs no new software, just messages timed and capped for the home they land in.
Begin with the cap. Of every householding use, summing frequency at the address is the cheapest to implement and the first one customers can feel.
Keep reading
Related guides on this site: Segment Sunset Reviews: Retiring Segments That Stopped Earning Their Keep Dormant Buyer Winback Segments: Splitting Sleepers by Why They Left Practical answers about personalization.
Frequently asked questions
How can a store identify households in its customer file?
By combining signals: shared shipping addresses, shared payment instruments, shared devices, and interlocking order patterns. No single signal is proof, so householding should be treated as a confidence score used for capping and reporting, not an automatic account merge.
What should be managed at household level?
Frequency caps, campaign suppression, catalogue mailings, and reach reporting, where the household experiences the sum of all accounts' messages. Personalisation, consent, and account security should stay strictly individual.
Can consent be shared within a household?
No. An opt in or opt out belongs to the person and the address or number it covers. One household member's choices never authorise marketing to another, regardless of shared addresses or devices.
Why not just merge household accounts into one?
Because households change: people move out, separate, and return, and a hard merge cannot be undone cleanly. A household flag over intact individual records gives the capping and reporting benefits while staying reversible.