Segment Refresh Cadence Through the Seasons
Segments built on behaviour expire as seasons turn, because the behaviour that defined them was seasonal. Refresh cadence should match how the segment is used: always on lifecycle segments update continuously, campaign segments get rebuilt against current season data, and seasonal audiences from last year should be re qualified, never simply reused.
Last year's holiday buyers are not this year's
The most seductive segment in any database is last year's seasonal audience: big, named, and one click away. It is also a trap. That list mixes people who bought gifts once for recipients they no longer shop for, deal seekers loyal to the discount and not the brand, and a core of genuine repeat seasonal customers. Mailing them all as one segment averages those groups into mediocrity and trains your deliverability on the uninterested majority.
For more on this, see this related guide and our earlier explainer on the same theme.
Re qualify instead. Intersect last year's seasonal buyers with any engagement or purchase since, and split the segment: the still active get the campaign, the quiet get a lighter reactivation treatment, and the silent get left alone. The segment shrinks and performs, which is the correct trade in every season.
Continuous, calendar, and campaign refreshes
Not every segment refreshes the same way. Lifecycle segments, new, active, lapsing, churned, should update continuously, because their job is to reflect the present. Behavioural campaign segments, category intenders, price watchers, need rebuilding against a rolling window matched to the buying cycle: thirty days for fast categories, a season for slow ones.
Calendar segments exist too: back to school shoppers, holiday gift buyers, January fitness starters. These should be rebuilt fresh each cycle from current signals, with last year's version used as a seed and a benchmark, never as the send list. Behaviour announces itself anew every season; the marketer's job is to listen in the current one.
Windows that match the buying cycle
The refresh window question is really a memory question: how far back does behaviour still predict? For consumables with a six week cycle, ninety days of history contains two full cycles and predicts well. For annual purchases, a ninety day window misses almost everyone's last purchase, and the segment needs a year of memory filtered by recency of engagement.
Readers also found this companion piece from our archive useful.
Set windows per category and review them when results drift. A segment whose performance decays steadily through a season is telling you its window or its definition is stale. Treat that decay as a maintenance alarm rather than a creative failure, because no subject line fixes an audience built from expired behaviour.
Hygiene rituals that keep segments honest
Give every segment an owner, a definition in plain language, a refresh rule, and an expiry date displayed wherever the segment is used. Segments without visible expiry get reused past their life by whoever finds them in the tool, and the resulting campaign gets blamed on the channel. Most stale segment disasters are governance failures, not data failures.
Prune ruthlessly at season boundaries. Archive last season's campaign segments so they cannot be accidentally mailed, and start the new season's set from current data. A shorter list of living segments, each with a known refresh rhythm, outperforms a museum of audiences nobody trusts but everybody fears deleting.
Frequently asked questions
Can we reuse last year's holiday segment?
Not as a send list. Use it as a seed: intersect it with engagement or purchases since then, split the active core from the quiet, and treat each group differently. The one click reuse mails gift only buyers and discount loyalists as if nothing changed.
How often should segments refresh?
Lifecycle segments continuously, behavioural campaign segments on rolling windows matched to the buying cycle, and seasonal segments rebuilt fresh each cycle from current signals. One cadence for all segments guarantees some are always stale.
How long a behavioural window should a segment use?
Long enough to contain at least one full buying cycle: about ninety days for six week consumables, up to a year for annual purchases, filtered by engagement recency so old behaviour does not outweigh current disinterest.
Who should own segment maintenance?
A named owner per segment family, with definitions, refresh rules, and visible expiry dates wherever segments are used. Governance, not data, is what usually fails when a stale segment gets mailed.