2025-09-03 · Jun Han

Send-Time Tests That Respect Korean Inboxes

Global playbooks rarely account for local commute patterns and evening study habits common among education subscribers in Korea. Copying a 9 a.m. US send window can land during Seoul lunch breaks or late-night cram sessions.

We segment send-time tests by cohort behavior, not by headquarters timezone. Trial users often engage after 8 p.m.; enterprise SaaS admins open during mid-morning. Your analytics export should inform buckets before you randomize.

Each test needs a holdout and a minimum two-week window to absorb weekly seasonality. Korean public holidays can skew a single week; we annotate calendars in the readout deck so stakeholders do not misread a dip.

When a winner emerges, document the window in the operator runbook. Send-time wins decay as product usage shifts—plan a quarterly revisit instead of treating the result as permanent.

#testing#localization#retention

← All posts