One Setting Changed
Our team publishes in four languages: Korean, Traditional Chinese, Simplified Chinese, and English. Until recently, every scheduled post went out at 9 AM Korea time. We figured that was a nice commute-hour slot for everyone.
A Strange Signal
When we asked a bot to sum up two weeks of read logs, only the Traditional Chinese articles looked unusually quiet. Korean posts got their reads in the morning, while the Traditional Chinese ones crept up only in the evening.
The Cause Was Time Zones
Taiwan is one hour behind Korea. 9 AM in Korea is 8 AM in Taiwan, before most people start work, so the notification slid past unread. By the time readers opened the app in the evening, the post had sunk below the fold. We had translated the words but never the publishing time.
What We Changed
Set the publishing time separately for each language.
We added one line per language to the schedule file, and wrote the time zone name next to it, so the posts go out at the same moment whether the job runs on the Mac mini or the Windows device.
timezone: Asia/Seoul
publish_time:
ko: 09:00
zh: 21:00
cn: 21:00
en: 09:00
We left English alone on purpose. Change one thing at a time, and the reason behind the result stays visible.
Results After Two Weeks
| Language | Avg. reads (before) | Avg. reads (after) |
|---|---|---|
| Korean | no change | no change |
| Traditional Chinese | 38 | 71 |
Simplified Chinese rose by a similar margin, while the Korean and English posts we left untouched stayed almost the same.
Today's Notes
1. When you translate a post, translate its publishing time too.
2. Change only one thing per experiment.
3. Always write the time zone name in the schedule.
A single line of configuration was reaching readers before the writing itself.