A practical business guide

How to schedule business automations across time zones

Set a clear local time promise and test how your scheduler handles seasonal clock changes and delayed runs.

Download the worksheet ↧

Choose what stays fixed

Choose the time rule before setting up the trigger. Use a named time zone when people expect a job at the same local time, such as a report every day at 8 a.m. in New York. Use a fixed time in Coordinated Universal Time (UTC) when several systems must use the same instant. Use an elapsed-time rule for work that needs a fixed gap between runs. These rules can produce different results when local clocks change.

Write the rule in one sentence: “Publish the daily exception digest once for each New York business date at 8 a.m. local time, with an alert by 8:20 a.m. if it has not arrived.” Define any weekday or holiday rule separately. Use a location identifier such as America/New_York. An offset such as -05:00 describes one offset from UTC and does not carry New York’s seasonal rules.

Set the clock for the trigger and the work

Find where the scheduler evaluates the expression. A workflow can run on a host in UTC while the trigger uses a local time zone. Check if the zone belongs to the trigger, workflow, account, or scheduler itself. Save the zone and intended run time with the configuration. Inspect previewed runs across a clock change, or use a disposable trigger with a safe test destination if there is no preview.

Pass the intended scheduled time into the job when the scheduler supports it. Base the business date and data window on that time in the chosen business zone. A delayed run at 12:05 a.m. should not silently switch to the next file or report. For a calendar-day data window, convert that local date’s midnight and the next local midnight into two UTC instants. Use the first inclusive and the second exclusive. A civil day can have 23 or 25 hours where the clocks change; adding 24 hours can leave out records or include records from the next day.

Decide what should happen in a missing or repeated hour

Some local times do not exist on the spring change date. Some occur twice on the fall change date. If a job must run once per business date, avoid a sensitive time when you can. If you must use one, record whether to skip, shift, or run it in a later recovery window on the spring date. For fall, record which occurrence should run or how you will accept either and suppress the other.

For example, Amazon EventBridge Scheduler documents that it skips a cron occurrence in the missing spring hour and runs only once for a repeated fall time. Google Cloud Scheduler explains that local-clock schedules can run unexpectedly around clock changes and recommends UTC when a fixed cadence matters. Check your actual scheduler’s behavior. If it cannot enforce the business rule, add a safeguard in the receiving workflow or choose another time.

Worked example: a daily digest across two time zones

Fictional Harbor Desk sends one exception digest per New York business date at 8 a.m. New York time, seven days a week for this test. A Phoenix employee must be available to receive it. The job’s daily identifier combines the workflow name with the intended New York date. On a replay, the receiver checks this identifier before it sends a second digest.

Expected times for the fictional digest
New York dateNew York localUTCPhoenix local
Mar 13, 202708:0013:0006:00
Mar 14, 202708:0012:0005:00
Nov 6, 202708:0012:0005:00
Nov 7, 202708:0013:0006:00

Harbor Desk cannot promise both 8 a.m. New York and 6 a.m. Phoenix all year with a single daily trigger. It keeps the New York rule and gives the Phoenix operator the seasonal schedule. The team logs the intended New York date, intended UTC time, actual start and the delivered digest ID. If two workers receive the same intended date, they need one shared record of whether the digest was sent. See the duplicate-action guide for ways to make that record safe.

Run this test before release

Use a separate workflow with a fake recipient or a stub. Compare actual starts and completed sends. Test before and after the clock change, even if your normal trigger is outside the affected hour.

SCHEDULE RELEASE CHECK
Business rule:
Trigger type and time zone:
Business zone and data window:
Alert time and owner:

Case | Expected result | Observed result
Ordinary day | One run at documented local time; UTC matches |
Spring change | Local promise stays fixed; day window is correct |
Fall change | Local promise stays fixed; day window is correct |
Missing or repeated hour | Documented skip, shift or deduplication |
Delay past midnight | Run uses the intended business date |
Two deliveries for same date | One recorded business result |
Scheduler restart | No silent loss; operator sees any missing date |
Public holiday or weekday filter | Documented calendar rule holds |

Save the configuration, expected UTC times and observed outputs.
Mark any provider behavior you could not test.

For a scheduled job that goes live, monitor actual completed outputs and check them against the business time promise. Add a separate alert for a missing result. If you need help testing an existing schedule without sending duplicate messages or creating records, start with the Workflow Automation service.

Put the plan to work

Start with one useful change.

Bring the process, the people involved, and a redacted example. A free 30-minute fit call can establish the next step.