Three timing modes
A fixed moment sends to everyone at the same instant. Right for an announcement that must be simultaneous, wrong for a promotion that would reach Sydney at 3 a.m. A local window opens the campaign in each zone when its clock reaches the start hour. A pace spreads the send over time. They combine.
Send now is simply a fixed moment that happens to be the present. A local window has a start and an end in the recipient's own time, for example 10:00 to 18:00: numbers in each zone start when their clock shows 10:00 and stop when it shows 18:00, so one campaign to a global list rolls westwards round the world over the best part of a day. A pace is a ceiling in messages per minute, or a number of hours to spread the list over. Any running campaign can be paused and resumed. More on the scheduling page.
Which mode for which message
| Message | Timing | Why |
|---|---|---|
| Promotion, newsletter, launch | Local window | Lands in daytime everywhere, inside quiet hours |
| Service outage or security notice | Send now | Everyone needs it at once |
| Live event that starts at one instant | Fixed moment, ahead of the start | The instant matters more than the local hour |
| Appointment reminders | A fixed moment per message, set by your system | Each one has its own time |
| Large re-engagement send | Local window and a pace | Protects the team, the site and the list |
| One-time codes | Immediately, through the API, bypassing quiet hours | A code is wanted at any hour |
How zones are resolved
Each number maps to a country from its prefix. For countries with several zones, the zone comes from the area code where the numbering plan allows it, and from the country's most populous zone where it does not. Mobile numbers in the United States are the known weak spot: they follow the area code the line was issued in, not where the person lives now.
The weak spot is easy to plan around. Mainland US time zones span three hours, Eastern to Pacific. A window of 12:00 to 17:00 keeps every recipient inside reasonable hours even if they have moved from coast to coast: a New York number whose owner now lives in Los Angeles receives a 12:00 Eastern message at 09:00, and a Los Angeles number whose owner now lives in New York receives a 17:00 Pacific message at 20:00.
Where mobile numbers carry no geography, the most populous zone decides. Australian mobile numbers all start with 04 and say nothing about where the line is, so a recipient in Perth, two or three hours behind the east coast depending on the season, receives a 10:00 start at 07:00 or 08:00 local time. For such countries, open the window an hour or two later.
Quiet hours inside your window
Each destination has a default marketing window in the recipient's local time, shown on its country page, such as the France page. For example: France 08:00 to 20:00 and not on Sundays, many European countries 08:00 to 21:00, Africa 08:00 to 20:00, Asia 09:00 to 21:00. Quiet hours apply inside your window, so messages go out only while both are open:
| Your window | Destination default | Messages go out |
|---|---|---|
| 08:00 to 22:00 | 08:00 to 20:00, not Sundays (France) | 08:00 to 20:00, Monday to Saturday |
| 10:00 to 18:00 | 08:00 to 21:00 | 10:00 to 18:00 |
| 18:00 to 23:00 | 09:00 to 21:00 | 18:00 to 21:00 |
| 07:00 to 12:00 | 08:00 to 20:00 | 08:00 to 12:00 |
Treat the defaults as the outer limit rather than the target; the window that suits your audience is often narrower. Transactional traffic, such as codes, receipts and alerts, is not marketing and can bypass the window: in the API, set quiet_hours to bypass, as explained in the quiet hours bypass for transactional messages. Promotional content flagged as transactional is treated as abuse under the acceptable use policy. The reasons behind quiet hours, consent and opt-outs are in marketing SMS done right.
Edges of the window
Messages not sent when the window closes wait for the next day's window. A window of 08:00 to 22:00 still stops at 20:00 in France, and a campaign launched on a Sunday reaches French numbers on Monday. That carry-over is harmless for a newsletter and wrong for an offer that ends tonight, so check that the list fits the window before you start; the arithmetic is below.
The other edge is late delivery. A message handed to a carrier at 17:55 for a phone that is switched off is delivered when the phone comes back, within the validity period, 48 hours by default. For content that loses its value, a message sent through the API can carry a shorter validity, in minutes, so it expires instead of arriving after the event.
Daylight saving
Daylight saving is computed in the destination zone at send time, so a campaign scheduled weeks ahead lands at the hour you chose. It matters because clocks do not move together: Europe and North America currently change on different weekends in spring and autumn, the southern hemisphere moves the opposite way, and much of Africa and Asia does not change at all.
A local window absorbs all of this: 10:00 is 10:00 on each recipient's clock, whatever happened in between. A fixed moment does not, because it is one instant and its local hour shifts whenever a destination changes its clocks. A weekly bulletin sent at a fixed 18:00 London time reaches New York at 13:00 for most of the year, and at 14:00 in the weeks when only one of the two has changed. For recurring sends to several countries, a local window is the safer habit.
Adding a pace
A pace is a ceiling, in messages per minute, or a duration to spread the list over. Use it when replies, calls or clicks land on something with finite capacity. Four hundred messages a minute with a 2% reply rate, for example, is eight conversations a minute: know who answers them.
Responses to an SMS tend to come soon after it arrives, so the pace, not the size of the list, sets the load on your team and your site. A pace also buys time to react. Watch the first minutes of a large send, and if the report shows something wrong, such as a broken link or a surge in one failure reason, pause the campaign while you look, then resume.
Pacing maths
Four lines of arithmetic cover most cases. The figures are examples; use your own.
- Duration from pace: list size ÷ pace. 60,000 numbers at 250 a minute take 240 minutes, four hours.
- Pace from duration: list size ÷ minutes available. Spreading 60,000 numbers over six hours means 60,000 ÷ 360, about 167 a minute.
- Pace from capacity: responses your team can handle per minute ÷ response rate. Four agents handling two replies a minute each absorb eight a minute; if earlier sends drew replies from 2% of recipients, the ceiling is 8 ÷ 0.02, or 400 a minute.
- Fit inside a window: window length in minutes × pace. A 10:00 to 18:00 window is 480 minutes; at 100 a minute it carries 48,000 messages. A single-zone list of 60,000 leaves 12,000 for the next day's window, unless you raise the pace to 125 a minute or widen the window.
When the window and the capacity disagree, capacity should win: a campaign that finishes on day two is better than a team that cannot answer on day one.
Scheduling through the API
The same timing is available per message on POST /v1/messages: send_at for a fixed moment, and a window with local_start and local_end for a local window in the recipient's time. Inside the window, quiet_hours set to default keeps the destination's quiet hours, and bypass lets transactional traffic leave at any hour. Add a validity in minutes, up to 4320, for time-bound content, and a campaign tag so scheduled API traffic groups like a panel campaign in the report. The scheduling reference has the details.
Worked examples
- Flash sale, 40 countries: window 10:00 to 18:00 local, default quiet hours, no pace. The sale opens first where the day starts first, in New Zealand and Australia if they are on the list, and last in the Americas. Avoid a Sunday launch if France is on the list, or French numbers hear about it on Monday.
- Appointment reminders: fixed moment per message from your system, transactional flag. The bypass makes any hour possible, not every hour sensible: compute
send_atso the reminder lands in daytime, the evening before an early appointment rather than at dawn. See also appointment reminders. - Re-engagement of 300,000 numbers: window 11:00 to 17:00, pace spread over three days, so the report of day one can fix the list for day two. Three six-hour windows are 1,080 sending minutes, so a ceiling of about 280 a minute does it.
- United States, coast to coast: window 12:00 to 17:00, so recipients who have moved still receive between 09:00 and 20:00 local time.
- Weekly bulletin to Europe and North America: each edition scheduled with a local window opening at 09:00 rather than a fixed hour in your own zone, so the spring and autumn clock changes never move it.
Something unclear or out of date? Tell us on the support page; corrections are folded back into the guide and the update date changes.
