Scheduling & timing
Good timing is part of good messaging. A pipeline gives you four controls that decide when a message goes out — so you reach people on the right dates, at a reasonable time of day, after a sensible wait, and never too often. This page explains each one in plain language, with practical examples.
The four controls at a glance
Active window
The calendar dates the pipeline runs between — a start date and an end date.
Daily dispatch window
The hours of the day messages are allowed to go out.
Delay cap
How long to wait before the very first send after an item enters the queue.
Frequency cap
The minimum gap before the same recipient is messaged again.
These stack: a message must fall inside the active window, land within the daily dispatch hours, clear the delay cap, and respect the frequency cap before it goes out.
Active window
The active window sets the calendar dates your pipeline runs between — a start date and an end date. Outside those dates, nothing is sent.
By default a new pipeline starts today and ends one year later, so it begins working straight away and keeps running until you decide otherwise. You can adjust either date to plan ahead or to pause:
- Schedule a future launch by setting the start date to a day that hasn't arrived yet. Nothing sends until then, and the pipeline shows as upcoming.
- Pause or stop a pipeline by setting the end date to a date in the past. Sending stops, and the pipeline shows as ended.
The start date is anchored to the very beginning of that day and the end date runs through the very end of its day, so both dates count in full — a pipeline that ends today keeps sending until midnight.
Daily dispatch window
Even on an active date, you probably don't want messages going out at 3 a.m. The daily dispatch window sets the hours of the day when sending is allowed. By default it runs from 09:00 to 21:00 — a sensible daytime range.
If a message becomes due outside those hours, it simply waits and goes out at the start of the next allowed slot. Your customers are never messaged in the middle of the night. Leave both ends open and messages can go out at any hour.
Hours follow your organization's timezone
The dispatch window uses your organization's configured timezone, so "09:00" means 9 a.m. where your team is based. Keep that in mind if your recipients are spread across other regions.
Delay cap
The delay cap is a waiting period before the first send. You set it in days, hours and minutes, and the clock starts when an item enters the queue. The values accumulate, so 1 day, 6 hours and 30 minutes means a total wait of 30 hours and 30 minutes.
This is useful when you want to give people time before you ask for feedback:
- Wait 1 day after a purchase so the customer has a chance to actually use what they bought before you survey them.
- Wait 2 hours after a support ticket closes so the resolution has had time to settle.
By default there's no delay, and messages go out as soon as the dispatch window allows.
Frequency cap
The frequency cap stops you from over-messaging the same person. It sets a minimum gap — again in days, hours and minutes — before the same recipient is messaged again by this pipeline. As with the delay cap, the units add up.
By default there's no limit. Set one to avoid survey fatigue:
- A gap of 30 days means someone hears from this pipeline at most once a month.
- A gap of 7 days keeps a weekly cadence at most.
When a recipient would be messaged again before the gap has passed, that item is held back rather than sent — you'll see it marked accordingly in the dispatch queue.
Failed sends retry on their own
If a send fails, the platform automatically retries it a few times — you don't need to configure anything. You can watch the status of every message, including retries, in the dispatch queue.
Putting it together
Imagine an email pipeline for a delivery survey. You set the active window to run for the next quarter, keep the dispatch window at 09:00–21:00, add a 1-day delay cap so customers have received their order, and a 30-day frequency cap so no one is surveyed twice in a month. A response that arrives at 11 p.m. tomorrow will wait through the night, clear its one-day delay, and go out the following morning — unless that customer was already surveyed in the last 30 days, in which case it's held back. Together these controls help you send at the right moment without overwhelming anyone.