Schedules
Give a piece its destinations and its moment and leave. Each destination becomes a real instruction with its own hour, and every minute the platform asks what is due and sends it, reading the words and the pictures as they stand at that second. Everything here runs on a clock you set or a button you pressed. Nothing watches your workspace for things happening and acts on them, which is exactly why you can predict what it will do while you are asleep.
Schedules, close up
Set by you
Runs on the clock
Where it lands
What is travelling those lines
This is the same map as the one on the front of Solutions, zoomed into one surface. Every arrow starts at a clock or at a person, because those are the only two things in here that start anything.
What it does
Scheduling a post writes one real instruction per destination, each holding its own hour. Three channels are three instructions, so they succeed and fail one at a time instead of as a lump, and each one can carry a different moment.
Nothing about the post is copied into the queue. The instruction resolves the content, the caption and the pictures at the second it goes, so an edit made on Sunday night is the version that lands on Tuesday morning. There is no stale copy anywhere to go out by mistake.
A schedule that belongs to a brand has its next slot worked out against that brand's own business hours, the ones set on the brand. Run several brands in several places and none of them needs you doing arithmetic. A schedule that belongs to the whole workspace has no brand to borrow hours from, so it runs on the platform clock.
A send whose moment went by while the workspace was unreachable is marked missed rather than fired hours late. This is the fear people actually have about scheduling, answered directly: a post for Friday morning will not appear on Saturday night.
A refused send is tried again after a growing wait, a small number of times, and then stops. It does not loop, it does not keep hammering a channel that has said no, and it stays on the piece with the reason it was given so you can send it again yourself.
Publishing this second and publishing next week run through one execution path. That is not tidiness for its own sake: it is the reason a scheduled publish behaves exactly like the one you watched work, rather than being a second implementation that drifts.
Pause a brand while you rework it and its scheduled work stops, while its slots keep moving forward. Come back a month later and it starts from today. Nothing replays, and nothing arrives all at once as punishment for taking the time.
One send becomes one instruction per subscriber, and each one passes the same suppression, duplicate and rate rules a single send obeys. A large list goes out steadily rather than in one burst, and anyone beyond the pace goes back in the queue for a later minute rather than being dropped.
Every workspace starts with the drain of the queue already switched on, which is why a scheduled post goes out without you on your first day, and with a daily health check of your connected accounts so the rest of the product already knows the state of them rather than making you wait while it asks.
The boundary, said plainly
Products in this corner of the market usually promise that the software will notice something and act on it. m18t does not do that, so this page does not say it. There are exactly these ways for work to begin, and you are standing at the start of all of them.
A moment
A piece of content, its destinations, and one hour in the future. That is a scheduled publish, and it is the strongest thing on this page.
A frequency
A saved schedule running one of the built in pieces of work, over and over, at the frequency you gave it, on the clock of the brand it belongs to.
A press
Publish now, or play on a schedule that was not due yet. It goes down the same path the clock would have sent it down.
There is no fourth. Nothing in m18t sits and watches your workspace for something to happen, so nothing here can surprise you with an action you did not ask for at an hour you did not choose. That is a smaller promise than the category usually makes, and it is the whole reason you can hand it a week of posts and stop thinking about them.
One send, end to end
One piece of content, three destinations, one hour in the future. This is the whole life of that instruction, including the part on Sunday night where you change your mind about a line.
Three separate instructions are written down, one for each destination, each carrying the moment you chose. The post is not copied into any of them. The piece keeps its own status, which stays yours to manage and starts nothing by itself.
Each instruction sits marked as queued against the hour you gave it. Holding it costs nothing, and changing your mind about it costs nothing either.
Once a minute the platform asks this workspace what is due. Four days of nothing due is four days of that question answered no: no run recorded, no noise added, nothing for you to read.
Open the piece, change the words, save, and leave the schedule alone. There is no frozen copy in the queue to go stale, so there is nothing to reschedule and nothing to cancel and redo.
The first instruction is picked up, reads the piece as it stands at that second, resolves the caption and the pictures for that one destination, and sends. The brand's own stored credentials are opened for it, so the send does not need anybody signed in.
Each destination is its own instruction. A refusal at one leaves the other two untouched, so two out of three going out is a real and legible outcome rather than a half finished job somebody has to unpick.
What went, where and at what time now sits on the content itself. A refusal sits there too, carrying the reason the far end gave, and it can be sent again from the same place.
A send that was never attempted, and whose moment is now further behind than a short grace period allows, is marked missed rather than going out late. A post for Friday morning does not surprise you on Saturday night. It stays on the piece, marked missed, and going out is a decision you get to make again.
A refused send is tried again after a short wait, then a longer one, then a longer one still. After that it stops on purpose rather than looping, and it stays on the piece with the reason it was given. It waits there for you rather than in your inbox, and one click sends it again.
Its scheduled work stops for as long as it is off, and its slots keep moving forward the whole time. Switch it back on and it starts from today. A month away does not come back as a month of posts arriving in one minute.
Where this sits
Schedules is not a place you spend time. It is the thing that lets you leave the other places, so every join here is a surface handing it work to hold.
The destinations. A queued send resolves the same channel settings, the same overrides and the same credentials an immediate publish resolves, because it is the same path.
Read PublishingThe piece the instruction reads at the moment it fires, and the place every attempt reports back to: sent, refused, missed, with the reason, and cancel for anything not gone yet.
Read ContentA newsletter is the biggest thing the queue carries. It fans out into one instruction per subscriber so the list paces itself against the same limits a single send obeys.
Read EmailsThe one place you pick a frequency from a dropdown rather than writing one. Choosing how often your sources are read saves a real schedule behind the screen.
Read NewsThe clock itself. A brand carries the business hours its schedules are worked out against, and pausing a brand is what stops its scheduled work without losing it.
Read BrandsYours to keep
Every one of these we would choose again.
Every start here is a time you set or a button you pressed. A workspace that only ever does what you told it, at the hour you told it, is a workspace you can leave alone without a knot in your stomach.
Because nothing is frozen at schedule time, correcting a piece is just correcting a piece. You never have to remember which posts were already locked to an old draft.
A missed hour is marked missed and left for you. The platform does not decide on your behalf that six hours late is close enough.
Switching a brand off costs you nothing on the way back in. There is no queue of a month's work waiting to punish you for taking a month.
A refused send is tried a few times and then left alone with its reason, on the piece it belongs to. Nothing keeps hammering a destination that has said no.
Anything not yet gone can be cancelled outright. Anything already gone is already gone, and the product says so plainly instead of offering you a button that quietly does nothing.
Questions
Keep reading
Where the queue delivers, and what each destination does with it.
The piece a queued send reads at the moment it fires.
The biggest thing the queue carries, one instruction per subscriber.
The clock a schedule is worked out against, and the pause that stops it.
Open the workspace, do one real piece of work in it, and the joins will show you the rest.