The calendar and what actually went out
A content calendar is usually a spreadsheet with a column for each channel, and by the end of a planning session it looks finished. The grid is full. What it doesn't show is that every cell still depends on a person being at a desk at the minute written in it, with the image to hand and the wording signed off. Tuesday morning rarely goes that way. The post goes out at half past eleven when the cell said nine, or it reaches LinkedIn and never reaches Instagram because the image was the wrong shape.
Nobody records the miss. At the end of the month the calendar shows the same full grid it showed at the start. What actually went out is a thinner set of posts, at hours nobody chose. Engagement then gets read against the plan, so a post that reached one channel at the wrong hour is judged as though it went out as intended everywhere. The calendar looks like a record when it was only ever a plan.
We've stopped treating a calendar as done at the point where the cells are filled in. The question we take into a planning session is narrower. For the posts planned last month, how many went out at the time and on the channels in the cell, and who can show that from a log. Most teams answer the first part with a guess and the second not at all. That is the answer, because a calendar that can't be checked against what happened is a list of good intentions with dates on it.
What the scheduler can be trusted with
Posting itself is mechanical work. We hand it to a machine. A scheduling tool takes a finished post, the wording, the image, the link and the channels it was approved for, and publishes it to each of those channels at the set time. The caption is cut to each channel's limit and the image cropped to its shape. It does this at nine on a Tuesday whether or not anyone is at a desk, and the same way on the last post of the month as on the first. A person does it worse, because the person has a meeting at nine.
The log comes second. Every post carries the time it was set for and the time it went out, channel by channel. At month end the calendar is a record, and the plan sits beside it. When a channel rejects a post, because the connection has expired or the image is over the size limit, the tool sends the failure to a named person within minutes. That message is the difference between a miss fixed the same morning and one nobody notices until the client asks where Tuesday went.
It can also hold the queue. A month of approved posts sits in the tool in the order they'll go out. Moving one means dragging it to a different day, and the log keeps the original date. The plan and the record then live in the same place, which is the arrangement the spreadsheet was standing in for all along.
What stays with a person
The writing stays with a person, and so does the choice of what goes in the calendar at all. A scheduler publishes whatever it is given at the time it is given. It has no view on whether the post is worth reading, so the quality of what goes out is settled before the tool sees it. Filling a month's queue is quick, and that is the risk, because a month of posts nobody would stop to read is a worse result than a calendar with gaps.
Judging the tone stays with a person too, on the day of posting as much as on the day of writing. A post written a fortnight earlier can land on a morning when something has happened in the client's industry, and a cheerful line about a product launch reads badly next to it. Nothing in the scheduler knows the morning has changed. Somebody has to be able to pull a post in seconds. So we ask for the next day's queue to be shown to a named person the evening before.
Replies stay with a person. A comment under a scheduled post arrives at an hour the tool chose, and the answer to it is a conversation with a customer. The rule is the one we hold to on follow-up sequences: the automation sends, and anything that comes back goes to a person. A tool that also answers comments is talking to the public in the client's name with nobody watching. One wrong reply costs more than every hour it saved.
The post has to be finished first
A scheduler changes when the writing happens, which is the part of the build teams don't expect. The cell that reads LinkedIn post about the new service is not a post, it is a reminder to write one. Nothing can be scheduled from a reminder. For a post to go out at nine on Tuesday, the wording has to be final, the image sized for each channel, the link built with its tracking and the post approved by whoever signs off. All of that has to be true before the week starts and not at ten to nine.
The writing moves earlier as a result, and it moves into a batch. Teams we've worked with end up writing a fortnight or a month of posts in a sitting, with the sign-off done on the batch. From then on the scheduler owns the month. The morning that used to go on finding the image and chasing the sign-off goes back to the team. A queue also needs a floor, set in days and chosen by the client. When the finished posts in hand drop below it, the tool tells a named person, and the next writing session goes in the diary before the channel goes quiet.
Sign-off is the step that gets skipped, and it is the one that can't be. Once a post is in the queue it will go out. The approval has to happen before it goes in, and the person whose name the channel carries has to have read it. We keep the number of approvers to one wherever we can. A post that needs a committee's agreement sits in an inbox past its date and the scheduler posts nothing, and that is a miss with a better excuse.
