How to use the content planner to run a week of posts

You sit down on Monday to write a week of posts. You write one, then check something, then write half of another, then find a link you meant to save, then it is Wednesday and the one post you finished has not gone out because you are not sure it is right.

The problem is not writing speed. It is that idea, drafting and reviewing are three different modes, and doing them interleaved means paying the switching cost twelve times instead of twice.

The planner is a board for keeping them separate.

What is the goal?

A week of posts that exist, have been reviewed, and are sitting ready — so publishing becomes a two-minute job you do from a list rather than a creative act you have to summon on the day.

What is the planner, and what does it control?

One board, three views, four lanes.

Board is the default: cards moving left to right. Calendar shows the same cards by date, with the week starting on whichever day your display preferences say. List is the dense view for bulk tidying. All three read one dataset loaded in a single query, which is why filtering and searching are instant.

A card holds the post type, the audience, the text, an optional image prompt, and a scheduled date. The underlying statuses are idea → review → ready, plus posted.

The planner plans, and it never posts. There is no publishing integration in Reply Pilots for it to use, and the header says so on every visit for exactly that reason.

How do you set the planner up?

0. Do the basic setup first. The planner generates from the same business context and voice as everything else, so the fifteen-minute setup is worth doing before you generate anything here.
1. Set your display preferences first. Timezone, time format and week start live in Settings, and week start drives which day the planner's calendar begins on. Getting this wrong once is confusing for a month.
2. Pick your post types and audiences. These are the two fields every card carries, and they are what make the board filterable later.
3. Add three or four ideas by hand before generating anything. A board with real ideas on it makes the generate step much better, because you are asking it to develop your intent rather than invent one.
4. Generate a post. One line of intent becomes a caption with hashtags. It writes a card in the idea lane — it does not publish, and it does not schedule.
5. Generate an image prompt if you want one. You get a dense visual description to paste into whichever image tool you use.
6. Resize if needed. Make longer or shorter adjusts an existing draft without you rewriting it.
7. Set dates only when a card reaches Scheduled. Dating an idea creates overdue cards for work you were never committed to, which is the fastest way to make the Overdue lane meaningless.

Generation costs credits and shows up in your usage like every other AI action, so you can see what the planner is actually costing you.

How do the four lanes actually work?

This is the part worth understanding properly, because it is where people get stuck.

There are four lanes over three statuses, and the fourth — Overdue — is not a status at all. It is derived: a card whose scheduled date is in the past and which is not yet posted. Overdue wins, so a past-due card sits in Overdue regardless of whether its status says review or ready.

The consequence that surprises people: setting an overdue card's status to "review" does not move it. The status changes, the save succeeds, and the card stays exactly where it was — because it is still past due, and Overdue beats status.

That is why every lane move settles the date as part of the same write. Moving to review or idea clears the date, because undated-in-review is a normal state. Moving to Scheduled rolls it to today, because "Scheduled, no date" is a contradiction. Posted leaves it alone, because the archive matches on status. When a move changes a date, the board tells you — it does not move it behind your back.

Comparison is by calendar day, not by timestamp. A post due at 09:00 does not turn red at 09:01.

What does a weekly workflow look like?

Batch by stage, not by post. Three sittings, about ninety minutes total.

Collect with no dates. An idea is not a commitment. Dating it creates an overdue card for something you never promised.

Draft without editing. The point of a drafting sitting is volume — get six cards to a rough complete state. Editing while drafting is the mode-switching this workflow exists to avoid.

Review as a separate act, ideally a day later. You will cut two of the six, and cutting is much easier when you did not just write them. This is also where your rules earn their keep — a post is more public than a comment, so an over-claim costs more.

Only Scheduled cards get dates. Then the Overdue lane means something: it is work you committed to and did not do, which is exactly the signal you want it to carry.

What are the best practices and common mistakes?

Do: collect undated · batch by stage · review a day after drafting · use audience and post type consistently so filters work later · archive posted cards so the board stays under its limit.

Do not:

• Date your ideas. The single most common mistake. It fills Overdue with things nobody promised, and once Overdue is noise you stop reading it.
• Generate straight into Scheduled. Skipping review is how an unedited draft goes out under your name.
• Edit while drafting. It halves your output and produces worse edits.
• Rely on the stat tiles reflecting a filter. They count the whole board on purpose.
• Treat the image prompt as an image. It is a description to paste into an image tool.

What will the planner not do?

It will not publish. No integration, no scheduler that fires, no queue that drains. The Scheduled lane means "ready and dated", not "will go out automatically". Marking a card Posted records that you posted it somewhere else.

There is one opt-in daily job that can generate a post for you on a schedule — and what it does is write a card on the board. It is off by default, owner-level, and it still does not publish anything.

It will not fetch your engagement numbers. The planner is not analytics.

It will not know what performed. The audience and post-type fields are for your own filtering.

It will not fix a thin voice. A generated post is only as specific as the context behind it, so if the drafts read as generic the fix is in the voice setup rather than in the planner.

That review gate is the feature, not a limitation. An unreviewed post published under your name is the one failure mode with no undo, and the board is shaped so that cannot happen by accident.

Troubleshooting

Your next step

Open the planner and add four ideas — one line each, no dates. That is the whole collect step, and it takes two minutes.

Tomorrow, set a 30-minute timer and generate from all four without editing any of them. The day after, review and date them. One pass through the cycle tells you more about whether the batching rhythm suits you than any amount of planning it.

See the documentation for the exact, step-by-step setup instructions.