Why "thirty post ideas" does not work
If all you ask is "give me thirty ideas for my company", you get thirty generalities. They are unusable, because there is nothing in them that only you know.
A calendar is not built from a list of ideas, it is built from topic buckets. Four or five recurring buckets you draw on every month — that way you do not have to work out what to post from scratch each month, and your feed does not turn into a disconnected stream of ideas.
The half day, step by step
1. TOPIC BUCKETS (30 min, HUMAN)
4-5 buckets, each answering one thing:
- "what the customer does not get" (explainer)
- "what we do differently" (position)
- "what happened at our place" (backstage)
- "what the client pulled off" (result)
- "what is interesting in the trade" (reaction)
2. RAW MATERIAL (30 min, HUMAN)
The last 3 months: support questions, the recurring lines
from sales calls, reviews, your own articles.
This is the input. Without it the machine gives platitudes.
3. CALENDAR (1 hour, MACHINE)
"From the buckets and the raw material below, put together
4 weeks, with [N] posts a week. In every row:
date, bucket, WHAT the post claims in one sentence,
format, and which piece of raw material it comes from.
Do not write the copy yet. If a bucket has too little
raw material, say so, do not make it up."
4. FILTERING (30 min, HUMAN)
Cross out whatever is not yours. What is left is the plan.
5. COPY (1.5 hours, MACHINE + HUMAN)
One request per post, with the calendar row as the brief.
One request = one post. "Write all twenty" gets you
twenty identical posts.
6. VISUALS (1 hour)
From a template, in a series.
The visual side calls for the same template-driven workflow that producing social visuals at scale sets out: fixed brand elements, an interchangeable layer.
Posting frequency: what the data actually says
You will find numbers on this everywhere, and it is worth knowing where they come from.
There is no official platform data on the relationship between posting frequency and reach. Neither Meta, nor TikTok, nor LinkedIn publishes any. What exists is tool-vendor analysis.
The largest-sample analysis of that kind was built on the data of more than a hundred thousand accounts, and it measures median reach per post. By it, on Instagram 3–5 posts a week show roughly 12% more reach per post than 1–2 a week; above 10 a week, 24%. On TikTok, going from 1 a week to 2–5 shows at most 17% more views per post. On LinkedIn the turning point is 2–5 posts a week.
Now comes the part nobody ever quotes alongside it: this is correlational data, not causal. Whoever puts out ten posts a week is typically a bigger, more practised, more active account. Frequency and reach move together — but it does not follow that frequency causes the reach. The analysis does not state the number of posts examined or the period covered either.
The usable reading: 2–5 posts a week is a sensible starting point on all three platforms, and increasing it has diminishing returns. But if the extra post is weaker, the number will not save you.
Does it matter that it was made with AI
Three platforms give a clear answer from a primary source, and all three are more reassuring than the trade gossip.
- YouTube: "the disclosure label by itself does not change how the system recommends the video, nor whether it can be monetised." That is a verbatim platform statement.
- TikTok: the platform gives the user a control for setting how much AI content they want in their feed. That is not a penalty, it is a preference — but its practical effect on reach can be real. On the label itself hurting reach, TikTok says nothing.
- Google search: what is forbidden is not the automation but the intent. Under the guideline it is a violation if you use automation "primarily to manipulate search rankings".
One thing is worth separating sharply, though: the algorithm's reaction and the reader's are not the same thing. The fact that rankings do not punish the label does not mean your own audience will not react worse to it — only your own numbers can tell you that, and it is worth looking before you believe anyone's opinion.
For a sense of scale: according to TikTok's official statement, more than 1.3 billion videos have been marked as AI content so far. That figure has not changed in the official announcements since November 2025. The platform also puts an invisible watermark into the content — but the exact scope matters: only into content made with TikTok's own tools, not into every upload.
What also has to be planned into the calendar
- The labelling. If the video or the image was made with AI, use the platform's own marker. In most places the label goes on automatically anyway if you worked with the platform's own tool — manual labelling covers the case where you used an outside tool.
- The fact-checking. Dates, prices, legislation, "market leader": these are the sentences that slip most easily into generated copy, and they are the ones that cost the most.
- The reactive slot. Leave one empty spot a week in the calendar for whatever is happening right then. A plan written a month in advance goes lifeless precisely for the want of it.
- The repurposing. A longer article turns into five posts, not one. When you plan the calendar, mark which row comes from which existing piece.
Common misreadings
- "You need one post a day." No platform data supports this. Even the available analysis only shows correlation, and the returns diminish.
- "The algorithm penalises AI content." YouTube says explicitly the opposite, and Google looks at the intent, not the tool.
- "One request gets you the monthly plan." One request gets you the skeleton. The copy needs a separate request per post, otherwise it all comes out the same.
- "The calendar is enough, the raw material will turn up." It works the other way round: the raw material is the precondition for the calendar. Without it the machine generates platitudes, because it has nothing else to work from.
What to take away
A calendar is not a list of ideas, it is four or five topic buckets plus raw material. The larger part of that half day is spent by the human — picking the buckets and filtering — while the machine supplies the skeleton and the copy variants. The same series logic applies to advertising, only with stricter rules: ad creative at volume is the post that covers that.