Automated editorial calendar: Notion + n8n to publish across all your social networks

Keeping three or four social networks alive without a dedicated tool always ends the same way: a forgotten spreadsheet, copy-pasting between tabs, a LinkedIn post republished verbatim on Instagram where it falls flat, and whole weeks without a single publication as soon as real work picks up. The good news: you already have the building blocks for an automated editorial calendar that actually lasts. A Notion database as the single source of truth, an n8n workflow that reads it every hour, an LLM that adapts the copy to each network, and the LinkedIn, Instagram or X APIs to publish. This guide assembles the whole thing, pitfalls included.

The Notion database as the single source of truth

Everything starts from one Notion database that the whole team can feed without ever opening n8n. The minimal columns:

• Content (long text): the "master" message, written without worrying about any specific network.
• Publish date (date with time): the point from which the post is allowed to go out.
• Target networks (multi-select): LinkedIn, Instagram, Facebook, X — one post can target several.
• Status (select): Draft, To review, Ready to publish, Publishing, Published, Error.
• Visual (files and media): the image to attach, if any.
• Published URLs (text): filled in by the workflow after publishing, for auditing.

Keeping a single master message and adapting it per network is more than a convenience. A study by Dolan, Conduit, Frethey-Bentham, Fahy and Goodman, published in 2019 in the European Journal of Marketing and based on twelve months of Facebook page data, shows that content types (informational, entertaining, relational) have distinct effects on users' active and passive engagement (Dolan et al., 2019). In other words, duplicating the exact same text everywhere wastes each platform's potential: the per-network adaptation we delegate to AI here is precisely what makes the automation worthwhile.

For the technical connection (integration, database sharing, credential), follow our guide to connecting Notion to n8n.

The publishing workflow, step by step

1. An hourly Schedule Trigger

A Schedule Trigger runs the workflow every hour. Make sure the workflow's timezone matches the one used for dates entered in Notion — it is the number one cause of posts going out hours off schedule, covered in detail in our guide to the Schedule Trigger, cron and timezones.

2. Reading the posts that are due

A Notion node (Get Many on the database) filters pages whose status is "Ready to publish" and whose date has passed. Using the JSON filter mode, the request sent to the Notion API looks like this:

{
  "and": [
    { "property": "Status", "select": { "equals": "Ready to publish" } },
    { "property": "Publish date", "date": { "on_or_before": "{{ $now.toISO() }}" } }
  ]
}

3. Marking "Publishing" before anything goes out

Immediately after the read, a Notion Update node switches each page to the "Publishing" status. This deliberately happens before any publishing API is called: if the workflow replays (restart, manual run, a second trigger firing while the first is still running), the step 2 filter will no longer return those pages. It is the same reasoning as webhook idempotency and duplicate prevention: a post stuck in "Publishing" that never went out is visible and recoverable; a duplicate published to your followers is not.

4. One branch per network, one text per network

A Split Out node explodes the "Target networks" multi-select to produce one item per post × network pair. Each item then goes through an AI node (AI Agent or a direct model call) with a prompt that enforces the network's constraints:

• LinkedIn: 900 to 1,300 characters, professional tone, hook in the first line, 0 to 3 hashtags at the end.
• Instagram: a shorter, more personal caption, a call to action, 5 to 15 hashtags, no link in the text (it is not clickable).
• Facebook / X: short version, one single idea, link included; for X, a hard length limit that the prompt must state explicitly and the workflow must verify.

Ask for structured JSON output ({ "text": "...", "hashtags": [...] }) rather than free text: it keeps the rest of the workflow predictable.

5. Publishing through the APIs

A Switch node routes each item to its network's branch. For LinkedIn, the POST to https://api.linkedin.com/rest/posts is detailed in our guide to automated LinkedIn publishing with n8n and AI; for Instagram, the two-step flow (create a media container, then publish through the Graph API) is covered in our dedicated guide to automated Instagram publishing. One crucial setting: configure each branch to continue on error (the node's error-handling option) so a failure on X never blocks the LinkedIn publication of the same post.

6. Updating Notion and notifying Slack

After each successful publication, a Notion Update node writes the post URL into "Published URLs" and, once all branches of a post have responded, sets the status to "Published" — or "Error" with details if at least one network failed. A Slack message sums it all up: content, networks that succeeded, networks that failed. Your Notion calendar doubles as your logbook.

Handling visuals

The image attached in the "Visual" column is exposed by the Notion API as a temporary URL (it expires after roughly an hour): download it as binary data at the start of the execution, then upload that binary to each network. Watch the formats: Instagram expects ratios between 4:5 portrait and 1.91:1 landscape, LinkedIn is more forgiving, so is X — a resizing step, or a 1080×1350 master image cropped per network, avoids silent rejections. If a post has no visual, you can generate one automatically, as described in our guide to generating AI images in n8n, with the "To review" status so a human sees it before it ships.

Queueing and spacing publications

Publishing eight posts in the same minute because they were all due looks like spam — to your followers and to the APIs. Two simple safeguards: cap each run at the N oldest due posts (a Limit node after sorting by date), and insert a Wait node of two to five minutes between two publications within the same run. Whatever is left waits for the next hourly pass — that is a queue behaving as designed, not a bug.

AI drafts and human review

The calendar is useless if it stays empty. Two sources can fill the "Draft" column automatically: your own blog articles (the AI generates a social variation whenever a new one is published) and your automated RSS monitoring, whose best items become commented post proposals. In both cases, the draft is created in Notion with the "To review" status: a human reads it, edits it, picks the networks and the date, then moves it to "Ready to publish". For an even smoother review loop, the human approval pattern with the Wait node and Slack buttons lets you approve straight from Slack without opening Notion. Resist the temptation to publish AI drafts unreviewed: brand voice and editorial responsibility stay human.

The pitfalls that break an automated calendar

• Expiring tokens. LinkedIn access tokens typically last 60 days; Meta's have their own cycles. Set up a scheduled reminder before expiry and an explicit "Error" status whenever a call returns 401 — otherwise you discover the outage three weeks and twelve posts too late.
• Duplicates on replay. "Publishing" status written before publishing, every single time (see step 3).
• Image formats. One file does not fit all networks; validate the ratio before uploading rather than deciphering cryptic API errors afterwards.
• One failure blocking everything. Error handling per branch and per network — never a global failure because one network is down.
• Timezones. A Notion date without a time is interpreted as midnight; require a time on every entry and align the workflow's timezone.

Key takeaways

A structured Notion database (content, date, networks, status, visual), an hourly Schedule Trigger, a "Publishing" status written before anything is published, AI adaptation per network instead of uniform cross-posting — a choice backed by social media engagement research —, independent publishing branches with per-network error handling, and a write-back to Notion (URL + status) paired with a Slack notification. Fed by AI-generated drafts that a human reviews, this automated editorial calendar runs for months without intervention, and every incident stays visible and recoverable right in the database.