Comparisons
Shared calendar vs publishing queue: not the same job
A shared calendar and a publishing queue solve different problems. One tracks intent, the other executes an approved version. Here is how to keep them apart, with examples of a calendar entry, a queue entry, and a draft that is not yet a version.
A shared calendar and a publishing queue do not solve the same problem. Picture Monday morning: the editorial team opens the calendar and sees an article on AI visibility scheduled for Thursday. That entry says who writes it, by when, and under which editorial plan. It does not yet say what will be published. The queue is different: it holds a specific version that has already been approved and is ready to run. Mix the two up and you start believing that “on the calendar” means “ready to go live”. It doesn’t. RankAI makes human approval of a version the default; automatic publishing is switched on per site and still goes through the checks. The queue executes what has been approved. The calendar organises what still has to be produced.
Two axes, not one
The split is not just “plan versus publish”. There is a second axis: who owns the status and who owns the deadline. In the calendar, status reads “in production”, “in review”, “approved for the queue”, and the deadline is the editorial date. In the queue, status reads “awaiting approval”, “approved”, “published”, and the deadline is the moment of execution. Two different owners, too: the calendar belongs to whoever coordinates the editorial plan, the queue to whoever controls execution. When one person runs both, an editorial deadline can easily be mistaken for a green light to publish.
A three-step procedure from RankAI
This is an operating proposal from RankAI, not a universal standard. It keeps the two tools apart without adding pointless steps.
1. Write the calendar entry
A filled-in calendar entry might read: “Thursday 14 — Article: AI visibility and the editorial plan — Author: Giulia — Status: in production — Deadline: Wednesday 13, draft ready — Notes: engine data review still missing”. Notice what is absent: no reference to an approved version. The entry describes work to be done, not content to be executed.
2. Write the queue entry
A filled-in queue entry might read: “AI visibility article — Version 3 — Approved by Marco on 13/05 — Site: main blog — Automatic publishing: on for this site — Checks: passed — Execution: 14/05 at 09:00”. Here the version is identified, the approval is on record, and execution is separate from the editorial deadline.
3. Spot a draft with no version
A draft with no identifiable version is, for example, a shared document titled “article draft” with no number, no approval date and no named reviewer. It cannot enter the queue: there is no precise object that human approval was given to. The test is not “the file exists” but “a version exists that someone approved”.
The point that often gets lost: the calendar and the queue remember different things. The calendar remembers intentions — what we meant to produce and when. The queue remembers decisions — which version was approved and under which checks. Delete a calendar entry and the trace of an already published version should not vanish with it; run a queue entry and the calendar should not mark itself “done” until someone has checked the outcome. RankAI keeps these two levels apart: human approval of a version is the precondition for execution, while automatic publishing, when it is on for a site, still goes through the checks. In practice, before moving anything from the calendar to the queue, ask yourself: “Do I have an identifiable, approved version?” If the answer is no, you are still organising, not executing. That avoids two matching mistakes: publishing a draft mistaken for a final version, or keeping a queue entry that has no approved content behind it.
Why it matters for AI visibility
When the editorial plan includes content built for AI visibility, the queue is where an approved version gets executed. RankAI’s public pages name ChatGPT, Perplexity, Gemini, Grok and Google AI. Claude and Grok are measured only after you connect them. Gemini and Google AI Overviews are not the same observation. So a queue entry should not just say “publish it”: it should state which site and which checks, because execution is what makes the link between the editorial plan and the observation of the named engines verifiable.
Verifiable limit: this procedure does not decide when a version is “good”, and it does not guarantee that a queue entry will produce a visibility result. It only establishes that the calendar and the queue are not interchangeable. For the full picture of the tools, see the features page.