Editorial planning
SEO editorial planning with approval: the FAQ's role
RankAI's FAQ connects what Google and AI engines read on your site to an editorial plan that passes through human approval. Here is how the editorialPlan page works.
If you are building an SEO editorial plan with approval and want to know how what Google and AI engines read on your site connects to editorial decisions, RankAI's FAQ is built for exactly that. On the editorialPlan page, observations about how engines read your site feed into an editorial plan that goes through human approval. When ChatGPT, Perplexity, Gemini, Grok or Google AI reads a piece of content, that data does not sit on its own: it becomes something to weigh before you decide what to publish and in what form. One point stays fixed: the default is human approval of a version. No draft goes live without that step unless automatic publishing is turned on for a site.
From engine reads to the plan: the steps on the editorialPlan page
The editorialPlan page is not a list of ideas. It is where observations about how your site is read become a plan with an explicit check. Here is how we do it, with criteria you can verify.
- Collect observations per engine and keep them separate. RankAI's public pages name ChatGPT, Perplexity, Gemini, Grok and Google AI. Keep the observations apart: Gemini and Google AI Overviews are not the same observation and should be treated as two distinct reads. Verifiable criterion: in the plan, every observation names the engine it refers to, and there is no generic “AI” entry that merges them.
- Connect only what was actually measured. Claude and Grok are measured only after you connect them. Verifiable criterion: before you connect them, the plan contains no rows attributed to Claude or Grok; after you connect them, the rows appear with the engine named.
- Send every version through human approval. The default is approval of a version: a draft becomes publishable only after that step. Verifiable criterion: in the plan's history, every published version has an approval recorded before publication.
- If you turn on automatic publishing, do it per site and keep the checks. Automatic publishing is turned on per site and still goes through the checks. Verifiable criterion: the setting is visible at the level of each individual site, and the checks stay active even with automatic publishing on.
Why keeping observations separate matters in the plan
An SEO editorial plan with approval only works if the information behind each decision is readable and attributable. If Gemini and Google AI Overviews reads were merged, the plan would lose the ability to tell where a signal comes from. The same holds for engines measured only after you connect them: until the connection exists, there are no observations to enter, and the plan has to reflect that absence rather than fill it in.
A RankAI working proposal, not a universal standard: treat the editorialPlan page as a register with two conceptual columns, “observed” and “decided”. Only reads attributed to a specific engine among those named on the public pages go in the observations column, along with a note on whether the measurement is available or requires the connection. The editorial choices that follow go in the decisions column, each tied to the version that was approved. The advantage is that the approval step stops being an opaque bottleneck: whoever approves can see which observation a version rests on, and whoever reads the plan later can retrace the path. The proposal does not replace the product's rules: human approval remains the default, and automatic publishing, if turned on per site, still goes through the checks. The method makes visible what would otherwise stay implicit in the handoff between reading the site and publishing.
One limit to keep in mind
The link between observations and the plan applies to the engines RankAI names and to the measurements that are actually available: Claude and Grok only after you connect them, and Gemini kept distinct from Google AI Overviews. An API response is not a person's screen, so the plan should not be read as a snapshot of what a user sees in their own browser.
For the full picture of how the page works, see the in-depth page.