Editorial planning
SEO editorial plan: approve it with Google, Perplexity and Gemini
How to use Google's reading of a page alongside Perplexity and Gemini answers to decide whether each item in an SEO editorial plan is ready, needs fixing, or should be dropped.
To approve an SEO editorial plan by comparing Google's reading of a page with the answers from Perplexity and Gemini, take each planned page, check what it tells Google and what comes back when the same question is put to Perplexity and Gemini, then decide whether the draft is ready, needs fixing, or should be dropped. In RankAI this comparison gives the approval of an SEO editorial plan a checkable basis: a page being well written is not enough, it has to hold up both in the search engine's reading and in the assistants' answers, and the two observations must be kept separate. Approval stays human and applies to a version; automatic publishing is turned on per site and still goes through the checks.
Why the two observations stay separate
Google's reading and the answers from Perplexity and Gemini do not measure the same thing. Gemini and Google AI Overviews are not the same observation: treat them as two separate sources, even when the engine's name appears in both. RankAI's public pages name ChatGPT, Perplexity, Gemini, Grok and Google AI; Claude and Grok are measured only after you connect them. That is why, at approval time, we compare what the page states to the engine with what the assistants return, without merging the two results into a single verdict.
The approval method RankAI proposes
What follows is an operational proposal from RankAI, not a universal standard: its purpose is to make the decision on each item in the editorial plan repeatable.
- Set the page's reference question. Checkable criterion: the question is written in one sentence and the draft contains a direct answer within the first paragraphs. If the answer cannot be found, the item goes back for revision.
- Record Google's reading. Checkable criterion: note what the page tells the engine — title, heading structure, entities and main topic — and compare it with the intent of the question set in step 1. List any discrepancies; do not fix them from memory.
- Query Perplexity and Gemini separately. Checkable criterion: put the same question to both assistants and keep the answers as two distinct observations, without attributing to Gemini what belongs to Google AI Overviews. If an answer does not cite the page or cites different content, flag the item.
- Compare and decide. Checkable criterion: approve the item only if Google's reading matches the intent and the Perplexity and Gemini answers do not contradict the page on key points. Otherwise fix or drop it, with the reason written next to the item.
- Approve the version and log the check. Checkable criterion: approval is human and tied to a specific version; if automatic publishing is on for that site, the check is still logged before release.
Where it fits in the workflow
In RankAI the default is human approval of a version: the comparison between the two observations feeds that decision, it does not replace it. Automatic publishing is turned on per site and still goes through the checks, so the method holds even when release is faster. Keeping the sources separate avoids the most common mistake: treating a Gemini answer as if it were Google's reading, or the other way round, and approving a page on the strength of a single observation.
The practical contribution of this piece is the two-column approval sheet we use internally: Google's reading on the left, the Perplexity and Gemini answers on the right, on separate rows. Each row carries the reference question, the observed result and the decision (approve, fix, drop) with a short note. The rule we set ourselves is that an item cannot be approved if either column is empty: a missing observation is a reason to postpone, not a green light. And when Perplexity and Gemini diverge from each other, we do not average them: we write down both answers and leave the choice to the human reviewer, because the divergence is useful information about the content, not noise to be smoothed over. This format makes the editorial plan checkable item by item and leaves a record of why a page was approved in a given version.
A checkable limit
The method does not guarantee that a page will be cited or shown: it measures the consistency between what the page tells Google and what emerges in the Perplexity and Gemini answers at the time of the check. Claude and Grok are measured only after you connect them, and Gemini and Google AI Overviews remain distinct observations: if one of these sources is not available, the sheet should be completed only with the observations actually collected.
For the full picture on approving editorial plans, see the in-depth page.