Editorial planning
How to Review Priorities With Your First Measurements
Your first measurements don't tell you whether an editorial plan works. Here is a written-threshold method for deciding what to revise, using only the evidence you actually have.
You have an SEO editorial plan with approval in place, and the first measurements are coming in. How should priorities be revised? At RankAI we start from a simple rule: revise only what available evidence can support, and keep every revision on record. Human approval of a version is the default; automatic publishing is switched on per site and still goes through the checks. On the technical trial, the limits are 3 articles per month and 3 in total. That means the first measurements have to be read within a narrow scope, not as a report card on the whole plan.
Why first measurements are not a verdict
With only a few articles published, measurements mostly show what was approved and in what order. They do not yet show whether an editorial line works. Revising priorities, then, is not a reshuffle driven by impression: it is a decision made with the available evidence, written down so the next review can compare against it.
The RankAI method for deciding revisions
This is an operational proposal from RankAI, not a universal standard. Its purpose is to make two review points comparable without inventing new criteria each time.
- Write the threshold before you look at the data. Use a sentence template to fill in: “I will revise the priority of [article or topic] if [signal] appears on at least [number] consecutive approved versions, within [interval].” A filled-in example: “I will revise the priority of article X if the same signal appears on at least two consecutive approved versions, by the next review cycle.” Verifiable criterion: the sentence exists, is dated, and names an observable signal.
- Check that the version is approved. Before attributing an effect to an article, confirm that the version under observation is the one a person approved. Verifiable criterion: in the revision log, the approved version matches the published one.
- Compare two reviews, not two impressions. Set the threshold written in the previous review against the signal observed now. Verifiable criterion: the decision cites the threshold and the signal, or states that the threshold was not reached.
- Stay within the trial's scope. With 3 articles per month and 3 in total, a review that touches more articles than are available is not supported by the evidence. Verifiable criterion: the number of articles involved in the review does not exceed those published in the trial.
A filled-in example of a reshuffle
Suppose three articles are approved in the trial. In the first review we write: “I will revise the priority of article A if the same signal appears on at least two consecutive approved versions, by the next review.” At the following review, if the signal appears on two consecutive approved versions, article A moves up in priority; if it appears only once, it stays where it is and the threshold is cited as not reached. Articles B and C are left alone until they have a written threshold and an observed signal.
Approval and publishing are two different things
Revising priorities does not change the default: human approval of a version still applies. Automatic publishing, if switched on for a site, remains subject to the checks. A priority revision is not treated as permission to publish without approval.
The contribution of this piece is a written-threshold template, designed to be filled in before you look at the measurements. The sentence to fill in is: “I will revise the priority of [article or topic] if [signal] appears on at least [number] consecutive approved versions, within [interval].” The added value is that the threshold becomes a verifiable object: it can be cited in the next review, declared not reached, or compared with a different threshold. That way, revising priorities does not depend on who remembers best what was decided, but on a dated text. The template does not say which signals to choose: that remains a decision for the site, to be written case by case. The useful part is the form: number, signal, interval, and a named article. If any of these elements is missing, the threshold is not verifiable and the review is not supported by the available evidence.
Verifiable limit
On the technical trial, the limits are 3 articles per month and 3 in total: a review involving more articles than were published cannot be supported by the trial's evidence.
Method limit: this article does not establish which signals to observe or how to measure ranking; it only describes how to decide revisions with the available evidence and with human approval as the default. For the full picture of the editorial plan, see the in-depth page.
What kind of call to action belongs at the end of an article?