Content quality scores
RankX AI gives each generated article a quality score and a GEO score: the share of pass or fail checks it passed. A blank score means not yet scored.
RankX AI puts two scores on every generated article. The quality score asks whether the piece is well made. The GEO score asks whether it is built to be quoted and cited by an AI answer engine.
Each score is the percentage of checks, or gates, the draft passed, and each gate is a plain pass or fail rather than a weighted judgement. A blank score means not yet computed, not zero: scores are computed at set points in writing, so a draft still being written has none.
Why pass or fail gates rather than a weighted score?
Every gate tests something objectively checkable about the text: a count, a threshold, the presence of an element. A weighted score built from subjective factors is a number you can move without improving the piece.
Ten gates passed out of eleven is a statement about the document. It is not a prediction that the piece will rank, so read the failed gates, not the number: the number is a summary, and the failures are your edit list.
What do the quality gates check?
The quality gates check that the article is complete, readable, original and safe for your brand.
| Gate | What it checks |
|---|---|
| Word count | The article reaches at least 80% of its target length, or a minimum when it has no target |
| Readability | Reading ease is above a threshold, so the text is plain enough to follow |
| Repetition | The text does not repeat the same phrases past a threshold |
| Writing style | No phrasing that reads as machine-written: banned stock phrases, em dashes and filler openers |
| Internal links | Links to your own indexable pages, two where the site has them. Passed automatically when there is nothing to link to |
| External citations | At least one link to an outside source. Linking the source you quote builds credibility |
| Originality | Not too similar to your existing pages, and not too similar to an earlier draft |
| Prohibited terminology | None of the phrases your brand has banned |
| Confidentiality | None of the phrases your brand has marked confidential |
| Keyword coverage | The primary keyword appears in the title, the H1 and the first hundred words |
Two of these read your brand settings rather than general practice, and they are the ones worth setting up. Prohibited terms and confidential phrases are how you stop an article saying something you would have to retract.
What do the GEO gates check?
The GEO gates check extractability: whether an AI answer engine can lift a passage from the article and use it as an answer.
| Gate | What it checks |
|---|---|
| Direct answer first | The article answers its own question at the top, rather than building up to it |
| Definitional opener | The opening defines the topic, within the first 60 words |
| Question-phrased headings | At least one heading is phrased as a question a reader would ask |
| Question and answer coverage | The article contains at least one explicit question and answer pair |
| Citation-worthy statistic | It contains a statistic that passed RankX AI's claim check |
| Schema completeness | Article structured data was actually produced for the piece |
The schema gate passes only for markup that exists. It reads the structured data RankX AI actually produced for the article, not a list of types the article ought to have.
The statistic gate needs a verified figure. It requires a statistic whose claim check came back verified, not merely a number somewhere in the text.
Where do the GEO gates come from?
The quality gates encode ordinary editorial standards. The GEO gates encode structural patterns that make a passage easy for an AI system to find and quote.
Research on large language models found that they use information at the start of a long input more reliably than information buried in the middle (Liu et al., 2023 (opens in a new tab)). Definitions and question-shaped headings follow the same logic: they match the way a question is asked, so the answer is easy to lift. None of this makes a weak article good; it makes a good article easy to quote.
How should I read a score?
Below 100 is normal. Some gates are legitimately out of reach for some articles: a purely instructional piece may have no statistic worth citing, and a new site may have nothing to link to.
A gate that fails for a good reason is not a problem. The internal-links gate passes automatically when your site has no indexable pages, because failing it there would mean nothing.
A failing keyword-coverage gate is always worth fixing. It is mechanical: the primary keyword is missing from the title, the H1 or the opening.
A failing direct-answer gate is the most valuable one to fix. It is usually a one-paragraph edit: move the answer to the top.
Where to go next
- Writing articles, for how an article is written and reviewed.
- Content briefs, where the reader questions behind the GEO gates come from.
- Content Studio, for what RankX AI's AI SEO content generator does and what an article costs.
- AI Readiness, the same idea applied to your whole site.
Questions
Does a GEO score of 100 mean my article will be cited by ChatGPT?
No. A full GEO score means the article has the structure answer engines can quote from, such as a direct answer, a definition and question headings. Whether an assistant cites it also depends on your site's authority and what else answers the question.
Can I see why my article failed a check?
Yes. Each failed check is listed in plain words, such as "Your primary keyword is missing from a key position", with the evidence where there is some. The list of failed checks is your edit list.
Do pages I wrote outside RankX AI get these scores?
No. The scores are computed on articles RankX AI generates. Published pages are checked by the Website Audit and AI Readiness instead.
Last updated