Low text variety: a weak AI-content signal

What it means

A weak statistical signal: this page's sentences are unusually uniform in length and its vocabulary is comparatively repetitive. Machine-generated and templated text often looks this way — but so does plenty of legitimate writing. This is a probabilistic hint, never a verdict that the content is AI-written.

Why it matters

It matters only as a nudge. Google's concern is low-value, scaled, or unoriginal content regardless of how it was produced; uniform, low-diversity prose is sometimes a symptom of that. But AI detectors are genuinely unreliable — they misfire hardest on non-native-English and technical writing — so we deliberately treat this as advisory, cap it at a notice, and only raise it on longer English pages where two independent measures agree.

How to fix it

1. Read the page and ask whether it says anything a reader couldn't get from ten other pages; add original insight, examples, or first-hand experience.
2. Vary sentence structure and length so the writing has a natural rhythm.
3. If the page is genuinely useful and original, you can safely ignore this signal — it is not evidence of a violation.

When it's not a problem

Ignore it for technical reference, specifications, non-native-English writing, or any page you know is original and human-reviewed — those legitimately produce uniform prose and are the documented false-positive cases.

How common is it?

4 audited sites in our corpus currently show this issue. The breakdowns below show which platforms, gatekeepers, verticals and countries are most exposed.