Numeric Evidence Sentence Count totals sentences that pair a number with measurement or evidence language — cues such as percent, study, survey, sample, data, measured, tested, average or rate. Decorative numbers and unverified claims can match, so human verification is still required.
| Factor ID | RG-AIX-023 |
|---|---|
| Family | AI access & extractability |
| Measurement | Count |
| Measured zone | Rendered sentences |
What it measures
This factor approximates how much of a page is quantified claim rather than general assertion. It is a lexical proxy for evidence density, useful for comparing a page with the field rather than for judging any individual sentence.
How RankGear measures it
RankGear counts sentences containing a digit together with a conservative evidence cue such as percent, study, survey, sample, data, measured, tested, average or rate.
How to optimize it
Use accurate, relevant numbers where they help readers evaluate claims, and explain methods, units, dates and sources rather than adding unsupported statistics. A page that raises this count with invented figures has made itself worse in the only way that matters.
Important considerations
- Decorative numbers and unverified claims can match the pattern, so the count overstates genuine evidence.
- Human verification remains required — the detector reads form, not substance.
- A low count can be correct for topics that are not quantitative.
- This is a measured correlation within a result set, not a guarantee: the value relates to how the pages ranking for your query are built — it does not, on its own, make a page rank.
Related factors
Part of the Factors reference · how RankGear measures · glossary.