The Exact Match Clustering Score is a legacy composite that describes where the exact search phrase appears and how tightly its occurrences are spread across a page’s cleaned body text, returned as a single number. It is a descriptive distribution statistic — not a target to maximize — and can come back non-finite when the phrase is absent, because the formula divides zero by zero.
| Factor ID | RG-SRP-003 |
|---|---|
| Family | Term Position Distribution |
| Measurement | Composite position/spread score |
| Measured zone | Cleaned, lowercased, trimmed body content |
What it measures
This factor reports a legacy clustering score derived from the positions of the exact search phrase within a page’s body. Rather than counting how often the phrase appears, it summarizes where those occurrences sit and how far apart they are, folding position and spread together into one composite number. Because it is built on exact-phrase offsets, it responds to how a page threads the target phrasing through its content, not to synonyms or partial matches.
How RankGear measures it
RankGear locates each occurrence of the exact phrase and records its offset, then computes the mean of those offsets, their population variance, and the UTF-16 length of the cleaned body. Those three quantities are combined into the composite score with the formula below. When the phrase never appears, there are no positions to average, so the calculation divides zero by zero and can return a non-finite result.
score = 100 × (mean / body_length) × min(1, population_variance / mean)How to optimize it
Treat this as an observation, not a target. Structure content for readers: establish the subject early, develop related sections in a logical order, and repeat the exact phrasing only where it genuinely aids clarity. Do not engineer phrase clusters solely to move this composite score — the number describes a distribution, and there is no value that is universally “good.”
Important considerations
- Positions are UTF-16 offsets — not words, sentences, pixels, or HTML byte positions.
- The cleaned body content excludes markup and may differ from the raw page source.
- The generated variation set feeds variation-based factors, so changes here can ripple into those signals.
- These are descriptive distribution statistics: lower or higher is not universally better.
- A missing phrase yields no positions, which can produce a non-finite score rather than a zero.
- This is a correlation-based signal within a measured result set. It relates to how a phrase is positioned; it does not on its own cause or guarantee a ranking.
Related factors
Part of the Factors reference · how RankGear measures · glossary.