Variation Density in Top 30KB reports how densely your keyword’s variations appear across the opening ~30,000 characters of the page plus its image alt text, expressed as a percentage of whole-word matches over the word-pattern count. It is a descriptive observation of early-page topical concentration, not a target to inflate — density that outpaces readable prose is a warning, not a win.
| Factor ID | RG-KWD-030 |
|---|---|
| Family | Keyword Usage & Density |
| Measurement | Percentage |
| Measured zone | Google-density top-30KB corpus plus image alt text |
What it measures
This factor reports keyword-variation density within RankGear’s compatibility corpus for the beginning of the page. Rather than counting exact keyword repeats, it looks at the variations of your target term and gauges how concentrated they are across the page’s opening region — the zone a reader (and a crawler) encounters first — combined with the text carried in image alt attributes.
How RankGear measures it
RankGear extracts alt text from the full HTML, strips selected elements, and truncates the remaining source to 30,000 UTF-16 units. It then appends all collected alt text, divides the count of whole-word variation matches by its word-pattern count, and multiplies by 100. A page with a zero word count returns 0.
density = (whole-word variation matches / word-pattern count) × 100
corpus = strip(HTML) truncated to first 30,000 UTF-16 units
+ all collected image alt textHow to optimize it
Treat this as an observation of early-page focus, not a number to push. Establish strong topical focus early in the page, use accurate and nonspammy image alt text, and let natural supporting terminology carry the variations. Writing clearly for the reader is what produces a healthy figure here; chasing the percentage directly tends to degrade the prose the score is meant to reflect.
Important considerations
- Alt text collected from beyond the 30,000-unit truncation point can still be appended to the corpus, so images late in the page influence the figure.
- Some markup tokens remain in the denominator, so the word-pattern count is not a pure prose word count.
- This corpus is defined by a byte-length truncation and differs from the visible first-screen (above-the-fold) content a reader actually sees.
- Do not target density at the expense of readability.
- Correlation is not causation: a value here tracks how a page relates to others within a measured result set. It is a prioritization signal, not proof that a given density makes a page rank.
Related factors
Part of the Factors reference · how RankGear measures · glossary.