Exact Match TF/IDF

Exact Match TF/IDF is a page-local keyword score that returns a small compatibility value: it checks whether any complete line of the raw body equals a keyword variation, then dampens that hit by the log of the page’s word count. It is not textbook corpus-wide TF/IDF, and it correlates with position within a measured result set rather than causing a page to rank.

What it measures

This factor combines a whole-line variation test with an inverse-frequency approximation based on the page’s raw-body word count. In plain terms, it asks a strict yes/no question — does any single, complete line of the body exactly equal one of the keyword variations? — and then scales that answer down according to how much text surrounds it. The result is a “compatibility” nod to the classic term-frequency / inverse-document-frequency idea, computed entirely within the one page rather than across a corpus.

How RankGear measures it

RankGear sets the match count to 1 if any complete source line equals a variation, and otherwise to 0. It computes 1 + ln(count) and divides that numerator by the natural log of the number of 3-to-64-character word tokens found in the raw body. A zero word count returns 0 before any division is attempted.

numerator   = 1 + ln(match_count)      // match_count = 1 if a whole line equals a variation, else 0
denominator = ln(word_token_count)     // tokens 3–64 chars long in the raw body (scripts removed)
score       = numerator / denominator  // word_token_count == 0 → score = 0 (no division)

How to optimize it

Treat this as an observation, not a target. Focus on comprehensive, naturally worded content rather than manipulating source line breaks or chasing this legacy score directly. Because a “match” requires an entire source line to equal a variation verbatim, deliberately shaping line breaks to trigger it produces brittle, unnatural copy with no reliable upside. Write for the reader and let the score fall where it may.

Important considerations

• This is page-local and not corpus-wide textbook TF/IDF — the “inverse frequency” term is approximated from the page’s own word count, not from a document collection.
• No whole-line match can yield a nonfinite numerator, so the score stays bounded even at the edges.
• Markup tokens can contribute to the denominator, since the count runs over 3-to-64-character tokens in the raw body.
• Use more direct keyword and topical metrics for editorial decisions.
• Correlation is not causation: this score relates to position within a measured result set and is a prioritization signal, not a guarantee that a page will rank.

Related factors

• Exact Match Term Frequency
• Exact Matches in A Tags

Part of the Factors reference · how RankGear measures · glossary.