Question 1

Cosine similarity

Accepted Answer

How closely the meaning of the page matches the meaning of the search, even when the words differ. We turn the search and each passage of the page into embeddings (vectors) and measure how close they are. On the Signals page: A positive correlation means pages that are closer in meaning to the search are ranking higher. This is usually one of the strongest and steadiest signals.

Question 2

BM25

Accepted Answer

The classic keyword-match score: how often the search terms appear in the page's body copy, weighted by how rare those terms are and adjusted for page length. Repeating a word over and over adds less and less. On the Signals page: A positive correlation means pages using the searcher's actual words are ranking higher.

Question 3

Title BM25

Accepted Answer

The same keyword-match score, measured on the page's title tag only. On the Signals page: A positive correlation means titles that use the search terms are ranking higher.

Question 4

Phrase coverage (trigram)

Accepted Answer

The share of the search's three-letter character pieces that appear in the page title. It catches partial matches like plurals and word variants that exact-word matching misses. This measure comes from the Yandex source code leak. On the Signals page: A positive correlation means titles that closely echo the search, including close variants, are ranking higher.

Question 5

Hybrid relevance (RRF)

Accepted Answer

One combined relevance score. It blends where a page stands across SERPrecon's relevance metrics using reciprocal rank fusion, similar to how search engines combine many signals into one ranking. On the Signals page: A positive correlation means pages that do well across relevance measures at once are ranking higher.

Signal	What it measures	Good sign
Relevance
Cosine similarity	Meaning match between page and search	Positive
BM25	Keyword match in the body copy	Positive
Title BM25	Keyword match in the title tag	Positive
Phrase coverage (trigram)	Close-variant match in the title	Positive
Hybrid relevance (RRF)	All relevance metrics combined	Positive
Intent
Title intent match	Title matches the results' main intent	Positive
Copy intent match	Body copy matches the results' main intent	Positive
Content
Word count	Number of words on the page	Either
Readability	How easy the copy is to read	Positive
Topical coherence	How tightly the page stays on topic	Positive
Title/body alignment	Title describes what the page covers	Positive
Commodity score	How much the page repeats competitors	Negative
Topic Coverage
Bigram coverage	Shared two-word phrases used	Positive
Keyword coverage	Shared keywords used	Positive
Entity coverage	Shared entities mentioned	Positive
Authority & Quality
PageRank	Estimated link authority (0 to 100)	Positive
Spam score	How repetitive or templated the text is	Negative
Structured data	Has schema.org markup	Positive
Page Speed
Largest Contentful Paint (LCP)	Time until the main content appears	Negative
Interaction to Next Paint (INP)	How fast the page responds to input	Negative
Cumulative Layout Shift (CLS)	How much the layout jumps while loading	Negative
First Contentful Paint (FCP)	Time until anything appears	Negative
Time to First Byte (TTFB)	Server response time	Negative

SERPrecon Signals, explained

All 23 SERPrecon semantic relevance signals at a glance

Semantic Relevance signals

Intent signals

Content signals

Topic Coverage signals

Authority & Quality signals

Page Speed signals

Correlation isn't causation

Insights-powered Solutions

Get Started with SERPrecon