Latent Semantic Indexing (LSI)
Latent Semantic Indexing (LSI) is a mathematical technique from 1980s information retrieval research for finding relationships between terms and concepts in a body of text — it is not a technique Google has confirmed using in its modern ranking systems.
Updated
November 12, 2026
Category
SEO Glossary
The term 'LSI keywords,' widely used in SEO content and tools to mean 'related terms you should include for semantic relevance,' is something of a misnomer — Google has stated it doesn't use LSI as a ranking technique, and the term as commonly marketed doesn't accurately describe Google's actual modern methods, which rely on far more advanced natural language processing and neural techniques like BERT and later transformer-based models. That said, the underlying advice usually attached to 'LSI keywords' — write comprehensively using related terms, synonyms, and topically connected vocabulary rather than repeating one exact phrase — is genuinely sound practice, even though the technical justification (LSI specifically) is outdated. It's worth knowing the term is commonly used loosely rather than technically accurately.
Example
A content brief calls for 'LSI keywords' around a target term — in practice this just means including naturally related vocabulary and subtopics, not literally applying 1980s-era latent semantic indexing math.
Related terms
Related Reading
404 Error
A 404 error is the standard HTTP status code returned when a requested page doesn't exist on the server — the classic 'page not found' response.
A/B Testing (SEO)
A/B testing in SEO is the practice of showing two versions of a page element — usually a title tag, meta description, or on-page element — to see which one earns a better click-through rate or ranking outcome.
AI Citation
An AI citation happens when an AI assistant like ChatGPT, Perplexity, Claude, or Gemini references or links to a specific page as the source for part of its answer.
