Latent Semantic Analysis (LSA)
Latent Semantic Analysis is an older way for computers to find which words and documents are related, by spotting words that tend to appear together.
Latent Semantic Analysis, or LSA, is a technique for working out which words and documents are related by looking at which words tend to appear together. If "espresso," "crema" and "portafilter" keep showing up in the same documents, LSA learns that they belong to the same subject, and that a document using them is about espresso even if it never says "coffee machine." It's also called Latent Semantic Indexing, or LSI, when it's used to search documents.
It's a research method from the early days of making computers understand text. In SEO, it's mostly known through a myth.
The "LSI keywords" myth
For years, SEO tools and articles have sold lists of "LSI keywords": related words you're told to sprinkle into a page because Google supposedly uses LSA to rank it. Google has never said it uses LSA, and the lists are usually just related searches or synonyms.
The useful part of the advice is plain good writing. A page that covers a subject properly uses the words that go with it. For example, Kestrel Roasters' guide to pour-over coffee will naturally mention grind size, bloom, water temperature and filters, because you can't explain pour-over without them. No list required.
What replaced it
The idea behind LSA, that meaning shows up in which words keep company, lives on in much more capable forms. Today's search engines and AI models turn text into embeddings, lists of numbers that place passages with similar meanings close together. That's how an AI platform can match a question to a passage that answers it in different words.
What it means for your pages
Write each page to answer a real question fully, in the words your customers use. Cover the parts of the subject a reader would expect. Put the direct answer near the top, so the passage a platform picks up actually contains it.