What is an LSI keyword?
LSI keywords are the words that naturally appear around a topic. A page about video production will also mention cameras, editing, crews, pre-production, color grading and delivery, and those surrounding terms are what SEO practitioners call LSI keywords. The initials stand for latent semantic indexing, a technique from information retrieval in the late 1980s that analyzed which words tend to occur together across a body of documents in order to infer meaning. The SEO industry borrowed the name to describe related terms, and it stuck, although the connection to the original technique is loose.
In practice the term means something simpler than its name suggests: the vocabulary a knowledgeable writer uses when covering a subject properly. Synonyms, subtopics, related concepts, the questions people ask, the tools and names that belong to the field.
Why it matters
Why LSI keywords matter
Search engines no longer match pages to queries word for word. They try to understand what a page is about and whether it answers the searcher's intent, and the surrounding vocabulary is one of the signals they use. A page that repeats its target keyword without the language of someone who knows the subject looks thin; a page that covers the related concepts reads as authoritative, to the algorithm and to the reader. LSI keywords, whatever they are called, are a practical way of making sure a page has that depth.
They also protect against the oldest SEO mistake. Writers who focus on a single keyword tend to repeat it until the page is unreadable, which is keyword stuffing and which search engines penalize. Writing with the full vocabulary of the topic produces better content and better rankings at the same time, because the two are no longer different things.
In practice
How LSI keywords work in practice
- 01
Finding LSI keywords is mostly research. The quickest sources are the search engine itself: autocomplete suggestions, the "related searches" at the bottom of the results page and the "people also ask" box all show the language searchers use around a topic. Keyword tools group related terms by topic, and reading the top-ranking pages for a query reveals what they cover that a draft does not. For specialist subjects, the best source is the person who knows the field, because they already use the vocabulary without thinking.
- 02
Using them is a matter of writing well. The related terms go where they belong, in headings, in body copy and in the questions the page answers, at the density a natural explanation would produce. A page on keyword research will mention search volume, difficulty, intent, long-tail terms and the tools involved because it has to, not because a checklist said so. The aim is complete coverage of the topic, which is what topic research and content planning are for, with the related terms as a check that nothing has been left out.
- 03
The method scales into site structure. Pages that cover related subtopics and link to each other form a cluster that search engines read as expertise on the whole subject, which is the thinking behind pillar pages and topic clusters. A glossary like this one is an example: each entry covers one term in depth and links to the terms around it, so the whole reads as a map of the field rather than a list of definitions.
- 04
The practical hand-off is a content brief. The strategist gives the writer the subtopics, questions and concepts a complete treatment has to cover, drawn from the top-ranking pages and the field itself. The draft is checked against that brief: covered, partly covered, missing. After publication, the query report in Search Console shows which related searches the page draws, and the gap between that and what it was meant to cover becomes the next edit.
Watch out
What people get wrong about LSI keywords
Believing that Google uses LSI. It does not, and has said so; the original technique is decades old and modern search uses far more sophisticated language models. What survives is the useful idea underneath: cover the topic with its real vocabulary. The second mistake is treating LSI keywords as a list to be inserted. Dropping twenty related terms into a page that does not actually discuss them is stuffing by another name. The third is confusing related terms with synonyms. "Film production" and "video production" are synonyms; "pre-production", "call sheet" and "color grade" are the related concepts that show a page knows what it is talking about.
