LSI Keywords: What They Are, What Google Says, and What to Use Instead

What are LSI keywords?

LSI keywords are words and phrases said to be semantically related to a main keyword. For a page about "apple pie", they would include "crust", "cinnamon", and "baking time". The idea is that search engines use these related terms to understand what a page is about. Google does not use LSI, but related terms still help writers.

In practice, when someone uses the term they mean one of three things. Synonyms, like "car" and "automobile". Co-occurring words, like "crust" next to "pie". Or words that resolve ambiguity, like "fruit" or "iPhone" next to "apple". Each of these is a real and useful idea. None of them needs the LSI label.

The honest name for all of them is related terms or semantic keywords. That is the term we use for the rest of this guide. If you came here to learn how to find them, you will still get that. Most related keywords SEO advice lands on the same method. Only the story behind it changes.

What does LSI stand for, and where did it come from?

LSI stands for latent semantic indexing. It is a real technique from information retrieval, the science of finding documents in a collection. Researchers at Bellcore, including Susan Dumais and Scott Deerwester, patented it in 1988 and described it in a 1990 paper called "Indexing by Latent Semantic Analysis".

The method builds a big table of which words appear in which documents. Then it uses math, a technique called singular value decomposition, to squash that table into a smaller set of hidden "concepts". Two documents that share concepts are treated as similar even if they use different words. That was a smart way to handle synonyms in a small, fixed set of documents.

Two limits matter for SEO. First, LSI was built for collections of a few thousand documents, not a web of trillions of pages that changes every second. Second, it dates from before Google existed. The SEO version of the term appeared around 2010, borrowed the name, and dropped the math.

Does Google use LSI keywords?

No. Google's John Mueller answered this directly on Twitter in 2019.

There's no such thing as LSI keywords.

Google does understand meaning, of course. It has used systems like RankBrain, BERT, and MUM to understand what words mean in context. Google explained BERT in a 2019 blog post as a way to grasp the full context of a word by looking at the words around it. None of that is LSI. It is far newer and far more capable.

So why do related terms still show up in ranking pages? Because a page that fully covers a topic naturally uses them. A recipe page without "crust" or "oven" is probably not a complete recipe. Google rewards the completeness. The words are a side effect of good content.

LSI keywords examples (better called related terms)

Here are LSI keywords examples for a few main keywords, sorted into the three types that matter. Notice how the "disambiguation" column is often the most important. It is how a reader, and a search engine, can tell which "apple" or "python" you mean.

If someone asks for an example of SEO keywords in general, the same table works. "rank tracker" is a head keyword. "rank tracking tool" is a variant. "daily rank tracking for agencies" would be a long tail keyword. Related terms sit around all of them.

How to find LSI keywords (related terms), step by step

You do not need a special generator. Google shows you the related terms for free, and a keyword tool speeds up the collecting. Here is the honest way to find them.

1. 1

   Search your main keyword and read the top 5 pages

   Open the top results and skim the headings. Write down every subtopic that appears on three or more pages. Those are the things Google already considers part of a complete answer. This one step usually finds half your related terms.

2. 2

   Pull Google autocomplete suggestions

   Type your keyword into Google and note the suggestions. Then add a letter or a question word and repeat. Our free Google Suggest tool does this in bulk and adds volume and trend data, so you can collect hundreds of suggestions in a minute.

3. 3

   Collect the People Also Ask questions

   The People Also Ask box on the results page lists questions searchers ask about your topic. Each one is a related term and often a heading for your page. See our guide on People Also Ask SEO for how to mine them.

4. 4

   Check the related searches at the bottom of the page

   Scroll to the bottom of the results page. The related searches block shows what people search next. Add anything relevant to your list. These often reveal the angle your competitors missed.

5. 5

   Expand with a keyword research tool

   Put your main keyword into a keyword research tool and pull the related keyword ideas with their volume and intent. Zutrix shows volume, difficulty, CPC, trend, and search intent for each idea. Sort by volume and keep the terms that describe a real subtopic of your page.

6. 6

   Group terms into sections

   Sort your list into 5 to 10 groups. Each group becomes a heading or a paragraph in your page. Terms that do not fit any group are probably a different topic and belong on a different page.

7. 7

   Write, then check for gaps

   Write the page for the reader first. Then compare your draft with the list. If a group is missing, ask whether the page is incomplete without it. Add a real section if yes, and skip it if no.

How to use related terms in your content

The goal is a page that reads like it was written by someone who knows the subject. That person uses the right words without trying. Here is how to get the same effect on purpose.

For the full writing process, from brief to publish, read our guide on how to write SEO content. Related terms are one step in that process.

Should you use an LSI keyword generator?

Tools like LSIGraph and various free LSI keyword generators are popular. They are not doing latent semantic indexing. Most pull Google autocomplete, related searches, and the words on top ranking pages, then present the list under the LSI name. That is the same data the steps above collect.

That does not make them useless. A generator that saves you 20 minutes of clicking is worth having. Just read the output with judgment. Half the terms will be noise, some will belong on a different page, and a few will be gold. The tool cannot tell which is which. You can.

One more distinction. Related terms are not the same as long tail keywords. A long tail keyword is a specific, lower volume search you might target with its own page or section. A related term is a word that belongs in a page about a topic. They overlap, and our guide on how to find long-tail keywords shows how to work with the first kind.