LSI keywords: why Google doesn't use them, and what works
Short answer Google doesn't use LSI keywords. Latent semantic indexing is a document-retrieval method patented in 1988, and Google's list of its ranking systems never mentions it. Google matches meaning with AI systems such as RankBrain, neural matching and BERT. What works: answer the questions your customers ask, in their words, instead of adding a tool's word list.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callDoes Google use LSI keywords?
No. Google publishes a guide to the systems it uses to rank pages. It lists BERT, neural matching, RankBrain, MUM, spam detection and more. Latent semantic indexing (LSI) is not on it, and the words "LSI" and "latent semantic" don't appear anywhere on the page (Google Search Central: A guide to Google Search ranking systems).
The system that does the job people credit to LSI is RankBrain:
"RankBrain is an AI system that helps us understand how words are related to concepts. It means we can better return relevant content even if it doesn't contain all the exact words used in a search, by understanding the content is related to other words and concepts." (Google Search Central: A guide to Google Search ranking systems)
So a list of "LSI keywords" from a tool is a list of related words, nothing more. Google doesn't need you to add them to understand your page.
| Myth | What Google says |
|---|---|
| Google ranks pages with LSI | Its ranking systems guide names no LSI system |
| You must add related words so Google gets your topic | Its language systems connect words to concepts on their own |
| More LSI keywords, better ranking | Repeating words until they sound unnatural is listed as keyword stuffing in its spam policies |
What is latent semantic indexing, really?
It's a math method for searching a fixed set of documents, invented before the web. Seven researchers at Bell Communications Research filed the patent in 1988. Its abstract explains the idea:
"The information is treated in the statistical domain by presuming that there is an underlying, latent semantic structure in the usage of words in the data objects." (Google Patents: US4839853A)
In plain words: count which words tend to show up together across many documents, then use those patterns to find a document even when it uses different words from the search. The patent was granted in 1989 and is listed as expired (Google Patents: US4839853A).
The idea was real and useful in its day. The leap SEO blogs made was assuming Google uses this particular method, then turning it into a tactic: sprinkle in the words that "go with" your keyword. Google has never said it uses LSI.
What does Google use instead of LSI?
AI language systems, launched over the past decade. Google's Pandu Nayak described them in 2022:
| System | Launched | What it does, per Google |
|---|---|---|
| RankBrain | 2015 | "the first deep learning system deployed in Search"; links words to concepts |
| Neural matching | 2018 | Matches "fuzzier representations of concepts in queries and pages" |
| BERT | 2019 | Reads how combinations of words express meaning and intent |
(Google blog: How AI powers great search results)
On top of these, Google describes a synonym system:
"...our sophisticated synonym system that allows us to find relevant documents even if they don't contain the exact words you used." (Google Search: How Search works, ranking results)
Its example: someone searches "change laptop brightness", the maker wrote "adjust laptop brightness", and Google connects the two. The same kind of matching can connect a furnace page to a search for "heater repair", without a word list.
So what are related terms good for?
Covering the topic, not feeding an algorithm. Google says it checks whether a page has relevant content beyond the keyword:
"...algorithms assess if a page contains other relevant content beyond the keyword “dogs” - such as pictures of dogs, videos, or even a list of breeds." (Google Search: How Search works, ranking results)
And its self-check for content asks: "Does the content provide a substantial, complete, or comprehensive description of the topic?" (Google Search Central: Creating helpful content)
Related words show up on their own when you answer the questions a buyer has. A water heater page that covers tank vs tankless, sizes, price ranges, how long the install takes and what happens to the old unit will naturally say "gallons", "gas line", "permit" and "haul away". That's the coverage an "LSI" list tries to fake.
| An "LSI keyword" list gives you | Customer questions give you |
|---|---|
| "best water heater", "water heater types", "hot water tank" | "How much does a tankless install cost?" |
| Words to sprinkle in | A section to write |
| The same list every competitor's tool pulls | Answers only you can give: your price, area, timing |
How do I find the related terms that matter?
Start from your customers, not a tool. Five free sources:
- Calls and emails. Write down the questions people ask before they book. Each one is a heading or a sentence on your page.
- Search Console. Its Performance report shows the searches your site already appears for. See Google Search Console for a small business.
- Google's search box. Type your service and read the suggestions, then the "People also ask" box. See featured snippets and People Also Ask.
- Your own words. If customers say "AC" and you say "cooling system", use theirs. The SEO Starter Guide's example: some people search "charcuterie", others "cheese board".
- Free keyword tools. For more, see keyword research for a local business.
Then answer each question where it fits. Don't chase every variation. Google's Starter Guide says its "language matching systems are sophisticated and can understand how your page relates to many queries, even if you don't explicitly use the exact terms in them."
Can adding LSI keywords hurt?
It can, if you force them in. Google's spam policies list "Repeating the same words or phrases so often that it sounds unnatural" as an example of keyword stuffing (Google Search Central: Spam policies). A paragraph built to fit "plumber near me, best plumber, plumbing services, local plumber" reads like that to a customer, too. How keyword stuffing is judged, and what happens after, is in keyword density.
Example
Illustrative cases with invented numbers.
An HVAC company's AC repair page (US dollars)
A tool hands over 40 "LSI keywords" for "AC repair". The owner instead lists the six questions callers ask: cost to diagnose ($89 here), same-day service, repair or replace, which brands, warranty, and service area. Six short sections. The page now says "compressor", "refrigerant" and "SEER" because the answers need them.
A bookkeeper near Vancouver (Canadian dollars)
Her page repeats "bookkeeping services Vancouver" plus a pasted list of related phrases. She cuts the list and adds what clients ask: monthly price from C$300, what she needs each month, GST filing and year-end handoff to the accountant. Fewer keywords, more answers.
A dental clinic adding a new service
The team wants to rank for "Invisalign". Rather than buy a word list, they add the questions from the front desk: cost range, number of visits, whether insurance helps, and Invisalign vs braces. Each becomes a heading.
FAQ
What are LSI keywords?
A name SEO tools and blogs give to words related to your main keyword. It borrows "latent semantic indexing", a 1988 retrieval method that Google doesn't list among its ranking systems.
Are LSI keywords the same as synonyms?
Tools usually mix synonyms with related phrases. Google says its synonym system already connects different words with the same meaning, so you don't need to list them.
Do I need an LSI keyword generator?
No. Your customers' questions, Search Console and Google's own search box give you better related terms for free.
Does LSI matter for AI Overviews?
No. Google says there are "no additional requirements to appear in AI Overviews or AI Mode" beyond normal SEO best practices (Google Search Central: AI features and your website). Clear answers to real questions help everywhere.
Different for you?
- A plugin is flagging keyword use: keyword density.
- You don't know what customers search: keyword research for a local business.
- You're rebuilding a service page from scratch: on-page SEO for service pages.
- You want to see which searches you already show for: Google Search Console for a small business.
- You'd rather have pages researched and written for you: SEO & GEO Agent.