Skip to content
×

Why is it misleading to explain Google rankings in terms of LSI keywords?

Explaining Google rankings in terms of “LSI keywords” is misleading because it conflates semantically related terms with a specific information retrieval technique. Semrush states that Google does not use latent semantic indexing (LSI) for ranking. The presence of related terms in content does not justify attributing ranking changes to LSI, either. A more accurate approach is to discuss how these terms add depth to a topic and clarify context, rather than treating them as a recipe for higher rankings.

LSI is the name of a technique, not a general term for related words

According to Semrush, LSI is an information retrieval method that uses relationships between words identified through latent semantic analysis (LSA) to find information, rather than relying solely on keyword matches. LSA analyzes patterns in text. For example, the English word “bank” may have different meanings in contexts involving finance and savings versus rivers and erosion.

But the ability to analyze meaning does not mean Google uses this particular technique. Semrush describes LSI as an older technique designed for smaller collections of documents and cites statements that Google does not use it. Calling “credit score” and “interest rate” in a credit card article semantically related terms is therefore more accurate than calling them “LSI keywords.” It also avoids implying that we know how a search engine works internally.

Understanding context is not the same as counting related terms

Semrush also discusses the Knowledge Graph, natural language processing, and AI as ways to understand entities, concepts, and context. It uses “Los Angeles to New York” and “New York to Los Angeles” as an example: the words are similar, but the search needs differ. A list of words alone is therefore not enough to fully understand a query.

  • Discussing interest rates in a credit card article can add relevant information.
  • The value of related terms lies in helping convey the content, not in whether they are called “LSI keywords.”
  • These articles provide no rule linking a given number of added terms to a particular ranking improvement.

A 2025 article on Neil Patel’s website also identifies the idea that “adding these terms improves rankings” as a myth and quotes John Mueller’s 2019 rejection of the term “LSI keywords.” That quotation has a specific historical context; it is not a complete account of current ranking rules.

An older article’s testing advice is not evidence of how rankings work

These articles do not entirely agree. A 2023 article on Neil Patel’s website suggests that if an existing keyword strategy is not delivering the expected rankings, one could select a page that already ranks reasonably well, add a small number of what it calls “LSI keywords,” and observe its rankings after it is recrawled. The original article also acknowledges that rankings may rise, remain unchanged, or fall.

This is only a test suggested by the author, not an official Google rule or a verified outcome. Even if rankings rise after the changes, that observation alone cannot prove that Google uses LSI or that the added terms caused the improvement. Filling gaps in relevant information remains a useful approach to content, but ranking problems cannot be attributed directly to “not including enough LSI keywords.”

As a practical example of content work, Procrustes turns questions that genuinely matter to clients into articles, with ongoing content planning, writing, and updates.

LSI keywords and Google rankings: perspectives from across the web

Should you still consider topic-related terms when writing an article?

Yes, but only where the content calls for them, not simply to fill out a word list. Semrush argues that semantically related terms can help an article cover relevant subtopics and perspectives; for example, interest rates may be worth explaining when discussing credit cards. This is a way to improve how a topic is covered, not a guarantee that including particular terms will improve rankings.

If two searches use the same related terms, does that mean they reflect the same need?

Not necessarily. The combination and order of words can change their meaning, so even queries using similar words may reflect different needs. Semrush uses trips between cities in opposite directions to illustrate how Google uses semantic analysis to understand context. Simply checking whether a page contains a set of related terms is therefore not enough to determine whether its content matches search intent.

If rankings rise after related terms are changed, does that prove the page was missing LSI keywords?

No. A 2023 article on Neil Patel’s website suggested that when an existing keyword strategy is not producing the expected rankings, one could modify a page that already ranks reasonably well and observe the results after it is recrawled. It also acknowledged that rankings could rise, remain unchanged, or fall. This provides only a before-and-after observation; the result alone cannot establish that the added terms caused the change, much less prove that Google uses LSI.

References

  1. The Ultimate List of SEO Myths: 21 Misconceptions Debunkedneilpatel.com · 2025-03-24T14:00:00+00:00 — Source used in the article above.
  2. What Are LSI Keywords & Why They Don‘t Matterwww.semrush.com · 2021-01-27T08:44:00Z — Discusses: Should you still consider topic-related terms when writing an article?; If two searches use the same related terms, does that mean they reflect the same need?
  3. 10 SEO Experiments to Grow Your Organic Traffic Performanceneilpatel.com · 2023-06-15T19:00:00+00:00 — Discusses: If rankings rise after related terms are changed, does that prove the page was missing LSI keywords?
Keep the conversation going

Join the discussion 0 comments

What are you working through? Share a question or experience and let’s find the next step.

No comments yet. Start the conversation.

Share your thoughts

← Keep exploring
LET’S TALK

Start with your brand.

Tell me about your website and what you want to achieve.

WhatsApp tony@procrustes.info
Scan to chat on WhatsApp