How should the SEO effects of so-called LSI keywords be tested?

When testing so-called “LSI keywords,” first establish which terms the source means, then observe ranking changes after adding those terms, rather than assuming Google uses LSI. The available material suggests editing a page, waiting for it to be recrawled, and comparing rankings, but provides no experimental results demonstrating effectiveness. Keep in mind that filling gaps in the content, changing wording, and testing for ranking changes are three different things.

Define the term list first—do not treat a tool’s label as a search mechanism

Semrush’s 2024 article describes what are often called LSI keywords as terms related to the target keyword but not synonymous with it. For example, when discussing credit scores, “credit limit” is a related concept, while “credit rating” is a synonymous expression. The article rejects the notion of LSI keywords and reports that Google has confirmed they do not exist; the supplied material does not include Google’s original statement.

The same article still emphasizes semantics and search intent, and introduces a tool’s Related filter for finding terms related to a topic. Before testing, therefore, record which tool and feature the term list came from, along with the reasons for choosing the terms. Do not assume that these terms correspond to a particular ranking mechanism simply because they carry the “LSI” label.

Neil Patel’s website suggests comparing a single page’s rankings before and after an edit

A 2023 article on Neil Patel’s website takes a different position: if an existing keyword strategy is not delivering the expected rankings, it may be necessary to add what it calls LSI keywords. The article suggests finding terms through Google’s autocomplete suggestions, choosing a page that already ranks well, and adding a small number of terms. Then, wait for Google to recrawl the page and check whether its rankings rise, remain stable, or fall.

This is the article’s proposed test, not an official rule or proof that the approach works. The article says the added terms should supplement existing content rather than replace the primary keywords; it provides no fixed waiting period, appropriate number of terms, or threshold for success.

Record content additions separately from wording changes

To make these observations easier to interpret, use the following suggestions to keep track of your edits. These are recordkeeping suggestions, not experimental standards that the sources claim have been validated:

• Keep the pre-edit version: Record the page, the queries and rankings being monitored, the added terms, and the date of the changes.
• Identify the nature of each change: Adding an explanation of “how credit limits affect credit assessments” supplements the content; merely substituting synonymous expressions is primarily a wording change. Record the two types of changes separately.
• Avoid confusing the effects of different changes: If you also change the title, rewrite paragraphs, or add information, record those changes too. You cannot attribute the results of the entire edit solely to the added terms.

A ranking change does not prove that LSI works

After the page is recrawled, you can describe how its rankings changed for the specified queries, but a single before-and-after comparison is not enough to isolate the effect of the added terms. If the page also became more comprehensive, you cannot conclude that the terms themselves produced the effect. A more accurate conclusion is “ranking changes were observed after this content update,” rather than “Google’s LSI mechanism rewarded these terms.” A lack of improvement does not mean that all related terms are useless, either.