So does Google actually use LSI or Latent Semantic Index(ing)? Technically, the answer is: No. LSI is often an incorrectly used term. However, Google does something very similar and the most recent update to this indexing is called Neural Matching.
John Mueller from Google has stated in a tweet from 2019, “There’s no such thing as LSI keywords — anyone who’s telling you otherwise is mistaken, sorry.”
This means LSI keywords don’t even exist? Then what does Google actually do in order to recognize synonyms, context, and language? Well, that calls for a longer explanation.
When Google Search first became available in 1997 it only gave results based on the exact words you used in your search. In 2009 Google’s Caffeine update overhauled their entire indexing system to improve search results with more relevant search results.
After that followed several Google algorithm updates that are constantly building off of each other and improving the way we search. More recently in 2018 Google introduced neural matching, which is an AI that helps google recognize synonyms and context within searches.
What Are LSI Keywords?
Latent Semantic Indexing is a system created to recognize synonyms and related-keywords in order to give the best search results. SEO and PPC agencies use this system to mimic Google’s way of understanding language and searches.
In 1992 Susan Dumais, a search engineer at Microsoft at the time, published a patent for LSI keywords. This eventually opened the doors of “context” instead of only ranking websites based on exact words in their exact orders. However, at the time, the internet wasn’t what it is today and her patent was mostly used to index known documents.
The idea behind LSI keywords is that it creates relationships between words and phrases. For instance, when someone searches for “red fruit” the search engine will show results for strawberries, raspberries, cherries, and other types of red fruits instead of only showing results for exact word matches for “red fruit”. Yes Google does do this, but there are a number of algorithms and Ai involved in their overall algorithm that is the secret sauce of Google Search.
So, What is Neural Matching?
Neural matching is an AI algorithm that allows Google to understand language much better. This is, in a way, similar to the idea of LSI keywords. For example, when searching for “why is my couch so staticky” the results will show answers, reasons, and solutions instead of just giving results that say the exact phrase in the exact order.
Neural matching uses “super synonyms” and descriptions to give better, more diverse search results. Danny Sullivan of Google says that they impact 30% of queries and help Google better connect words to concepts.
How can I use this to benefit my SEO?
Now Google can understand questions even more thoroughly. This way results that would have been more difficult to find before are ranking higher and are easier to find. Results have become more intuitive so it’s easier to rank higher if you have content that contextually relates to what people are searching. There are a lot of other ranking factors that help your SEO, but this one is pretty big!
You can look at the “People Also Ask” (PAA) section of Google that answers questions related to your search query.
Key Takeaway
Latent Semantic Indexing (LSI) is a patented system from 1992 that doesn’t exist in Google. But even top ranking websites still use the term to describe Google’s recognition of synonyms and contextually related search results when referring to Google’s ranking factors. Google has their own confidential algorithm, that uses context and synonyms to give better results. Neural matching is their newest AI algorithm addition that makes searching more intuitive and can help SEO ranking by showing results with the best contextual relevance.
Conclusion
Just because Google says LSI is not a thing does not mean you should abandon the concept. Google is using synonyms and language grouping to understand content, context, and intent. Smart creators and web developers are still using tactics like Natural Language Processing and LSI to cover topics in their entirety.
