Google LSI – Is Latent Semantic Indexing A Bit Of A Myth?

For those of you who aren’t aware LSI stands for ‘Latent Semantic Indexing’, many marketers jumped on this bandwagon not so long ago claiming Google were using it to identify the meaning of words.

I am not going to go into the maths of LSI but will try and give a brief explanation.

Semantics involves assessing the meaning of words and not just the spelling of it. It does this by highlighting and recording word relationships linking up millions of words.

For example; Plane – Jet (Two different words that often mean the same thing) Car – Automobile? I am sure you get the idea.

‘Latent’ means buried, implicit, hidden or not explicit.

LSI is about finding the real meaning of words, finding which words relate and then indexing a large amount of documents by relevance.

Soccer -> Relates to Football (in the UK)

Wing -> Relates to Bird or Plane

LSI takes care of all this and indexes appropriately.

The three main components of LSI are;

Singular v Plural

– car / cars
– place / places
– baseball / baseballs

Changes in Tense

– grow, growing and grown

– playing – play – played

– work – works – working

Same meaning different spelling

Pupil – Student

Buy – Purchase

create – make

I think you’re getting the idea, LSI is awesome!!! But are Google using it???

Let me start by saying that Google have not patented anything relating to LSI, but I am not going to write extensively on this as it is, well, boring. The point is the fact that it supposedly exists in the Google algorithm is a bit of a mystery.

If you want to find out more about LSI or how it works I suggest you watch the video at Stompernet. It explains the background to LSI in a lot more detail and how the technology works.

I am an SEO so all I want to know is does LSI figure in the ranking algorithm? If it doesn’t I don’t care regardless of how great it is;

The only way to find out for sure is to test it.

The Test

Let’s take a look at this table of results;

Google LSI

As you can see from the chart all the keywords mean either the same thing or something very similar, however there are massive difference in the results.

I’ll leave it to you but do you really think Google understands the meaning of words like a human does??

Just to clarify the above graph shows the difference in the total search results returned and when pages in the top 5 matched between the keywords.

What Is Google Actually Doing?

If I knew that I would be a very rich man, however the other extreme to LSI is simple word recognition, the site with the most keyword mentions wins? We know this isn’t how it works from my post early last week about on page optimisation issues.

It used to be enough to create a site with the same keyword pasted throughout multiple pages however in the modern world of search diversity is key. You need to build a domain that is relevant to a niche and not a keyword, use variations, use Google’s keyword tool to understand what other keywords are used with your main phrase.

It is clear from my home page that I target certain keywords, however I don’t consistently use them in every page title, I use multiple variations and the same goes for my anchor text.

I use a range of keywords in the anchor text;

– search engine optimisation
– google seo
– google search engine optimisation
– seo
– seo techniques
– seo tutorials
– Search engine marketing
– Internet marketing


They don’t all ‘mean’ the same thing but a person searching SEO is no doubt looking for tutorials, advice or services of some sort; they are all keywords a particular type of searcher will use.

So when were thinking about what keywords to use in the content on our site we should be thinking in terms of search links, think about how Google target their adwords, not just on the keyword entered but on the search trends of the user. Using search trends it is easy to see how Google can spot an unnatural site.

What I really want you to get from this post

There are hundreds of search theories out there, stuff that the ‘SEO Guru’s’ discuss and say ‘may’ be a ranking factor, people then jump on the bandwagon and before you know it we have a new (100% fabricated) ranking algorithm. The truth is ranking on Google is about being smart with keyword selection and implementation. If you have a site about super cars and you use the keyword “super cars” there is a good chance you will rank for it, you will not rank for “fast cars” if you never use the keyword on the basis they mean pretty much the same, it’s a myth.

If you want to rank you use your main keyword in the title tag, you create content using different ‘search’ variations, you build links to all your content using varied anchor text, that’s it, read the theories if they interest you but don’t ‘read’ too much into them.

You may also want to read Google Semantics, An argument for LSI and Another LSI Myth Buster, all are great reading.

I have a busy week a head posting this so posting might be a struggle. I have also a large queue of emails in my inbox and I will try and get back to everyone as soon as possible. Please don’t stop emailing me with questions as I enjoy helping out and that is what this blog is all about, learning how to really do SEO.