The Definitive Guide to Latent Semantic Indexing (LSI)

What Is Latent Semantic Indexing?

it’s a kind of old hat for most search engine optimization specialists these days because it’s been swallowed up in Google’s Rankbrain AI machine learning code.

Latent means hidden or dormant or not yet revealed, so the idea is that any keyword (or key phrase) has other meanings that may not be so obvious, but are connected to their surrounding sentences and article topic.

In other words, if I use the word ‘pants’, it could have multiple meanings depending on the topic – or surrounding references – or the way it’s being used figuratively.

From this, you can start to appreciate just how clever we humans are! We can decipher all sorts of meanings even from a simple expression on someone’s face without a word being spoken.

Why Does LSI Matter?

So the idea of LSI and why it matters is to make some sense of the context of a piece of writing in order to decide its category and sub-category, and in particular, where it may be useful in terms of an answer for some search query.

Using our example word ‘pants’, if someone searches for ‘incontinence pants’ and Google shows up a page entitled “Why these new bicycle tires are pants”, you can see there’s going to be a mismatch (and Google will get voted down for this answer by the searcher – who at some point may decide Google is just no good anymore).

When it comes to writing articles that answer problems, you need to make quite sure that every word you use is semantically correct for the topic you are writing about.

Having said that, there’s no problem using figurative words, metaphor or crazy synonyms (such as ‘sick’ meaning’ good’ for example), it’s just that you may confuse Google and other search engines in the process – and so your article may not get as high a ranking as some competitor.

Latent Semantic Indexing and Search Engine Optimisation (SEO)

And from that, it’s simple to deduce that using semantically linked words and phrases that make sense is going to help you optimize your articles for more than poetic prose just for the sake of it.

But that doesn’t mean losing your writing “voice” you may have spent years developing, it just means ensuring that you sprinkle the right LSI keywords and phrase throughout your articles where it makes sense to do so – and perhaps replacing some less obvious phrases with something more helpful – at least from the reader’s perspective.

What are LSI Keywords?

We’ve covered some examples already, but what do LSI keywords and phrases look like?

To answer that, you need to know something about maths and vectors, and how an algorithm can be generated that will make some sense of a whole bunch of connected words in an article.

And the best way to start that is to think about intent. What is an article trying to do? Who is it trying to serve? What form is it taking to do this?

All these things matter. But what matters more is clarity. As we’ll see in a moment.

The bottom line is to ensure your sentences make sense. That they answer the problem that the article promises to answer in its title.

Google Keyword Planner

Before we go down that path, let’s see what Google has to say about it using their Keyword Planner tool.

Or rather, let’s not!

The problem with Google’s planner tool is that it’s all about advertising, not article writing. it’s the only thing left in search that still has some meaning when it comes to keywords.

And that’s because it’s easier for Google right now to have people bid on specific sets of keywords than it is to hope that somehow LSI-related keywords will get picked up when advertising using CPC (cost per click).

This is why Google tries to be helpful by offering suggestions, but you will find in practice it’s better to figure out the whole journey of your prospect in the first place, then ask them what they’re actually searching for, and use that data to plan your paid advertising campaigns.

As Google says, their keyword planner is only a guide. No results can be guaranteed. And worse still, every competitor is also using it.

too see A Guide to Analyzing Competitor Backlinks to Build Your Own Strategy

latent semantic indexing
The Definitive Guide to Latent Semantic Indexing (LSI) 3

Rankings and Traffic

It’s vital that you understand the difference between ranking an article and getting traffic from it.

It’s not hard to rank an article, it just depends on what it is ranking for.

And every article that’s worth reading isn’t just ranking for 1 key phrase. They’re often ranking for hundreds of phrases.

If your article is perfectly optimized using LSI keywords and phrases, then it has the best chance of ranking for hundreds of keywords, and if your intent in the article is obvious, eg. if it was to get someone to buy something, then a search engine like Google will pick up on this semantically, and start showing ads on the search results for it.

Discovering your LSI status for an article is vital if you’re going to understand what Google thinks of it (and remember that it’s Google that decides where your article fits semantically in its index – not you).

So the ideal article will rank at the top of Google for its main content, and also rank for 100 to 500 other words as well. It’s these other words (that may or may not be long-tail) that bring in the traffic.

LSI Keyword Examples

Let’s use the article you’re reading right now (ie. this one) to see some examples.

We know the article is all about Latent Semantic Indexing. We also know that it is abbreviated LSI (which you can see in the very first headline).

Google understands that LSI = Latent Semantic Indexing – the hidden (latent) meaning in that abbreviation is already in Google’s database of synonyms, polynyms, acronyms, and abbreviations.

And you can see that the above paragraph is very clear about its meaning. There’s also alternative words and phrase that can connect the contents of an article together even further.

For example, Latent Semantic Analysis (LSA) is another term used, which may or may not be similar depending on certain academic criteria. And that using the phrase ‘academic criteria’ builds more context into the use of the term LSA in the argument.

In short, as an article is built up using everything we’ve talked about so far – acronyms, alternative words – synonyms, polynyms, related phrases, figurative speech, and some new concepts such as colloquialisms, we build up a strong case for whatever it is we are trying to convey.

Example LSI Ranking Page

The page you’re reading now is ranking #1 on Google for its title text: “The Definitive Guide to Latent Semantic Indexing”. Yet it has zero links and far below average domain authority according to MOZ.

This is because it precisely answers the promise posed in the title, and uses Google to verify that promise in its content

Google is now ranking articles on merit alone where that merit is justified. This is good news for all of us. As I’m always saying, writing good stuff now gets rewarded.

Semantic Tools and Generators

Now you know all about LSI basics, why not generate a few LSI terms and phrases yourself.

A great and simple tool to do this is LSIGraph over here. (press the Ctrl key at the same time you click if you want this to open in a new tab – it’s Cmd + click on a Mac).

But knowing a bunch of LSI-related words is one thing – knowing whether Google is even remotely interested is quite another. To find that out.

LSI and Organic Traffic

The whole point of ensuring your article can be found and indexed semantically by Google is to get you more visitors. And if your site is commercial, then more visitors usually means more sales.

Whilst I always recommend working out how to use paid advertising to ensure a continuous stream of traffic to your site, why not ensure your content fits with Google’s aim of delivering high-quality solutions to its searchers as well, so you can benefit from tons and tons of free traffic.

That is why LSI ultimately matters. It’s getting you noticed by Google for the right reasons.

Mobile Semantics

Mark Zuckerberg, founder of Facebook is the first person to publicly announce his company was going to adopt a mobile-first strategy.

And he was dead right to do so (notice I’ve used the phrase ‘dead right’ here – this is pointless prose from an LSI point of view, but it adds ‘voice’ to my writing – and over time that voice may well be picked up by Google’s AI machines to mean something relevant to intent – or something else we don’t yet understand).

Mobile is the fastest-growing medium by far. And it’s mobile that first drove the concept of search using Voice. Right now, 1 in every 4 searches on mobile is done using voice. And it’s rising fast.

That means, search engines need to understand speech from a semantic point of view as well as the written word – and they are very different things due to the way we are taught to write in school.

This is worth bearing in mind as you develop your writer’s voice. Speaking out aloud whatever you’ve written is the simplest way to understand whether you’re writing from your voice or from some academic voice. This is going to matter a lot more in the future (TOP TIP: your voice will become infinitely more important than an academic voice over time).


Hashtags give people quick references to things. But their biggest use is in search. The clearer the tag, the better the search. But the longer the tag, the harder it is to remember.

Right now, it serves very little use in terms of ranking, but a lot of use in terms of specific result delivery.

Red LSI Flags

The key change in search over the last decade has been the overuse of keywords. And that applies to the overuse of LSI just as much, with the only difference being that the more synonyms you use, the less clear it is whether a keyword is being stuffed or not.

At least that’s the opinion of many SEO gurus, but I disagree with this. Google has more Ph.D. engineers working on AI than any other company, so it makes complete sense to guess that they understand this better than us.

There is a 99.99% chance that if people start using LSI-related phrases to keyword stuff, it will be obvious what’s happening, and a Google penalty will ensue.

Start by checking your keyword density. Then compare the combined densities of related LSI-oriented phrases. That’s a really tough analytical job because no one knows exactly how Google does this in the first place.

How To Write LSI Oriented Articles

This takes us nicely on to the final subject of how to write articles using LSI in order to give you a lift up the rankings.

And the answer is quite simple, don’t! Instead, do the analytics so you understand the space in which you are writing. Ensure you cover all the bases as dictated by Google, then write the best article you can use your research plus your own knowledge and voice.

Google will then reward you for the good stuff. That’s the bottom line!

Where to go Next

Read the following article on page 1 ranking to find out more about what you need to do next if you want a bigger audience, more followers, and more sales.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.