Tuesday, 26 October 2021

What On Earth Are LSI Keywords – And Do They Really Matter?

what-is-latent-semantic-indexing-all-about

LSI (or latent semantic indexing) keywords are supposed to help search engines contextualise your content and better understand the focus of your web pages. (Source: backlinko.com)

Last month, self-styled SEO guru Brian Deen posted a short video explaining how to use LSI keywords in your content: 

Unfortunately for Brian, LSI keywords are a fairly controversial subject and several well-regarded members of the digital marketing community decided to haul him over hot coals for (allegedly) misusing the phrase:

Putting aside the obvious problems with online toxicity and pervasive bad manners, this spat forced us to reflect on just how inaccessible SEO must feel to traditional marketing managers, CMOs or business owners who are simply too busy to keep up with all the buzzwords used by people “in the know”.

It’s like being the only person who wasn’t told about the dress code at a drinks party – or the kid that turned up to school with a brand-new yo-yo, only to find their friends had all moved on and bought Ataris.

Thing is, most of the underlying concepts are fairly simple. Once you strip away the jargon and really hone in on the theory, it becomes obvious that Google must be using some sort of semantic indexing to understand the context and content of your web pages. Irrespective of whether or not you should call them LSI keywords.

It also seems perfectly clear that you should be using a good spread of semantically-linked search terms in your content. But we’re on a mission to demystify the foggy and somewhat impenetrable world of digital marketing here, and we don’t want to get ahead of ourselves just yet.

To clear up any confusion - and explain why everyone spent the end of September arguing about LSI keywords - we’re going to walk you through the basics of semantic indexing right here; looking at what LSI keywords are, how they work and whether it’s really safe to discount them completely. 

What is Latent Semantic Indexing?

Latent semantic indexing, or latent semantic analysis is amathematical technique that’s used to discover statistically-significant relationships between different words and phrases. 

For those who don’t speak geek, it’s a way of analysing several bits of content to work out whether content about a specific subject ought to mention other - semantically related - search terms so that search engines like Google can contextualise your writing. 

A good (but basic) example would be the relationship between “jaguar” and “car”. Articles featuring the word jaguar could be about wheeled vehicles or jungle-dwelling predators but if your article contains the words “jaguar” and “car”, a search engine using latent semantic indexing would be able to tell that you were talking about the brand and not the spotted jungle cat. 

Why Are LSI Keywords So Contentious?

There’s a lot of confusion about LSI keywords. Some people think that they’re complete nonsense. Other people think they’re the key to supercharging your SEO campaigns and achieving truly groundbreaking results.

Google the phrase, and it becomes evident that even the experts are very confused about the true importance of latent semantic indexing. 

Ironically, this is largely an issue with semantics. Way back in 2016, Google published a paper that talked about teaching algorithms new ways of understanding the context or meaning of a piece of content. (Read it here if you’ve got a few hours spare!)

In the paper, they highlight the shortcomings of traditional language processing models and advocate for new ways of understanding the context of search queries. 

Some people claim that this is incontrovertible proof that Google is actively looking for LSI keywords in your content. 

Other people point out that the article in question was published six years ago, that latent semantic indexing is one of many machine learning models that can be used to understand the wider meaning of an article and that the paper in question actually highlights the shortcomings of traditional LSI techniques. 

And it’s worth acknowledging that Google’s own John Mueller says the same thing:

But everyone agrees that Google is working hard to understand the context of search queries and web content. 

In fact, Google has two dedicated NLP (natural language processing) algorithms called BERT and SMITH that are specifically designed to analyse and understand the meaning of words and sentences. (Source: newbreedrevenue.com

latent-semantic-indexing-machine-learning

What Now?

These “deep learning” algorithms trawl through thousands of web pages every day; training themselves to recognise the relationships between different phrases; the tone of your content and - on a very basic level - the way various words alter or change the meaning of a given sentence or paragraph. (Source:searchenginejournal.com)

Why? Well, it’s important to remember that Google is in the business of providing answers to your questions. If the search engine giant can’t offer up results that match the intent of your query, you might be tempted to start using an alternative search engine like Duck Duck Go or - perish the thought - Bing. 

So the boffins over at Google HQ spend a lot of time and effort trying to figure out how to understand minor nuances in the language we choose to use. 

We’re talking about things like the difference between “easy weight loss tips” and “the risks of rapid weight loss”. (similar articles with a completely different focus and a very different target audience). 

And one of the key ways that Google does this is by looking at the peripheral phrases that are used in relation to that main “weight loss” keyword. 

Obvious things like “risk” or “benefits” but also language like “metabolism”, “calories”, “antioxidants”, “glycemic index” or “metabolic acidosis”; words that are often used by experts in the field, when writing advice or thought leadership about a specific topic under the larger “weight loss” umbrella. (Source: SEMRush).

How Does This Help Me Write Better Content?

As a web marketer with a keen interest in outsmarting your competition, you’ll want to take advantage of this increasingly-sophisticated NLP by weaving a good spread of related keywords into your content. 

Keywords that Google increasingly expects to find in natural, authoritative and useful articles designed for people and not computer algorithms. Keywords that set your content apart from the shallow, rushed or repetitive articles that no doubt clog the SERPs you’re targeting. 

You can do this by reading industry-leading articles about your chosen topic, taking note of the common words and phrases used in relation to your target keywords and then brainstorming ways of using them to improve your own content.

For example, a marketer writing about engine tuning might head over to AMG’s Ultimate Guide To Engine Tuning, and make a note of the fact that the content makes many mentions of:

  • Petrol engine
  • Miles per gallon
  • Performance
  • Turbo
  • Throttle Response
  • Torque 
  • Rev range

A savvy marketer would then try to weave all of these keywords into their own engine tuning content, so as to ensure that Google understands that they’re talking about the same subject, in a similar level of detail as this top-performing guide.

But you don’t have to do all the legwork yourself. You can also use free Google tools to track down these NLP or (dare we say it) LSI keywords in next to no time at all. 

To do this, head over to Google’s Natural Language tool, paste in content from an article that’s ranking well for the keyword you want to target, and watch as it spits out a bunch of semantically-linked keywords that’ll help you rank your content.

Now, it’s worth pointing out that this isn’t a magic wand. It’ll help you rank, but there’s a lot more to getting content to perform well in SERPs. Your writing needs to be unique and engaging, informative, accurate and shareable. It needs to be optimised for people and capable of improving your brand. 

But if you can weave in some semantically linked keywords?

Well, it’ll certainly help Google and other search engines make the connections needed to understand the specific topic you're targeting, and that means it’ll be just a little bit easier to reach your target audience. Sounds like a win to us!

connections-semantic-indexing

Struggling to produce your own content?

Knowing the ins and outs of semantic indexing is all good and well, but actually finding the time to write detailed and interesting content can be incredibly challenging. Particularly if you’re trying to look after a website, grow a brand and keep the rest of the C-suite happy.

If you’d like help from seasoned industry professionals, we offer dedicated content marketing services that’ll help you get found in search – and stay ahead of the competition. 

Content Planning & Creation  Engage and convert new customers with killer content collaborations. Learn More

New!  A plain-talking digital marketing podcast  Available in all the usual places  Grab it here