We have been seeing in the last 2 years something called latent semantic index (LSI), used by search engines. So, what is it exactly? Here is a quick definition:LSI gives the possibility to search engines (especially Google), to associate keywords between them, like a human brain would do. For example we might associate milk with the color white, a ferrari with the color red. That’s exactly what search egines are now doing and learning: the keyword Ferrari will be associated with sports car, speed, red, and expensive keywords, and this is key to take this in consideration when you create content, and also when you build your website.
Let’s make a step backward, and remember how search engines used to give credit to a page some years ago.
Your page had to target one expression, and there was an old trick called Keyword stuffing, which relied on BOMBARDING your page with your targeted keyword or expression. If I want my page to rank well for the expression “tennis shoes”, this keyword would have taken most of the content and therefore this would have make spiders believe that the page was HIGHLY RELEVANT for this specific expression. This practice has worked some time however pages looked rubish as it was repeating the same keywords.
Focusing on quality, Google came up with the Keyword density filter to fight against this. If the density of your main KW was above a certain rate, it was considered as spammy for Google, and therefore your page was not showing on Google anymore. This was a big step in filtering fake content, but this was not enough as you could just put any text in your page and put your targeted keywords randomly in the page… Google needed something else to be sure they are showing only qualified content for their users. And that’s how LSI came up. The whole idea is based on making association between keywords, and index them using those associations.
So now, how can you take advantage of it? Let’s take again the Ferrari example. If you want to create an article about the last Ferrari model, you should use every related keywords in your page: Here is a small list: Speed, F440, F480, red, Porsche, Lamborghini, and the list is long… Now, if you think of this, a genuine article about the last Ferrari model SHOULD CONTAIN THOSE KEYWORDS and this clearly helps Google to decide the quality of an article.
Article writing is not the only area in SEO where LSI is applicable. Indeed, you can define the whole structure of your website using that mindset. The different categories and sub categories should be created with LSI in mind, and how those categories are associate with each other. It’s the same with Internal/External linking. LSI changed the SEO rules making the industry more mature (that’s my opinion).
One of the rule which changed recently is actually the way to optimize the HTML Title and H1 tag on a page. 2 years ago, it was adviced that both should match exactly the keywords you’re targeting (being almost the same). It is different today, and the SEO industry recommand to have use synonyms and variations between those 2 tags, rather than using the same keyword or expression.
I embedded below a slideshow explaining a bit more on details what LSI is, and how we can use it for a website optimization.
{ 6 comments… read them below or add one }
Philippe, you had not written anything for such a long time, I had stopped looking at your blog. (Sorry) But this morning I found this article and it is so interesting and opens so many interesting thought opportunities I am so glad I looked. For my own purposes, though, it poses a question: as you well know, culture has a huge influence on how people think, so concepts and word associations will differ from language to language and from culture to culture. Do you think that Google’s LSI work takes at all into consideration “real people” trains of thoughts and way of thinking based on language and/or culture, or does it use only as basis American way of thinking and associating words? Let me know what you think. Thanks. Adriana.
Hi,
Very good write up! And thanks for embeding my slide with your article!
@adriana
Thanks for checking out my blog. I’ve been very quiet recently indeed at my blog. I should be more involved, but it takes some time. It’s good to hear from you anyway. You have a good question.
I’mot entirely sure that Google takes the language in consideration when they index pages, and when they make association between keywords. Those associations are made through (it is just assumption) an algorythm decided by Google. Now, should they make the difference per language? I’m not entirely sure. I think the linguistic magic here is only the outcome of semantic: i don’t see it as a way to create it.
Does that make sense?
@Mercy Livi
Thanks Mercy, I think your presentation is really clear and very easy to read. I don’t think seo people are aware enough about this. great presentation!
Oh yes Philippe, it makes sense, and actually it is exactly as I thought too. Thank you.
@Philippe –