If you have been dutifully consuming articles on search engine optimisation on the Internet, you have probably run across articles about latent semantic indexing (LSI) that look like that higher level calculus class that still gives you nightmares. If you're not sure what we mean, go to Wikipedia and check out the phrase 'latent semantic indexing' to get the flavor of the discussion.

What is Latent Semantic Indexing?

Latent Semantic Indexing, or LSI, is a way for search engines to view and rank web pages not just on the basis of keyword density, but also on the basis of other words that might be expected to be present in content related to the keyword. For example, a web page about roses might also be likely to include words such as hybrid, pruning, garden, fertiliser, Japanese beetles. This analytic approach allows the search engine to determine whether the content of a web page is thematically relevant to the search being made (eliminates web sites that sell 'Just Like Roses' aftershave) and establish an index of usefulness, e.g., de-emphasise sites that sell 'roses, roses, roses all roses, all the time') and elevate sites that provide information about 'rose gardens and the cultivation of hybrids, including pruning and other special care issues'.

History of LSI

According to Peter Nesbit, Latent semantic indexing was created by Google and other search engines to counteract the way that e-entrepreneurs were manipulating search engines for the purpose of generating ad revenue by using duplicate information or high keyword density pages of little value to the searcher. Google purchased a company, Applied Semantics, to develop LSI early in 2003. Since that time, sites that rely on keyword density, reciprocal linking or single keyword anchor text have dramatically dropped in terms of Google page rank. (See our SEO 101 articles for more information on page rank.) Sites that provide authentically useful content, as determined by the constellation of thematically related words that are present on a page, have gained in Google page rank. Although Yahoo and MSN are still more focused on individual keywords, all search engines are moving in the direction of reliance on LSI algorithms.

Strategies for dealing with LSI

First of all, even though there are many, many purveyors of LSI-optimising software in the marketplace, you should recognise the limitations of the software before making a decision to purchase. While there is no way of predicting exactly how a specific search engine treats data, there are ways to increase the potential for higher position in search engine results:

  • Provide authentically useful content that draws your potential buyer to your site. This could include articles, a forum, etc.
  • Forget about placement and density; use a natural flow of writing.
  • Use sets of related words in your content.
  • Forget about reciprocal links; seek out high quality inbound links. (Read our article on linking strategy for more information.)

Bottom line, the more useful your site is to searchers, the more appealing your site will be to Latent Semantic Indexing search engines and the better your ranking will be.

Now that you know how the search engines work, let's focus on how to find the best keywords to use. Read the next article on our SEO series, Keyword Research, to learn more.