Effective Internet Marketing, using LSI and Theming
The traditional approach to writing articles for internet marketing has been to identify one prevailing search term, then write an article of several hundred words, the main purpose of which is to utilize that chosen phrase to the point of absurdity. Never mind that the resulting article is virtually useless, the whole point was to get the attention of search engines.
Google to the rescue!
Fortunately, the Google Search Quality Group has taken steps to eliminate this ridiculous abuse of the net. Their revised search algorithms are designed to evaluate the content, rather than simply match keywords. (See our related article, “Content Farms and Google”). By developing software which utilizes a process called Latent Semantic Indexing (LSI), they have enabled computers to emulate human thought processes. The resulting algorithms are used to downgrade articles which utilize excessive repetition of keywords and to upgrade those which contain useful information. In simple terms, Google now recognizes multiple synonyms* for the primary search term. For example, a search for “Laptop” may also trigger recognition of “Notebook”, “Computer”, and even several of the dominant brand names of such devices. In other words, Google is no longer searching just for keywords, but is instead seeking themes.
How do I write with themes?
In order to write for the latest algorithm, an author really needs access to current trends. Several software options exist, each of which evaluates current search results, then provides information on what words are being linked with a specified keyword. It doesn’t matter that an article offers accurate and useful information, if it doesn’t include words which potential customers – and search engines – actually use in searching.
To identify current search trends, it is necessary to have access to information on recent searches. No static list can accomplish this. Even a perfect and complete list of keywords used in searching for a laptop computer in January of 2001, for example, would have very limited value in January 2011. Marketing in 2011 must use words actually in common use in 2011. In the same manner, 2011 keywords will likely be of only limited value in 2015.
Even a well built website with excellent, instructive articles will benefit neither the owner nor potential customers – unless they find one another. “Finding” is usually a function of searching, and searching always uses keywords. The latest Google algorithms do a marvelous job of matching search terms to content, but writers still need to use words which make the connection. That connection now goes beyond the user’s choices, and includes theme words which Google has determined are related. How can machines which think, literally, only in 1’s and 0’s possibly go beyond the traditionally expected counting of keywords? How could such machines actually recognize useful content? To avoid redundancy, we will assume that you can understand the system when you read the next few paragraphs. In a nutshell, though, the system takes an average of several top rated articles, and counts recurring words. Any words which occur frequently are assumed to be a part of a relevant theme. This includes words which might never occur to you and me as search terms. The other side of that same function is that articles which contain redundant use of three or four word phrases are automatically downgraded. Include your title phrase every hundred words or so, and you will see your rating drop.
Identify the best words
OK, so how do writers identify the best words? If you are serious about your articles being “above average”, it is absolutely necessary to identify current search trends. This means that you must have an active source of live, current information, such as that offered by “Wordtracker”. There are several good (and some not-so-good) clones, as well. Understandably, this sort of service costs money to develop and maintain. You can find free trial versions, but for long term access to current trends, you’ll probably find it necessary to subscribe to a quality tracking service.
In greatly simplified terms, these programs take an average of several top performing web pages which use your chosen keyword, then count occurrences of related terms. A report is generated, showing actual counts of specific words in those articles. More sophisticated programs allow you to specify how many pages to use in setting the standard. More pages = more accurate predictions, but also more time to process.
LSI Keywords
For a good introduction to the concept and process take a look at “LSIKeywords.com”. Enter the term “Laptop”, and give it time to process. On the date of this research, the top 5 responses were as follows:
Single Word
|
|
Two Word Phrases |
|
Three Word Phrases |
|
| Laptop |
310 |
Hard Drive |
55 |
drive bare drive |
42 |
| Laptops |
262 |
Return Policy |
43 |
Notebook Hard Drive |
38 |
| Drive |
122 |
Drive Bare |
42 |
Hard Drive Bare |
36 |
| Hard |
89 |
Bare Drive |
42 |
Internal Notebook Hard |
32 |
| RPM |
87 |
Notebook Hard |
41 |
Delivery Collect Store |
30 |
Brand names do count in these processes, but you may be surprised at which brand names are most effective. The above search revealed the following names / counts:
| Single Word |
|
Single Word
|
|
Single Word
|
|
| Acer |
42 |
Toshiba |
40 |
Macbook |
38 |
| Samsung |
31 |
Asus |
24 |
Dell |
23 |
Who would have guessed those ratings?
Synonyms*? Well – not exactly!
Early in this article, we used the term synonyms*. That term is not accurate, in the literal sense. Words which may be treated as synonyms under the Google search algorithm would not likely occur to the average person as such! This is where the term “theming” comes in. The new Google algorithm (announced in February, 2011) responds to the theme of the search term, rather than just the literal meaning of the word. In effect, when Google sees “Laptop”, it now thinks “Computer”.
In the above search, it was clear that “Laptop” and “Laptops” were desirable keywords. Far less obvious would be “Drive”, “Hard”, and “RPM”, yet use of those terms seems to be a definite positive factor in elevating the rating of an article. Most people would guess that “Dell” might be the highest rated brand, but from this investigation it appears that in a Google search, “Acer” may be the most valuable!
It’s crucial that you remember, though. Commonly used search terms constitute moving targets. These figures are true as of the date of this research, but “commonly used” is not a static term. The same search process, done a year later, will no doubt produce different results. If you were to try an identical search now, you would not likely see an identical outcome.
Write for the Panda!
If you really want to succeed under the latest Google (Panda) standards, start with a well written, informative article which does not make excessive use of keywords. Then download a current word list on your chosen topic. This process will reveal which words are most valuable now. Evaluate your article for use of the highest rated words. If they aren’t in there, try to fit them in without disrupting the flow and value of the “piece”. In effect, you may use keywords 100 or more times, yet without redundancy. Your readers will benefit from truly useful content, and neither the readers nor Google will be offended by excessive repetition.
To pull all of this together: Don’t rely on your own intuition; don’t assume that logic or your own experience will be enough to fully maximize your use of the net. Don’t try to utilize somebody else’s list from a previous date, either. There is no adequate substitute for knowledge, and even you don’t know it all! Use the best tools to stay current, or watch your site disappear into the middle of the pack.
Tags: lsi and theming, theming, lsi, keyword tracking, tracking tool, wordtracker, secondary keywords, primary keywords, keyword density, keywords, keyword, search engine optimization, SEO