Rustybrick comments on a discussion thread over at HighRanking recommending that SEOs should avoid trying to optimize a single page for too many key phrases.

While there’s some useful stuff here, it seems to me that the discussion is perpetuating an underlying confusion experienced by many people trying to improve their search engine marketing campaigns by working on on-page factors such as optimizing for key phrases: namely, the distinction between avoiding ‘optimizing’ for many key phrases, on the one hand, and trying to optimize for just one key phrase, on the other hand. The two are not by any means the same thing. Does optimizing for two key phrases count as optimizing for many? How about three? Seven? One hundred and nineteen?

But apart from the silliness of arguing over an arbitrary boundary between ‘one’ and ‘many’, the discussion seems to me to obscure completely a much more fundamental — and, in my view, much more important — distinction, the distinction between the relevance of content, and the optimization of content. In a nutshell, the distinction is this: real content relevant to a given key phrase may (or may not be) well optimized for that key phrase, and well optimized content may or may not be relevant to that key phrase.

Why?

Remember that the real aim of a search engine is NOT to find pages that are well ‘optimized’ for a search term; the real aim of a search engine is to find real content that is relevant, in the minds of actual thinking human beings, to the search phrase. (See the comments on ‘thinking like a search engine’ in the marketing section of a site we’ve mentioned before, aimed at the mental health area.) But semantic relevance does not necessarily equate to syntactic optimization. To put it differently (and fairly obviously), syntactic optimization (factors such as key phrase frequency, etc.) does not uniquely determine semantic relevance. We can easily imagine two different pieces of text, displaying exactly the same level of ‘optimization’ in terms of key phrase frequency, etc., but which are completely different in terms of actual relevance — in the mind of a user — to a given topic. (Any number of algorithmic or even random transformations will do the trick, re-ordering words or sentences or paragraphs in such a way that the text reads very poorly, yet still displays the same level of ‘optimization’.) In the dark ages of the web, it may well have been true that search engine algorithms made no attempt to capture semantic properties of content, and made do with more primitive measures of key phrase frequency, etc., but this is true no more.

(Need proof? Just experiment with asking Google to ‘define’ particular words. At the top of the SERPs, you’ll find a link which leads to a set of different definitions which Google has extracted from various web pages relevant to the given word. Only very rarely does a ‘definition’ page contain the word ‘define’ or the word ‘definition’, yet somehow Google has managed to figure out that these pages are offering definitions. How did that happen? I search on “define xxxxx”, where ‘xxxxx’ is some word or phrase, and Google spits out a list of pages which don’t even contain the word ‘define’! Magic! This ‘magic’ happened because Google’s algorithms are designed to detect semantic relevance; they are quite a bit more sophisticated than the SEOs who think semantic relevance and syntactic optimization are the same thing.)

Now, some people might object that semantic properties supervene entirely on syntactic properties — in other words, that at some level of fine detail, the syntactic properties of text do determine all its semantic properties. This might well be true, but it’s irrelevant. Why? It’s irrelevant because at such a level of sophistication — i.e., at such a level of detailed syntactic analysis that we really can capture semantic properties, a level of detail that Google and other search engines are no doubt striving to reach — we may as well talk about semantic properties anyway. After all, one is capturing the other! In other words, even if it is the case that syntax does ultimately capture meaning, what is really important is that content is relevant to a given key phrase for a human reader, and not (merely) that it displays certain syntactic properties. And in that case, we are back to the primacy of the actual relevance of real content, and not questions like how many key phrases we might want to optimize for.

So, if you want to ask the question about how many key phrases you should optimize a page for, go ahead. But most actual content, which would be judged by human beings to be relevant to a given topic, will automatically be relevant to more than one key phrase, because there will be more than one way of describing that topic. (And as demonstrated by the ‘definition’ example above, sometimes making content semantically relevant may actually mean that you don’t even use one of the words in the key phrase!)

The real ‘answer’ to ‘optimizing’ your page for a search engine is to make its content real and semantically relevant to the topic. As search engine algorithms grow in sophistication, and come to approximate the semantic judgements of real human beings more and more closely, these old discussions about optimizing for key phrases will become more and more obviously irrelevant.