Saturday, June 25, 2005


SEO of course stands for Search Engine Optimisation, while SOM stands for Snake Oil Merchants, the many peddlers of “satisfaction guaranteed”, “instant” lotions and potions that are supposed to propel any website to the top of its SERPs. More about FED below.

SOM are of course in their element when it comes to SEO, simply because it’s actually very, very hard to prove (or disprove for that matter) that any given method “works” (or not). So, SOMs can claim anything they want, mostly unchallenged.

All Google watchers, including myself, are of course a little (some a lot) guilty of making claims we simply can’t really make hard. We jump to conclusions easily without reflecting much on our claims.

Suppose you decided to try something like increasing the keyword density (KD) of a page currently ranking say between 11 and 20 in the SERPs, in a bid to get to the first page.

So you modify the KD and hey, waddayaknow… the page turns up at, say, #5! Success!

But what have you proved? Very little indeed. You didn’t for instance have a “control” or “blank”, a similar page, ranking similarly but which you didn’t modify. You probably didn’t check the rankings of the surrounding, competing pages: what happened to their SERPs? Is your gain simply due to their loss? Have you gained more inbound links during the "experiment"? Was there an algorithm shake up going on?

There could be any number of reasons why your page improved in rankings, many of which would be totally unrelated to the KD modification.

But it’s very tempting to draw that simple conclusion:

Increasing the KD of my page improved rankings, ergo increasing KD works”.

To “test” the validity of your conclusion I suggest wagering a little bet with yourself. If you applied the method to another page and it worked you win the bet and a cool $1,000,000! But if it doesn’t work you lose the bet and you lose your life [it’s a cruel world!]

Would you accept the bet, based on the flimsy “evidence” you have at hand? I bet you a cool $1,000 that you wouldn’t: intuitively you know you’re on shaky ground.

SERPs of any given page depend on perhaps over a 100 factors, many on the page, many off the page, many outside of your power. Making simple changes to any single factor may well “work” but it doesn’t prove much at all.

The “many factors problem” is very common in many walks of life. I was confronted with it almost continuously in my capacity as a material development engineer, in a previous life. As the properties of any given material usually depend on a large number of factors, such as composition and manufacturing methods, material development engineers have to be very careful in drawing conclusions about any modifications they make to the contributing factors and the effect on the material's properties.

“Many factor problems” are usually tackled by means of so-called “Factorial Experimental Design” (FED). In these experiments, a number of factors is manipulated in a precise and controlled manner, holding all others at constant level. Simple statistical analysis then allows to draw justified and solid conclusions about the influence of the various factors on the response variables (such as the material’s properties).

An example of tackling many factor problems
is applied to social research on this site.

Can FED be applied to SEO? In theory, yes, in practice it would be very difficult. FED is based on the assumption that the researcher has complete control over all factors that may influence the outcome of the experiments and that’s simply not the case in SEO.

And that’s why most SEO claims are “guesstimates” at best, rather than precise, founded conclusions… That, of course, includes most of my own guesstimates.



At 10:11 PM, Anonymous Anonymous said...

This is an awesome article. I learned a lot of great tips and pointers. Thanks for the information!!

At 5:31 PM, Anonymous Anonymous said...

Although this article was published a few years ago, I think everything you said still holds true. Every SEO Company can only promise so much in terms of results, because often times what clients want is simply impossible to achieve. Great article.


Post a Comment

<< Home