• Register
• Log in

SEO Theory - SEO Theory and Analysis Blog

Algorithm analysis, Web community relationship analysis, SEO practices and techniques, industry news, etc.

• Home
• About SEO Theory
• SEO Glossary

How not to write an SEO analysis

Posted by Michael Martinez on November 19, 2007 in Competitive Analysis

I like Fantomaster. He’s a smart guy with a no-nonsense attitude. So when he links to someone’s analysis I usually follow the link and see what’s up.

I was disappointed with this study of the possible correlatione between PageRank and search results because the paper doesn’t even begin to describe the data that was used, how it was collected, or what the conditions of the study were.

In other words, we’re presented with a handful of graphs, some mathematical gibberish, and links back to the Web site.

There is no credibility in presenting a study like this. You have to show people what your methodology is. You have to show people how you collected the data (or where you got it). Your work has to be replicable (even if your results are not replicable).

These faux SEO studies do nothing but sow SEO myths and misinformation across the Web. Maybe Ralph saw the data and understood what the study was based on. Maybe he just liked the fact that the data feels about right. It does feel about right for certain types of queries, but the SEO community depends on links so much that unnatural search results will almost ALWAYS have a close correlation between PageRank and search results.

So depending on PageRank for your competitive analysis is like shooting yourself in the brain and asking why you have a headache. You’re manipulating PageRank, dudes. All you can validate by studying PageRank in the SERPs is the statement that SEOs push content through links.

That doesn’t mean search engines rank content by links. It means that SEOs promote content through links.

Those are two entirely different, distinct concepts.

In a completely natural search result the impact of linkage should be barely noticable. If you’re only studying highly competitive queries you’re wasting your time and resources documenting a well-known fact: SEOs promote content through links. SEOs compete through links. Blackhats compete through links.

Duh! Hello!

You don’t need to run a six-month study to show people the obvious.

If these graphs are based on non-competitive queries, MAKE THAT CLEAR. Tell us what the keywords were. Tell us how the link research was done (it’s completely invalid if you grabbed your link data from Yahoo! or Google). Tell us how the PageRank updates were factored into the study (where is the dividing line between the “normal” Toolbar PR updates and the October Link Sellers Slaughter?).

We can do without any more SEO studies of Toolbar PageRank. Not because there is nothing to learn from looking at Toolbar PageRank in the SERPs, but because these SEO studies all prove to be worthless. Last week I asked people for case studies because I feel that actual case studies provide unique and valuable insight into actual experiences.

Statistical analysis of nonsense data isn’t going to benefit anyone. We do statistical analysis here at 1st Query. I assure you, we’re not wasting our time looking at whether there is a correlation between PageRank and search results. In a competitive query, yeah, you’re going to find that kind of correlation. You don’t need to do a study to figure that out.

UPDATE: I should point out that Ralph did not wholly endorse the study, either. He asked some questions, too. I’m just a little annoyed that people prefer to engage in “SEO lite analysis”. That don’t cut the mustard with me.