Seo Test - Real Search Engine Tests

Forget everything you have learned so far about SEO and the Google search engine and let’s just put ourselves in their hypothetical shoes for a moment. Forget all your suspicions, opinions and everything you have heard others declare as fact and just for the next ten minutes, let’s presume what Google tells us is the truth… for a website to perform well in the Google search engine then it should strive for the following.

• High quality
• Useful, information-rich
• Unique content
• Relevancy
• Well structured
• Natural Backlinks / PageRank

Those are some of the key-points you will find in Google’s webmaster guidelines, which according to them is what you need to pay attention to, in order to create a website that ranks well in their search engine. I don’t know about you but I have no reason to believe they are lying about the fact they want high quality, relevant websites to show up at the top of their SERP’s, so let’s go with that and think about how we could make that happen if we owned a search engine with a high budget.

The first thing I would do is a general study of common practices and behaviours of all kinds of websites and the people that create them. Are the majority of high quality sites generated by hobbyists or webmasters? How do they create their pages. Do they know HTML or use junk like Frontpage etc.  I would use this information to decide importance of things like header tags, anchor text etc and how often is it utilised in the general high quality website. (btw I think header tags are over-rated).

How can you tell a computer to decide if a website is information rich and useful? Well obviously you can’t but what you can do is gather all sorts of common patterns and create sets of likelihoods to take a reasonable guess. This could be even more effective if we hire some language experts and have them work closely with some expert mathematicians and computer programmers.

In the past, guessing the likelihood of a site being high quality would rely a lot on Pagerank (backlinks) but nowadays I’m not so sure as backlinks are too easy to create unnaturally. We must not presume that no matter how good an algorithm is at decoding information, we can leave it indefinitely. People will always figure a lot of it out and find ways to make better predictions.  For instance, the majority of today’s backlinks are created by SEO’s for business and MFA sites more so than the person creating a high quality website for pure hobby. This means our algorithms will need constant reviews and updates otherwise the whole Pagerank thing becomes pointless.

Unique content isn’t so hard to find (or create) , all we’d need to do is decide how to deal with it. Don’t forget in Google’s own words.. Duplicate content is “mostly not deceptive in origin”. It isn’t penalised as such, it’s more a matter of what version will carry the most weight.

Relevancy. How do you get a machine to decide if content is relevant? Well I guess it can be done to an extent and produce reasonably accurate results, somewhat similar to the way you would check for quality content.

Well structured. This isn’t so hard to figure out, not foolproof, but easy to spot the difference between a site well structured and most of those thrown together in minutes or computer generated. Although the computer generated ones have got better, they can still leave traces of irregularity.

Okay, so here I am, the owner of a search engine trying to program algorithms that will make these decisions for me. So far I have a fairly good idea of how to spot the obvious and to make reasonable assumptions at whether a site is good or bad. All I have to do now is decide the best compromise between achieving those results and not penalising those that are great sites but don’t work too well with the algorithms.

This can be really difficult, almost impossible to decide, so the best thing would be to make a help page public so that we can inform the honest webmasters what their best practice is and what they should avoid doing.

Hmm.. this is all sounding a bit familiar isn’t it?

Ok that’s the hypothetical part over, how does this help us? Well let’s have a think about it.

Again, presuming Google are telling the truth, we can summarise. Relevant, high quality, unique, useful, well structured and backlinks all help us get better ranked results. An algorithm can’t produce perfectly accurate results but can produce good results if the human side of things take notice of what the algorithms don’t like. I.e.. non relevant, duplicate, etc as well as things that the robots don’t work too well with, script, flash, frames etc.

Right, we have two choices here.

1. Believe everything Google tells us and go with it.
2. Believe everything we read on blogs and SEO forums, or should I say, baclink, backlink, backlink, backlink…

Now is there anything I might be missing? Damn I’m trying my hardest to think if there is any way we can put this to the test. Nah, I just can’t think of, … um, er, wait a minute? what if? No don’t be silly, that could never tell us anything surely? Hmm, I wonder. Oh bollocks to it, I tell you what let’s just try it and see what happens if we type a keyword into Google and take a look at the top ranking websites!!

Three minutes later…

Well fuck me! I weren’t expecting that? I took a look at the top ranking websites for a few different keywords and guess what, most of them are well structured, unique, average to high in quality, definitely relevant, not driven by script, frames or flash. The only thing I noticed that seemed to have a widespread variation of importance was pagerank and backlinks. For some reason they didn’t seem to matter too much to what was at the top of the list compared to relevancy. If anything it seemed the most common element between all those top ranking sites was indeed relevancy.

Something is obviously not right here. I have no doubt wasted three minutes of my time because Google must be lying. Fuck it, I’m going with backlinks!!

Posted on October 28th, 2007