Pointlessly Ranking High In SERPS

Lijit is a useful tool, and so is the new-ish Google property, FeedBurner. Between them, they describe the searches that lead people to this blog. Looking at the searches that lead people here, provides some insight into how search engines determine rank, and the techniques that SEOs use.

As of the day this article was started (2007/09/10), we rank number 2 for “XYXXY”. Not a very competitive keyword. I used it in one article here on 25th May 2007, about four months ago, to illustrate a trademarked company name - though of course, I’ve now used it again… How about the search for “Unable to find the published output for this request (For Run Cheque )”? Number 2 on Google in September. By 6th December, xyxxy has slipped far down, but we’re up at number 1 for the query about cheques for an article published in November 2006 - nearly a year ago.

Yahoo shows us as number 2 for Xyxxy in September, but intriguingly manages to spot that I had derived this name from the amazingly ancient Colossal Cave game, in which there are two codes - “plugh” and “xyzzy”. I’d guessed that some business or other would be using one or both of these, and so I twisted the word, slightly. Yahoo inferred the usage. Cool. By December 2007, we’re down to number 14 for this query. The business site for the company of that name does beat us, at least.

MSN in September also shows us as number 2 for Xyxxy, and brings up the Japanese site, xyxxy.com, as well as the Colossal Cave adventure link. By December, we’re number 3.

What about the reports question? We’re near the top of the second page of results on MSN and invisible (well after the tenth page) on Yahoo.

The more astute of you will have realised by now, that it is unlikely that we wanted to achieve Search Engine dominance for these searches. So why do we appear? Why does the position for these non-competitive terms change with time? Does this allow us to infer anything about search engines?

SEO and In Page Factors

There’s two main techniques to SEO. Getting inbound links, and sending page rank to the right pages in your site.

There’s a basic rule to SEO which is that for non-competitive keywords (where other people haven’t been using SEO techniques, which, these days, means that there isn’t a lot of money to be made off them, except accidentally or for very niche businesses), you can rank well by using in-page and in-site optimisation. When you have competition, inbound links will provide the rank.

Neither of these searches has clearly related text in a link within the site. There’s nothing here that says “xyxxy” and links to a page called “why we use xyxxy” or anything like that. It’s not in any page metatag, description or other component. Just some mentions in the text of a single article.

So we’re ranking for non-competitive terms that the site is not optimised to rank well for. That means that we’re gaining rank for these search queries, because of inbound links. But no one points to us for these terms (at least, at the time of writing). What’s going on here?

Page Rank and Relevance

Most users will link to a site by referring to the main domain name. So when referring to Google, they’ll use www.google.com or just google.com as the domain part of the URL. This means that when a site has material of interest below the home page, that search engines need to spread the weight of the home page to other pages in the site, so that users can go to the specific content, even though the references are only to the home page.

There’s a lot of ways that Search Engines could use to spread the weight from the home page to interior pages. Ask’s experimental “Teoma” search engine appeared to use an ingenious system - if the site was part of a collection about, for example, financial services, then weight was spread to pages about financial services and not to pages about, hmmm, rubber duck collectors. This meant that sites needed to be part of a wider community to get real weight - very different from Google’s citation based model.

The citation model used by Google assumes that if your site has been found relevant for any content by a number of other sites, then your site is relevant for whatever you put on it. This is, essentially, why Google suffers from paid links. If sites were contextually relevant then you couldn’t send (much) rank from a site about financial services, to a site about pest control.

As it is, if you can build PageRank for a unique niche, you can use that to improve Google position in the things you care about, but that may be harder to get links for. Googlers encourage putting content irrelevant to the business, on a business website - though I believe that this is actually to the detriment of the internet and searchers. I’ll explain why, in another article… focus… Actually though, that is the point. Focus.

Learnings

What you should be taking away from this is that SEO depends on the nature of the search engine ranking system, and that Google uses a citation model. Because of this, anything you put on your site to attract links, will build Google PageRank. Having got PageRank, you can push it around your site using in-page/in-site factors.

The signature of having built some kind of value from inbound links, is that a number of irrelevant keywords will be ranking highly.

Other Stuff You Should Know

This is *NOT* a definitive guide to SEO. I’m just describing an aspect of how SEO works for Google. It is an important one, though, because if you do experiments and take lessons from low competition keywords and then try to apply them to high competition keywords, the game is different and you’ll fail to get as much traction as your effort implies… and you can know this because you now know that you can rank for nonsense that you don’t care about.

In other words: Lessons learned from low competition keywords do not imply that identical techniques work for high competition keywords.

Why is that relevant? Because I see numbers of snake oil salesmen who demonstrate their special, undocumented-by-Google techniques, and their mastery, by ranking highly on non-competitive keywords. Those techniques will fail when used in places that count - competitive keywords.

What does the decay rate tell us?

The trend for this small sample is representative of the behaviour of many search engine results that I see. Over time, results show up lower and lower on the page. The more competitive the search, the faster the decay.

Couple this with another observation - a freshly published article from an established site with tolerable pagerank, will get high prominence within hours of being published. The position of that article (the rank) will tend to decline with time, unless that page attracts inbound links.

While there’s ebb and flow in the specific techniques that gain rank, the main ones still work:

• Lots of varied inbound links from a variety of aged sources.
• Fresh content.
• In site optimisation to send PR to the right places.

SEO is largely an investment exercise - you spend now in the hopes of gain in a period of months, rather than paid search where you might see the effects of advertising payments having an effect on sales in a matter of hours. These observations about pointlessly high ranking suggest that the fundamental techniques of SEO still work… and that you should be designing sites to use intrinsic SEO techniques, writing and revising material to maintain freshness and currency. There’s nothing here that should surprise anyone - search engines want to deliver the best search results, which will tend to be pages from aged sites with topic relevance and fresh content, which are referenced widely by other sites.

There’s a lot of churn in how results are shown, and a continuous ferment of changed ranking, when looked at on a daily basis - but the long term nature is such that the right techniques deliver lasting value. The ways in which search engines choose rank may lead to unexpected behaviour. In the case of Google, their choice of ranking system has an economic impact leading to links themselves having a value. That will cause changes to the way that Google uses links, adding further complexity to the already evolved Page Rank idea - but mostly will fail to affect users who do not use paid links.