When a word or phrase is typed into a search engine, the search engine has to somehow decide which sites are relevant to what was typed in and has to decide to somehow rank them in order. All search engines have different algorithms which place different emphasis on different things in the hope that they will each be better and deliver what the searcher is looking for. it is not uncomon to rank well in one search engine and not another. While search engines keep their ranking algorithms secret, there are a number of basics that they have in common.
One way to consider it, is that the search engines want the most relevant sites ranked higher, they want the most important sites ranked higher, and they want sites that are ‘gaming the system’ to be ranked lower. Based on this, the ranking algorthms could be considered, hypothetically and conceptually, to be a very complex version of the simple formula:
ranking = importance x relevance +/- quality signals
If each of these three components are considered seperatly, this knowledge can be used to help search engines appropriately rank sites.
Search engines want to rank the most important sites first. They all do this by considering the link popularity of a page (ie how many links there are to a site). Google call this PR or PageRank – other search engines don’t really have a name for it. The more links there are to a page or site, the search engines consider the page or site to be more important, so they will want to rank it higher. More emphasis is given by Google to links from higher PR sites and all search engines may give more emphasis to links that come from other important (authority or expert) sites or from sites that also rank high for the same keyword(s). There has been speculation, but no evidence that one-way links are better than reciprocal links. This is considered due to reciprocal exchanging of links is a sort of ‘i’ll vote for you if you vote for me‘, so could be artificially increasing the link popularity (ie importance of a site). There are number of complex theories or algorithms that have been speculated as being part of this, including the Hilltop theory and Topic Sensitive Page Rank. To increase the importance of a site, more links are needed, preferably from other important and high ranking sites.
Search engines want to try to rank the most relevant sites higher. To do that the search engine spider or crawler visits the site and sends its content back to the search engines data base for analysis to determine what the page is about. It asks something like, what keyword(s) is this page relevant to?
To be more relevant to a particular keyword(s) you need to help the search engines. The most important place for the keyword(s) that you are targeting for a particular page are:
- The title of the page
- H1 (heading tag)
- Highest density of the words in the text on the page, but still kept as low as possible with naturally flowing text (search engines are getting smarter and may be able to detect unnatural lists of keywords)
- In the anchor text of links (the text you click on) pointing to the page (from within your site and and from other sites)
These are the most important ways to help a search engine know what a page is about. There are other less important ways, such as the keyword meta tag, having some of the keywords in bold and italics, in the alt attribute of images, etc, but they often do not appear to carry as much weight as the above more important ways. Different search engines in their ranking algorithm give different emphasis on different aspects of these factors. For example, Google does appear to put much greater emphasis on the anchor text of incoming links (ie off-page factors), where as Yahoo appear to more emphasis on the on-page factors.
Other ways that search engines could also determine relevance is that if there are links to the page from pages or sites on a similar topic.
Search engines also want to look for quality signals from the site as part of the ranking formula, in the hope that better quality sites are ranked higher and poorer quality sites are ranked lower.
What might be some of the things that search engines could look at for a poorer quality signal (this does not mean they are actually doing this, but they could):
- are there any hints that the site is participating in spamy link exchanges?
- are they linking to ‘bad neighbour hoods’?
- are there other copies of the content elsewhere?
- are the linking to viagara or casino sites, and the site is not about viagara or casino?
- are they linking for the benefit of the user or for the benefit of search engine manipulation?
The final ranking:
Search engines want to take all these factors and a whole lot more and put them in the melting pot to rank a page in response to a search query according to a hypothetical formula of:
ranking = importance x relevance +/- quality signals
Knowledge of some of the basic factors that go into the formula can be used to understand a pages ranking and applied to improve it. All search engines weight different factors differently, which is obviously why the search results from each one is different. Just how well they are achieving the goal of ranking the more important and more relevant sites higher is certainly open to debate as everyones perception of relevance and importance will be different and there are always people gaming the system with spam.
NB – notice that search engine submission has nothing to do with how a search engine ranks a site.
Final word on content:
The quality of the content of a site is extremely important for the visitor to a site, but computer algorithms are dumb and can’t judge that, so “content” was not considered above in the hypothetical ranking formula. However, with good content, it will be easier to get other sites to link to you, so content has a significant indirect affect on the hypothetical ranking formula.
For more detail on optimizing a site for search engine ranking, we recommend Aaron Wall’s SEO Book.