SEO Keyword Experimental Results
In a recent post I gave details on an SEO experiment to see how the rankings change with time. Well, the results are in and are presented below:
Graph 1. Cheap Flights
Graph 2. Christmas Gifts
Graph 3. Dog Kennels
Graph 4. Black Circles
Graph 5. Politics
Graph 6. PHP
The graphs show the rankings of websites competing for the top-ten positions for a particular key-phrases over a fourteen day time period. Looking at the number-one position, we can see that in general they are flat over the entire time period.
One thing that is clear from the data is that the URL appears to play an important part in the selection of the top sites. In five out of six of the keyphrases, the top positions were held by sites which include the key phrase in the domain name. Interestingly, in the one case where the key-phrase was not included in the domain holding the top position, there was a variation in position, not that this is statistically significant, yet.
In order to extract useful information, we chose search terms that we thought would show degrees of competition and topicality. To some extent this is shown in the data. The most volatile changes occur in the data for cheap flights. However, topical, competitive terms such as Christmas gifts show do not show the degree of competition that we expected. Perhaps this is because Christmas gifts is still a rather generic term despite being topical.
Politics shows an almost static results with time. We chose this keyword because we thought that it would not be a term that would be sought after term for online businesses. Similarly for black circles, but it turns out that this is strongly connected with tyres and does show a fair degree of competition.
PHP was also chosen as a non-competitive word on the basis that it freely available open source software. Looking at this data we can see that the rankings for several sites, move in unison. Could this be a sign of non-competitive term? Rankings that move together, could still be uncompetitive if occasionally, sites pop-into the index and displace a number of search results.
It has not escaped us that looking at the graphs, it may be possible to define a better metric of competition from collected data. Ranking data is easy to collect and in some sense is directly linked to how competitive a key phrase is. Others methods of measuring competition rely on gut feeling or the weighted average of measurements returned by various indirect search queries and on-site factors.
We have more data for the other six key phrases to be analysed and this will be presented it in a future post.
Sir Pumpkin Longshanks
Programmer, Research and Development
