tkik - 12:52 pm on May 17, 2010 (gmt 0)
I just discovered that google seems to tune in searchresults for what they think are synonyms for the words beeing searched for. When we searched for a term "airwidgets alu", for which we have a document with exactly that term in the title and in the text, we become a very strange mix of answers in the serps, like "widgets alu" "aluwidgets" "flexwidgets metal" etc. What we see is that google highlights all those different terms (which are indeed also different things!) in their serps.
This leads me to believe that google, whenever it sees that the number of hits from their database is too small, they
1.) start building a set of search terms with synonyms (airwidgets= widgets, aluwidgets, flexwidgets,... alu=metal, aluminium etc.)
2.) get the results of these queries too
3.) mix all results into one.
Now by the last step, they make a horrible error IMO. They do not check the closelyness of the result to the searchterm being searched for by the user. instead, it seams to me that they just use their original ranking criteria like backlinks etc. to merge the final result list.
For those who had documents that are optimized for long tail, they will suddenly find their document to compete with documents for (slightly) different terms too. In the worst case, these slightly different terms are being high volume words where people have been optimizing for a lot (backlinks etc).
This would then lead to your ranking getting worse.
We have a multilingualsite that has been hit massively in the german part, but not at all in the dutch part. When comparing the results from google.nl and google.de, we found that in the german google, there is often/always a list of synonyms on the top of the serps. In google.nl, there is none. The behaviour described above can not be found in google.nl, meaning, when you search for airwidgets, only documents that contain airwidgets will be listed in the serps.