In preparation for my Internet Librarian talks, I was playing around with Google, looking for how much my search results varied based on how I constructed my search. I was surprised to see how much they differed.
First, I tried searching carbon neutral winery and then searching winery carbon neutral. Not only were the results significantly different, the first search retrieved 93,500 results and the latter retrieved 195,000. (My hunch is that since there are plenty of sites that have the exact phrase "carbon neutral winery", Google delivered the results from a "shallower" search than for the less common wording of "winery carbon neutral".) I also found that, among the first 20 results of each search, there were five results that appeared in only one or the other search result.
My other surprise was the difference in results between standard results and limiting the search results to web sites updated in the last year, a technique I often use to get more relevant results. There was very little overlap between the two search results and, while I still prefer the results with the last-year filter, I am reminded that I could be missing some good material that way.
I'm finding this more and more with Google. For my Internet Librarian London talk I also did some tests - and got completely different results for these searches:
allintitle: ~animal ~nutrition; allintitle: ~nutrition ~animal; allintitle: ~animal ~animal ~nutrition; and allintitle: ~animal ~animal ~nutrition ~nutrition. (I wanted to show that there were differences but didn't expect how different they would all be).
Animal nutrition is one of my test searches - and using the ~ gives lots of easily explainable variations e.g. Pet Food comes up. Each of these searches gave completely different results with little overlap in the first few results (and some real surprises
A search for (no quotes) logged into google (and history not cleared) and the same in incognito mode on Chrome only had 4 out of the top 10 results common to both searches - again showing wide variations with only the top 3 results being the same.
Lesson is that lots of factors influence how search results appear on Google and it's worth changing things round to see if anything new appears.
Posted by: Arthur Weiss | October 22, 2012 at 04:30 AM
It should also be instructive that you won't see the same results if you were to try the search again, or a third time with the same terms. Goggle tracks you searches and adds to the algorithm what it THINKS you want and gives those results instead of an non-selective search.
Sometimes thinking you can go back to the search and locate an item, you can be disappointed as the hit is not there.
Posted by: Paul T. Jacksonq | October 22, 2012 at 11:26 AM
You are, of course, turning off "instant search" when testing. The increasing extent to which searchers are being second guessed by search engines is seriously undermining the value of these search engines.
I feel the search engines are becoming my enemy and I must resort to tricks to try to outfox them. Just waiting for a rogue innovation team to provide a new engine to knock Google, Microsoft, et al, out of the water just for the freedom loving joy of it.
Posted by: K Xi | September 13, 2013 at 09:53 PM