|Does Google Places contain a 'randomiser' element in its algo?|
| 12:41 pm on Jan 10, 2011 (gmt 0)|
So, Google places.
I've noticed something....weird, and I was curious as to whether anyone else has observed it.
At the same computer, same time, no Google login, 3 different browsers (IE, FF, Chrome) all provided 3 completely different results for the same search term on Google places for one of my clients. On FF they appeared on the 2nd page. On IE, the 4th, on Chrome the 6th. My decision to now work solely in FF when reporting is, of course, unrelated and completely coincidental. Honest 'guv.
Perhaps, with Places, Google has introduced a 'randomiser' to their results, to prevent accusations of favoritism and anti-competitive behavior? Has anyone else observed this phenomenon?
| 3:35 pm on Jan 11, 2011 (gmt 0)|
Personally I have no idea if Google uses randomizer or not.
However, one thing that comes to my mind is personalization through browser cookies. It might be that you use FF most often while surfing this client's site/place page followed by IE and Chrome which could be the result of this ranking difference.
My suggestion that you should clear the cache of all three browser and run the search again.
| 1:31 pm on Jan 12, 2011 (gmt 0)|
Ah, I think you may be onto something there. I had 'use search history' turned off, but it appears that only affects natural serps.
I may neglect to tell clients about this :)
| 10:07 pm on Jan 12, 2011 (gmt 0)|
I've seen no sign of a radomiser, per se, but other factors also figure into getting differentiated results.
If they're all on the same computer, each browser still reflects the same IP address.
In addition, Google has multiple data centers, each of which will have slight differences from one another at any given time. Google has seemed to sometimes assign different browser users to different data centers when delivering up results.
Finally, they sometimes may test different results, differentiating by browser type as well as other unique user factors, so you could be getting assigned some different results based upon them testing something.
| 10:51 pm on Jan 12, 2011 (gmt 0)|
|differentiating by browser type as well as other unique user factors |
I have heard rumblings of them doing this. From what I remember it was suggested that they choose a browser with a low visitor percentage and roll out changes to that group first... kind of like a 'soft' rollout.
| 3:20 am on Jan 17, 2011 (gmt 0)|
I haven't noticed a randomising effect but the phenomenon of two browsers getting different rankings has been noted before.
| 5:52 am on Jan 17, 2011 (gmt 0)|
One more possibility. Each browser has different cookies, so conceivably each could have a different location preference set.
(Setting is in the left-hand margin of the serps page, under the list of main vertical channels... ie under Images, Videos, News, Shopping, More).
Prior to the introduction of the user-set location preference, I remember that Google would be highly reactive to recent location-based searches, and, if I weren't signed in, would flounder for a search or two after I cleared cookies.