Forum Moderators: open
My logs now contain a disproportionate number of referrals from continental Europe, and my actual sales outside the UK, for what was previously a high ranking UK site - but sadly no longer - have gone up 50 fold since the Florida update.
The initial hypothesis is that although commercial results are less relevant, they are now being dished up in their charming and random way, all over the place!
There is no way Google should be serving up the same page in #1 position time and time again, when there are potentially thousands of other online business equally relevant to the query.
Given the "power" that Google has over where people go on the net, I think it is a perfectly responsible move.
Sure, the person that used to be a rock solid #1 for "expensive blue widgets online" will likely be a bit miffed, but at least it will save Google from numerous class action law suits from the owners of sites #2 - #999.
Dave
Dave
If that were the case then each data centre would show a different set of results for any give query. Each server will run its own instance of the query algorithm, and if that included a strong random number generator then each instance would almost certainly generate a different key with which to randomize the results.
The idea that Google is randomizing results is totally absurd!
I dont know about that, have you checked through your logs.
I have been checking referals , I follow back the phrase and path used to find me in Google, sometimes my page is there other times it isn't?
I'd say thats fairly random.
Karl :)
Randomised results, particularly for commercial searches have been predicted for months now.
Really? I don't see the results changing that much - they seem pretty stable, just odd!
superscript,
sorry - haven't answered your post (blush) but very interesting - one of our sites get's quite high foreign traffic for foreign searches, but I'll have a look and see if it's shifted and get back... the only shift I've noticed is down!
The idea that Google is randomizing results is totally absurd!
If that were the case then each data centre would show a different set of results for any give query.
I don't agree with you. Perhaps G is using some pseudo-random key based on date or something similar, so the results change for one day to another but not from one datacenter to another. And there are many other ways to pseudo-randomize.
I don't know if G is randomizing the SERPs, but I can assure that it is technically possible.
In some cases the randomizing is good for the web, but it is dangerous. If the results in SERPs become random, no more SEO will be needed, so nobody will improve and optimize the pages, nor exchange links, etc. So, if G abuses of this technique (I'm not sure even if they are at least using it), it will hurt the global quality of the web.
Greetings,
Herenvardö
will bring you results from the UK, Australia and many non related results.
While I appreciate that it is technically possible to pseudo-randomize, the results aren't changing on a daily basis. They have been totally stable for days now. 2 datacentres do show a slight variance, but always the same variance.
Diversification yes - a need to fine tune the broad matching criteria - yes. But no randomization.
[edited by: merlin30 at 9:48 am (utc) on Nov. 28, 2003]
in their charming and random way
Apologies for self-quoting, but I was being ironic, and not suggesting the results are genuinely random - they just look it!
But...
I agree you couldn't randomise the data in the datacentres this rapidly - but surely you can randomise the effects of a filter quite easily - and there's now little dispute a new filter (or filters) is in place.
Certainly my position changes almost hourly on a UK search.
[edited by: superscript at 9:49 am (utc) on Nov. 28, 2003]
Dave
I know, Dave - the users have to learn to search properly, right? :)
Regarding regional SERPS, the Australian ones look stable to me but the smaller index exposes glaringly poor results, particularly for adult terms. Sticky me for examples if you like - they boggle the mind.
I certainly agree that no page has the right to be at #1, and as Google spiders the web faster and recalculates relationships faster then results will change accordingly - in fact, I argued that case in a previous thread. But that is quite seperate from Google providing random results.
It matters not *how* it happens, only that is does happen.
Dave
Is anyone seeing examples of 'search in country only' searches giving unusual results post Florida?
More specifically - and this might clear it up - are there any forum members in Greece and Italy seeing more UK results? We hardly had a single sale to these countries before the Florida massacre, but now you would think we'd been running a TV campaign over there ;)
Where he disputes the commercial filter / SEO filter idea (which has never made much sense) and blames broad-matching instead. Well worth a read.
Its more about what other pages are saying, in general terms about your page. PR may now be used in a more indirect way - its not necessary for your site to have a high PR, but if a site with high PR links to and talks about your page in terms broadly relevent to the query, then you will have a chance of being on the radar. This would favour sites that have naturally garnered a rich set of back links indicating the many different qualities that your page has. SEOd sites have managed to garner a set of backlinks that talk about their page in very narrow terms - a few keyphrases. Doesn't really tell Google much about your page in general. Of course - if your a page doesn't really have very many useful qualities it's not going to score well.
I do think that the algorithm will be fine tuned in the coming months, but I think the trend is clear.