Forum Moderators: Robert Charlton & goodroi
I'm not seeing this one yet, but I'll keep an eye out. I don't usually check the same SERP frequently enough to catch a pattern like this.
between three different firms all day
Probably should make that clearer! It's not rotating every search, there can sometimes be several hours between the change. This has been going on for a few days now.
I noticed it because SERPS in our industry have been fluxuating wildly so I've been checking them alot - but on settling down there is only one position that now keep changing.
It may just be a coincidence, but it is interesting to see someone else mention position 4.
on some of the high traffic SERPs I monitor... #4 is often the place where universal search will inject something it thinks to be very on topic. It's not always news or video or anything near as clearly distinguishable... and the little line that tells you it's universal search isn't always present either.
( it's still better than when universal search puts the injected site listing(s) into the #1 position )
an ultra high competition, very broad keyphrase comes to mind where the position of these cool looking news and video listings with their thumbnails can seriously affect the traffic coming in. pretty obvious though... where it's either taking up 1 or 3 slots at #4 ( down to #6, in either case... below my site ) or 1 slot at #1 ( *cough* ) ...and displays news, thumbnails, stuff like that.
when displayed #4 and beyond... the injected news can be from blogs and forums as well, and in such cases they aren't highlighted as such. There isn't even a thumbnail. ( one test blog entry of mine made it, and was simultaneously the #1 'blog entry on topic' on CNN / in Google blog search... not much traffic though. )
I don't quite get the criteria, but it seems that if there's something that 'should be relevant' ( for example official pages or the holder of a trademark ) you sometimes see commercial sites in these spots too. Most of the 'how come this and that URL is there twice in the top ten?' cases seem to originate from this pehnomenon. Sites ranking well in web, blog and news search at the same time are usually able to grab an additional spot.
...
Also, results in universal search too are changing all the time.
with that said I don't know if this is what you're talking about.
There doesn’t seem to be any sense to these constant changes, sites jumping places all the time; but maybe its just part of the speed with which they calculate you’re worth now.
People keep saying they need a way to move beyond link pop, maybe they have.
I agree though. I think we're seeing the first steps into AI search algorithms. I honestly believe this new algorithm is learning as it goes. Sadly, it seems as though it is the crawler and filters talking to the algorithm, not user activity talking to the algorithm.
The results are not random enough. The results are not stable enough. This appears to be a learning curve which is why the results switching is slowing down as time goes on.
several days ago i noticed it with a very very competitive key phrase in my industry.
anyway i have been ranking around 15 for a long time - then it would jump to position 4 and then it would go back to say pos 15 then when i check hours later would go to 4 again.
this only happened for a few days last week for me and has since gone back to around pos. 15 for last 2 or 3 days.
This would work fairly well on commonly searched for terms.
How to choose the urls that are worth experimenting with would be another algorithmic choice, and the whole system could hook together in a big loop to inform machine learning or AI.
I noted two recent buzzwords on Googlers' lips - serendipity and diversity. A system like we're theorizing here could supply that kind of element to the first page of the SERPs.
[edited by: tedster at 3:56 am (utc) on May 7, 2008]
Re-read Miamacs post about 20 times.
Realize Google is trying to successfully monetize UNIVERSAL SEARCH, ya know, that thing they had a massive infrastructure change about? (What was it called again?)
This is, in fact, the anniversary week of when i first noticed and started bitc...err...complaining about Google throwing "news, blogs, videos" into the #4 position for "somewhat" related terms.... ie news terms for strictly commercial terms.
In other words, it's nothing new and re-read miamac's post another 10 times to let it sink in.
So I agree that using position #4 in creative ways is not new. But rapidly swithcing out different websites through that position is new.
As soon as we arrive at pos #4 it starts generating business (personally I think this position is the best - clients have the chance to look through a few sites, arrive at position #4 and think that they've searched enough and are ready to make a decision).
Mr/Mrs Google, please make us permanent at this position.
The observation that started this thread is three different websites were cycling through the #4 position. That sounds to me like a new use of the "plaything" slot that the Universal Search infrastructure created.So I agree that using position #4 in creative ways is not new. But rapidly swithcing out different websites through that position is new.
Right, but some are reading waaay too much into this.
The "switching" in and out of #4 is most likely a side-effect of the algo trying to fit Universal search properly into the results
NOT
the much-bandied-about-and-still-3-years-off, idea of Goog incorporating analytics-testing-AI into the algo.
(Didn't we have this exact same discussion for the #6 situation?!)
So while it's may be fun to envision Goog purposely "collecting data about which site should fit where according to click-thru rates", that technology being used LIVE is a couple years away.
As I commented in the #6 threads, the top 3 results of most terms are quite a bit ahead (in terms of ranking factors) than the rest of the SERPs.
So it's only natural to see the #4 spot switching off and on between the rest of the sites.
I see this everyday for tons of terms where we (and 2 competitors) rank #1-3 (and never fall below it) and also for terms where we are in the "second tier" of sites that rank #4-10
The "switching" in and out of #4 is most likely a side-effect of the algo trying to fit Universal search properly into the results
NOT
the much-bandied-about-and-still-3-years-off, idea of Goog incorporating analytics-testing-AI into the algo.
I agree that we are getting the two concepts confused but I don't think they are exclusive. I agree that the #4 switching thing has to do with the universal search system. But, it is also fair to assume that this spot is also being used to gather intel on upcoming sites that would normally not be in that area of SERPS (i.e. AI ranking rather than purely algorithmic).
Pure speculation: could it be demarcation by human review of sites that appear to be "worthy" of a high quality score, and rotating them in and out to test, to see if CTR and bounce rate corroborate it?
[edited by: Marcia at 10:44 am (utc) on May 7, 2008]
The site is very related to the search term however the site targets a section of this pluralised keyterm and not so much the keyterm in its entirety. For example we sell 'blue widgets' and the site performs at position 4 for the key term 'widgets'. This search term is quite general which other sites who offer a range of these products might have fewer bounce rates and higher click throughs as they have a larger product line.
I also saw this happen at the start of this year for the same site for specific search term like 'red widgets' which we started to rank for at pos 4. However we don’t sell 'red widgets'... we sell 'blue widgets'. We seemed to start performing well for 'red widgets' because we are a respected site and we linked out to our partner site who does sell 'red widgets' from the home page. The anchor text was 'red widgets'.
It seems to me that the site for these search terms is being tested to see if the Google or the user indicates that we should be at this position. Not quite AI but more user patterns. Our bounce rates for this search term and most probably the click through rates may not be that great but our business increased in correlation to the increase in traffic for this ralated keyterm. I guess our site convinced those looking for 'red widgets' that it was our 'blue widgets' that they realy wanted after all!
I hope that the site sill performs as it does for the term 'widgets' but if the 'red widgets' example is anything to go by this might be short lived.
Not so great for the upstarts and those with a shorter product line.
Footnote: It only seems to be term related as the rest of the site currently remains unaffected.
[edited by: Robert_Charlton at 5:42 pm (utc) on May 7, 2008]
[edit reason] widgetized [/edit]
NOT
the much-bandied-about-and-still-3-years-off, idea of Goog incorporating analytics-testing-AI into the algo.
I was skeptical of them (up until recently) actually capturing user data, i.e. click through numbers, bounce rates and such, and implementing that information into the algorithmic ranking process on a wide scale basis, but their certainly capable of it, and I'm sure they have experimented with it.
The only question is are they doing it as a core part of ranking sites currently. Perhaps the answer is no, but again you have to believe their capable of it, or darn near close.
The only question is are they doing it as a core part of ranking sites currently.
I'm convinced it is and that most everything we discuss here is. Plus, there is a whole lot more we don't discuss either because it is not public knowledge or some of it is just so darn technical that it makes our brains hurt. Google Patents for example. I've read just about all of them and I am 100% convinced that they are doing all of the above and a whole lot more! The Gorg continues to assimilate at an exponential rate.
I'm convinced it is and that most everything we discuss here is. Plus, there is a whole lot more we don't discuss either because it is not public knowledge
For the sake of argument, let's say I and my competitors know exactly the click-thru rates and, even conversion ratios of a highly searched term....
IF Google was using this data live then I would expect very different SERPs that what I currently see. Again, nothing crazy, but...
#3 spot would be #1 (by a large margin),
the #1 SERP would be #4, etc
Certain name-brand American sites wouldn't be found on the "international" serps and vice-versa.
I just don't buy it.
IF Google was using this data live then I would expect very different SERPs that what I currently see.
What you currently see and what I currently see are most likely two different things. Granted, I think they are far from being perfect in the process. But after reading 50 or so patents, some of them multiple times, I am convinced they are doing quite a bit. But again, it is still being perfected.
What exactly do you think they are doing with all those Ph.D.'s over there? They're perfecting their "bread and butter" by the minute.
And, they are not playing ping-pong and discussing the quantum theory behind Stargate-SG1. They have their own GooGate. ;)
[edited by: pageoneresults at 7:42 pm (utc) on May 7, 2008]
What exactly do you think they are doing with all those Ph.D.'s over there? ;)
Playing ping-pong and discussing the quantum theory behind Stargate-SG1?
=P
lol, if you edited your posts on my boards, i would re-edit them to make you look silly :)
Nothing involving Universal search here, only the use of the experimental spot. The phenomenon never occurs when universal search items appear in the results.
You contradicted and answered your own statement
"The phenomenon never occurs when universal search items appear in the results"
Come on folks.
Let's try to use a little scientific method here.
If it never occurs when Universal Search items appear, then that means what?
That it's connected to US?
or
it's connected to something else?
If you want to argue which types of pages are appearing at #4 (which seems to be random by everyone's observation) then that's a different issue.
It's the "experimental spot" precisely because that's where a majority of US SERPs appear.
(Again, I point you to my initial complaints about US, where US results only appeared at exactly #4)
Don't make the algo more mystical or complicated than it needs to be.
It simply isn't.
[edited by: whitenight at 7:20 am (utc) on May 8, 2008]
Um, no I didn't. If you put your underwear in your fourth drawer, and you put your gun in your fourth drawer too, that doesn't make your gun a pair of underwear.
And there is nothing random about the pages that appear in the #4 slot. In fact it is quite non-random whenever I've seen it.
Without repeating all the exact same arguments for when this same theory was proposed in the #6 issue...
Can anyone, ANYONE explain to me why it's #4?!
Why not #3, #6, #8?
Most importantly, why aren't we seeing all the top 10 pages rotating thru the #4 spot?
I repeat, why isn't every single top 10 SERP result being "tested" at the #4 spot? What about the #3 spot? #2?!
If anyone can even give me a half-way logical answer based on how Goog has "tested" parts of the algo in the past, then I'll entertain the argument.
Saying we can put someone on Mars, and actually doing it are two different things.
Just because something is "possible", doesn't make it so....
Same with "patents", "theories" and, "predictions"...
There's a reason I mentioned Stargate...
