|Are we entering another Panda user-engagement calibration phase?|
I'm seeing a confluence of events that suggests we may have entered another Traffic Shaping [webmasterworld.com] period, where Google is again measuring user engagement and behavior over a range of varying results, and is possibly throttling traffic to keep data normalized for comparison purposes.
From patterns I've observed in earlier tests Google has run, I'm thinking that Google has evaluated previous Panda results, made heuristic changes in the algorithm, and is now running another "calibration test" to further check user behavior in response to changes.
I suspect that there will be a return for some of what we have called "zombie" traffic....
January 26 2011 Change - Back to "Zombie Traffic"
In that thread I commented....
|...almost every new Google search feature I've been seeing seems to be built for disambiguation, including (but not limited to) the returning of general terms in localized serps. Instant and Instant previews, eg, are in a way an extension of Suggest... and a different angle on the choices. |
You can also see variations in Suggest that aren't geographic, but which are cyclic....
Over the past several days, I've been seeing changes to various Google search refinements, suggesting that Google may now be running tests for different queries and for a slightly adjusted algo.
Sitelinks and mini-sitelinks have recently changed noticeably. Pages that I assumed would produce low user satisfaction are in fact dropping out.
Multiple results from the same domain for some queries are now being returned for longer queries, almost as if they're now checking user satisfaction on long tail.
And, as many have reported, serps have again been jumping around rapidly, yo-yo fashion.
Instant Auto-complete suggestions that I watch made a big cyclic change roughly a day or two ago.
Additionally, new genre labels introduced in US Google News results [webmasterworld.com...] suggest to me that Google may have gotten heuristics for document level classifiers to the point where they're testing them live, at least in a limited fashion on News, and that the timing of their introduction might be meshing in with system-wide tests.
For reference on how Google might be monitoring some of this, for those who haven't followed, check this thread....
Panda Metric : Google Usage of User Engagement Metrics
I'm wondering how others who have followed the above-referenced discussions are seeing how these ongoing pieces of the puzzle are fitting together.
This isn't to replace the monthly serps thread, and... please... it isn't a thread to complain about Google or your traffic.
You mean Google is testing some new changes which are causing the YO-YO effect?
Robert, I think you are spot on...You have very well summarized what I am seeing too and this has been happening for some time now.I initially saw a set of pages being pushed up for some keywords and these have now been replaced by another set of pages from different domains.
But google does keep a few domains intact on the front page, while the others are pushed into oblivion.However, there is a constant churn in positions.
what i am not certain is whether they are testing the heuristics for the document level classifiers or they have picked up a different set of niches (or keywords) to run this UE test.
If you remember backdraft7 and ohno were reporting traffic throttling and fluctuations a few months ago.But they don't seem to be complaining here recently :) Might be that google is done with evaluation of their niches.
|I initially saw a set of pages being pushed up for some keywords and these have now been replaced by another set of pages from different domains. |
By "initially", I actually meant march and early april. From april to this date, I am actually seeing a completely different set of pages in the mix on the front page.
Here is what I'm seeing in my niche.
Top 10 Results are fluctuating wildly but some domains are intact.
One hour #10 becomes #1 then drops to #11
Second hour #3 become #1 and then drops to #4.
Third Hour #5 becomes #1 and then drop to #5.
I've even seen a #30 becoming #2 and then dropping to #30 again. Is this what all you guys are talking about?
I'm certain that google is testing something but I don't know if It is "google throttling" and how long does it lasts.
One more thing, it seems like they are doing this particularly on some niches. I'm seeing SERPs from other niches and they are calm. No changes.
March and Early April "different sets of pages" have finally settled down to?
|March and Early April "different sets of pages" have finally settled down to? |
Those pages appeared on the front page post panda and they have disappeared now.I am also seeing a few pages (not all) that they replaced making a comeback to page 1. The other pages that remained intact from pre-panda continue to remain in the mix and these are from big domains like wikipedia, etc.
When it comes to throttling, it is probably accomplished by having different result sets on their various data centers.For example, even within the same country, what you see in sydney may not be seen in Melbourne and so on.
Thanks. One more thing, does the Google throttling has some sort of "ceiling" so that the result which were #1 before the throttling started can't rise till the throttling is going on?
Here is a real example.
The keyword is : abc
It is jumping the the SERPs like this :
12 May #1
13 May #70
14 May #50
15 May #43
16 May #59
17 May #42
18 May #51
19 May #54
Not sure what's up.
I don't think there is any ceiling on the ranks.It is just that the results will be different for different data centers (regions).
But I am not sure of how will the SERPS look like after these tests.Probably, people who have experienced traffic throttling and these kind of UE tests before can help.
How are you monitoring the ranks?
I'm using sescout
Traffic throttling- Description
Typically means you have a ceiling level of traffic. It comes in two varieties- total traffic and Google traffic. Once you hit your traffic allocation for the day, you get no more. Usually it's a bit more subtle- specific referrals taper off, or get replaced by those with a lower search volume.
Classic Symptom: "Buzz Saw effect" when you view your daily traffic. Especially evident when you get a traffic spike for one referral, only to see a drop for other key referrers.
NOTE: Best seen from headline stats. The more you drill-down, the less you are sure of what you are seeing
Traffic Shaping - Description
Typically two things happen simultaneously. First, Google starts showing your pages to people with different user intentions (or just turn off user-intent filtering). See Symptom (a) below.
Secondly, Google slips your site into SERPs that it did not appear in before, while removing it from others. See symptom (b). Actually, what they do is some subtle re-ordering while rotating Universal Search features through other positions, affecting your exposure.
a) Traffic on normal referrals exhibits strange behaviour. The oft-reported Zombie Traffic just will not convert. Potentially, other terms start converting better.
b) Referral patterns change. So the traffic coming from google is arriving in unusual proportions when broken down by keyword.
NOTE: You cannot see this effect in your headline stats
KEY OBSERVATION FOR BOTH
Perceived rankings do NOT change significantly- apart from possibly a yo-yo on a SMALL number of terms. Also, there is nothing to stop a hybridisation of these two tools, keeping parts of the site in "normal" conditions, while "testing" others
Thanks a lot! This is something I never read about before. May I know the outcome of this in a bit easy language?
In the example I've given about regarding my site. I've been slipped into the SERPs where I've never been before and bouncing from here to there.
Do I fit in any of these two conditions?
This is exactly what I've started to see as of 10AM on Tuesday. Some drops on core terms, but primarily restricted to long tail terms where product pages with thin content were appearing. These long tail terms, in some cases, have been bumped 5-10 spots, in other cases are gone completely, and in other cases, haven't changed at all - yet traffic for the terms that "haven't changed" has dropped - which in my mind does validate Robert's theory.
Overall I'm seeing about a 10-15% drop in Google traffic since Tuesday.
Ok, I've been doing some comparisons. I do not see clear evidence of Traffic Shaping in my logs. Our traffic is well up Y/Y, and above trend too (uptick from start of May). However, it is behaving normally, and there has been no significant disruption in referrals (proportions have changed due to uneven traffic acquisition)
The US has been ahead of the UK in the Panda deployment, so that might be it. Or it could be that different types of site are being tested. So predominately ecom were evaluated Sept/Oct 2010, where predominately info were affected by the Panda Chronicles.
|So predominately ecom were evaluated Sept/Oct 2010, where predominately info were affected by the Panda Chronicles. |
I believe this to be true. It is a set of sites different from what we heard when folks like backdraft7 and ohno were around, that are now being evaluated.
I have absolutely no idea what is going on other than several of my sites have completely disappeared overnight or been buried so far down it's not worth looking for them.
Traffic so far today is at about 25% of normal...I'm getting too old to play these silly games any more!
This is EXACTLY what happened to me on May 13. When did this happen to your sites?
My all terms are buried in deep pages like page 4, page 5, page 6 and my SERPs are bouncing like a yo-you. Rank for my pages get +25 and -27. I was #1 for all those terms. Top 10 results are changing every minute I refresh the page.
95% of my terms have dropped or say 100%. Overnight. I've 70 pages of unique 1000 words content targeting several long tails and all SERPs are nowhere to be found.
It is not happening with you alone. Can you check If your SERPs are also bouncing wildly? Also monitor the top 10 results and let me know If those results are changing.
Its SITEWIDE Drop!
One thing, according to me, your sites which have dropped should be in the similar or same niche.
I believe google is testing out some SPECIFIC niches and DOMAINS at the moment.
One more thing, I'm seeing wrong forum posts ranking with almost no content and ZERO backlinks at all. Those are 6 years old post. If google is just testing waters then OK but If this is Panda's idea of quality, Google is done.
I'm in specialised construction products basically from Brazil, China, India, Italy, Portugal, Spain and Turkey with a few other countries thrown in.
|Those are 6 years old post |
Yep, I'm seeing that a lot plus some very irrelevant old directory sites with very little information on them and the odd awful image.
Alright. Same thing as me. Keep watching the SERPs and inform us about any updates. I'm sure our sites are going to come back.
For what it is worth, the SERPS for one particular two-word keyword phrase are nearly identical to how they have been appearing for the last two months.
With one exception. A site that has been around for a long time (not mine) and probably has the BEST backlink profile in terms of links from quality websites with relevant content has slipped from about the middle of the first page to the middle of the second page.
Kind of a shame, really. But otherwise, pretty much all the same.
Different results, mixed with old, always changing. Heavy G crawling, including long 410'ed pages.
I don't necessarily think it's a calibration phase, rather a transition phase and these SERPS are irregular as they always are in between. Maybe a major update? But who knows
Update: Today we are seeing a big disruption in referrals. Nothing conclusive, but it looks like the same phenomenon I described in the Traffic Shaping [webmasterworld.com]thread
Shaddows, I am foreseeing something big. Some old pages getting hits from real people through SERPS, suggesting churns.
It is May 23 and let us see what google has in store.
indyank, let's hope. My referrals are same as post-Panda but very different from the past few days. May 23...think Goog had a 90 freeze on Panda?
|By "initially", I actually meant march and early april. From april to this date, I am actually seeing a completely different set of pages in the mix on the front page. |
Yep, my fear is that content has been devalued a lot. In April several sites I know were hit after gaining on February panda and their content is not an issue. They are down a lot compared to pre-Panda and the sites that gained a lot were just popular sites, not superior content on the subset of pages in question.
So sitewide user metrics, as defined by Google, might even overshadow content.
[edited by: walkman at 4:35 pm (utc) on May 23, 2011]
walkman, I am just foreseeing some changes which may or may not happen but I lost hopes as Google's intentions get more transparent.The changes may not necessarily benefit affected sites.