Many Weeks since the Panda Update - Any Improvements? [part 2]

[rustybrick]

< continued from [webmasterworld.com...] >

So, still, no one is seeing any significant improvements?

[edited by: tedster at 5:00 pm (utc) on Mar 25, 2011]

[tedster]

That does seem to be the way it is right now - but I also expect it to be tweaked in the near future. The emphasis on "brands" was part of Eric Schmidt's regime, but the area of original source is also a focus for Google, especially in News syndication where they are trying some new met-tags to see if they can help.

Google has conflicting elements in the current algorithm (freshness, authority, original source, popularity) and I'm convinced they are going to work on that balance. The Scraper Update in January 2011 (really, an original source update) was a start but it didn't do a very good job. Still, it shows that Google knows they have an issue here.

[kd454]

I just got another good KW that popped back to #3 from no where that was hit by Panda.

Funny thing about this instead of it bringing my homepage back it found a "thin" content page I missed/did not delete about the subject and put it there instead.

I found another fresh/orignal 500+ article on the same keyword/subject and replaced it and added a couple of words to the title and desc, but left the url the same. I have no idea what Google will think of this.

[walkman]

kd454, I am noticing many changes too but nothing to improve total traffic. Started yesterday for me. Overall Google traffic is still way off. I have deleted many pages and added some things JUST for Panda but no luck so far.

Tedster, I think this started with the long tail update: Matt Cutts said that we want to send long tail searches to more established sites. Not the exact word he used but the same message. So they will assume that for "somewhat obscure search" CNN has X more chance of having the better match, simply because it's CNN, not because of what the text on page says. That's how I read it.

[tedster]

walkman - yes I think there is a connection, maybe even going back to the Vince Update [webmasterworld.com] in 2009, before the Caffeine integration.

[ascensions]

One really odd facet of this whole thing that conflicts with your opinion Tedster, is the complete removal of entire domains from the index. No penalty, just gone. Nothing under the site: command. These aren't sites that are in any which way ones you'd ban... and even if you thought the content was "thin", you could knock it down a few hundred pages and retain the content... but that's not happening. PR5 and up sites of 5 years of age- gone.... like Davros and "The Stolen Earth"... Where's Doctor Who when you need him?

[browsee]

@walkman, I thought DaniWeb is back to normal based on their Alexa stats. Looking at the comments on Google forum, they are still down.

[TheMadScientist]

These aren't sites that are in any which way ones you'd ban... and even if you thought the content was "thin", you could knock it down a few hundred pages and retain the content... but that's not happening. PR5 and up sites of 5 years of age- gone....

I think you're forgetting the VAST amount of information they're pulling and pushing around ... They're bound to have errors and 'misses', even prior to this system ... Sometimes when I read posts it seems like people think they look at the sites and pages and then make a determination, but they really can't do that on a system the size of their ... What they have to do is perform the calculations and then look at the results (<-- That's a huge distinction) ... If they looked at all the site(s) and pages not included to make an evaluation of the system it would take eons, because there are many more non-results than there are results in the top 10.

When the situation is results oriented, not page or site 'missing' oriented, they can determine 'are these better or worse' than what was previously here ... Then like Singhal said: They go back an try to improve again ... To really 'get' why some of the things that happen, happen like they do, imo you really have to understand how they have to look at things ... If you have a DB of 1,000,000 pages for a single phrase, are you going to review them all if you are trying to present them for visitors, or are you going to perform a set of calculations and look at what you're presenting, then try to improve on that ... You really CAN'T get into looking at every page or even every site with a system the size of theirs, any more than an individual could review a million pages and try order those ... The pages can change or be updated at any time, so what you saw today might not be what's there tomorrow ... So, you have to have a system that is 'reasonably good' and then improve on it from there WITHOUT getting caught up in trying to review each and every result NOT presented.

I'm not saying they don't ever look at what's not there, but rather: What's there and how that compares to what was there prior to a change is much more important.

[ascensions]

They're bound to have errors and 'misses', even prior to this system ...

I agree, as does anyone who write algorithms. One simply line of code can go un-noticed for years yet leave a huge gap in your end-results.

However, elimination of entire domains is not an "error" I would equate with the caliber of skill Google has and not a likely part of an intended algorithm function. If it was intended its nothing short of a computerized attempt at prejudice. A "cleansing" of the index based on their mathematical interpretation of what doesn't fit their "blonde hair, blue eye'd- website" ideals and it sound vaguely like they may be assuming their vision of the internet is what everyone should be forced to think like as we interact with it.

[TheMadScientist]

However, elimination of entire domains is not an "error" I would equate with the caliber of skill Google has and not a likely part of an intended algorithm function.

The chances are they don't even know about it is closer to my point ... It 'fits' (or doesn't) a pattern detected (or not) by an algo, and even if they do know about, reworking the algo to include one site (when they say they do not have an exception list for this piece) doesn't seem like it would be a quick process, but my well be 'fixed' in an updated version ... What I'm saying is: I doubt it's 'on purpose' and 'accidentally dropping' an entire domain when they're switching to a site scoring system of this magnitude seems entirely possible (actually likely) to me.

[ascensions]

I've seen Tron..... If Google sucks me into the grid, please tell someone at Bing to come save me.

[TheMadScientist]

lol @ascensions

[falsepositive]

Looking better on my end over the last 3 days. Monitoring things though. I did a TON of changes.

[crobb305]

Looking better on my end over the last 3 days.

I see some improvement also. Main phrase that fell about 600 positions has gone back to the top 100. That's a big improvement since Saturday.

[My_Media]

Hey Guys,
For my UK, Panda landed on the same day as the US, Feb 24. My London traffics dropped 60% on Feb 24. So again Google has random picking the site for this surgery.

[superclown2]

Yes I'm in the UK and the same thing happened. My site has a lot of links from the USA and I wondered if this was relevant.

[ken_b]

I get the impression that there haven't been any big sustained improvements reported. Mixed bag for me. Some improvement in traffic, but also some increased loss of rankings, depending on the section of the site. Hard to know if the traffic increase is because of better raking for some terms or if it's just following a several years long pattern of seasonal traffic ups and downs. Need to dig down farther in my stats I guess.

[Jane_Doe]

I get the impression that there haven't been any big sustained improvements reported.

Some pages that had minor flaws that I updated seem to have moved up. The pages that had the most issues are still in the dog house even though they have been totally refurbished.

Overall though the serps for the areas I am in are still seeing big changes. Some of my sites that did better in Panda, ones that I haven't changed in over a month, are still seeing some continued increases over the past couple of weeks.

[BrianS]

My_Media

For my UK, Panda landed on the same day as the US, Feb 24. My London traffics dropped 60% on Feb 24. So again Google has random picking the site for this surgery

Panda did not land on your site on this day, it landed on it's US search results and consequently your US traffic.

Panda specific site/page classifiers would have been implemented many months before, perhaps even upto a year before? repeatedly tested and tweaked before actual making use of.
Using these Panda specific classifiers + many of the existing ones, it was the Panda Filter Layer that was applied to the US search results on this date.

[outland88]

Jane you're going to get the changes with basically any attempt at improvement. Most is the same as in the past. It’s when they come around with another supposed quality re-evaluation like Adwords that it hurts. Many people though were dealing in years old static junk to begin with. A few changes may just send the same pages back into its past pattern. Plus many people are going to feel like geniuses or they can beat the algo for a few months. All you can do though is play the hands dealt to you and try and improve on it.

[Shatner]

Things took a step back today.

Previously some of you may remember that I reported new content was ranking well, even if pre-Panda content remained buried.

Today all that new content which was ranking well, was deranked and dropped as much as -100.

Does Google's Panda penalizer just take the weekend off or something and then re-nuke everyone on Monday?

[walkman]

@ Shatner
I think we see and we'll see many ups and downs as google tests, moves data around and changes things. But I guess we'll know when Panda 'penalties' are lifted, just as we knew when it hit us. So far, nothing, just a day up 12% here, a day down 9% there and so on.

[ascensions]

Ironically, this was a "farm" update, but "farms" are outranking me with MY content that was written years

Now there's a demotivational poster... that's going on to my wall.

[AlyssaS]

I think there was another tweak today. My one Panda affected site moved back to page four from it's heady position at #2 on page 1 since 22nd March (prior to Panda it was at the top of page 2). My other sites continue to remain unaffected.

I think G did it's usual thing of opening it's filters before making a tweak and then closing them again.

I keep an eye on the stats of the larger content farms as it's easier to spot trends looking at large sites that you don't own :-) Take a look at this graph of Hubpages from quantcast (you will need to change the graph to US only to see what I mean).

[quantcast.com...]

There's a spike up around the 25th of March.

Quantcast also tracks the stats of Articlesbase, another site that got dinged by Panda. Curiously their stats show no spike:

[quantcast.com...]

Did some digging to see what the response of each of these two sites to Panda was. Here's what Articlesbase did [blog.articlesbase.com]:

1) increase their detection of spun articles (they delete any they think are duplicates
2) increased the minimum word count to 350 from 250
3) stopped authors putting links at the top of the article, they can only place them in the author bio section, which remain no follow.

Hubpages have taken a different approach [hubpages.com].

1) they changed their ad layout so there was no ad within the content
2) they required that intros to hubs should be text and image, no amazon, ebay or rss feeds right at the top.
3) they required hubbers to increase the text to make sure there was at least 50 words for every Amazon or ebay product featured.
4) they observed that hubs that had fallen back had rss news feeds and other feeds, and the active hubber population appear to have taken this to heart and gone on an rss deleting spree.
5) active hubbers also appear to be deleting any of their own hubs that appear to them to be weak, plus have gone on a rampage to root out spam on the site and report it (which hubpages deletes as soon as they can). The spam includes stuff that isn't duplicate, but just badly written and yucky.

I just thought peeps would be interested to see the differences, and to see how hubpages spiked when articles base didn't. So maybe with the filters temporarily lifted, they looked better to G than articles base?

Any comments?

[tedster]

If you change from "people" to "pageviews" the spike for hubpages becomes twice a big, matching their pre-panda levels. Don't what that means altogether, but it's an interesting factoid.

[My_Media]

Guys,
With the other side of the world is doomed to drop in traffics, that means there another major drop coming to our existing dropped traffics.
What is Google trying to do?

[crobb305]

If you change from "people" to "pageviews" the spike for hubpages becomes twice a big

Check out the "page views per person". Enormous (unrealistic?) spike around March 25th (coincident with the increase in traffic). Both of the hubs AlyssaS mentioned show an increase in page views per person beginning Feb 24th. Could more pageviews per person after the update = quality indicator (to the algorithm)?

[tedster]

With the other side of the world is doomed to drop in traffics, that means there another major drop coming to our existing dropped traffics.
What is Google trying to do?

We need to look at a bigger picture. I work with some sites that dropped, but with more sites that increased due to Panda. The total of search traffic didn't change, but essentially Google's message is "up your game". It's not all doom and gloom - just another change to deal with. Prediction: big changes will continue to happen in the future.

[walkman]

Just looked around Hubpages and I have to feel Google's pain. People are becoming 'experts' on guitars, heart surgery and rockets at the same time just to get someone to click on an Amazon or adsense link. The way that system is setup those pages will/should gravitate towards page 1 since the authors will link at them from other pages, comment on each other hubs, twitter, FB and so on.

To summarize, looks that no one has really come back--as far as we can tell--from Panda. A temporary spike here and there, a page or two doing better, others worse, but no "I'm back!"

Is it possible that no one has fixed their sites to please Panda? I doubt it. Assuming people tell, this points out to the fact that either Google is gathering user data (unlikely IMO) or that they haven't run the re-calculation to reassign the quality score.

[browsee]

Is it possible that no one has fixed their sites to please Panda?

I fixed everything I can think of. Right now, I have zero thin content(except login and register), only quality pages. I don't have many images on the site, so we are adding images as well.

[walkman]

browsee,
have your Pandalized pages been indexed by Google? If so, when?