Panda 20 - a Full Update - 2012-09-27

[SnowMan68]

Seoroundtable has reported that Matt Cutts confirmed a large Panda update occured on 9/27.

Update: Matt Cutts told me this is a fairly major update to the Panda algorithm affecting 2.4% of search queries. More on this tomorrow.Barry went into more detail at searchengineland [seroundtable.com...] [searchengineland.com...]

[edited by: Andy_Langton at 7:23 pm (utc) on Oct 5, 2012]
[edit reason] Corrected link [/edit]

[claaarky]

Zeus, I think that the Panda rating of existing sites would be reset once all their Google traffic has totally gone for a long enough period, say 1-2 months, and the site would become eligible for traffic again, like a new site (assuming they were only affected by Panda). Like a kind of extreme version of the noindex solution many have claimed helped their sites recover partially while they improved the quality of those sections of their sites.

However, if the site quality hasn't improved when traffic returns then the site will just be demoted again.

Regarding where the metrics come from, I still haven't seen a quote from Google that says Chrome data isn't used for Panda (I've seen a quote that says data from Chrome is not used in the main algo, but that's not the same as saying it's not used for Panda).

It's definitely not Google Analytics though. The data in there is too inaccurate (even Google have said GA data should be used as a guide, not relied on 100%). Chrome data though - that would be very accurate.

[diberry]

Leosghost, I just Bing'ed this Wireshark, and I think it's over my head, but that's really interesting info. Wish we'd had it during this thread: [webmasterworld.com ]

[Leosghost]

I'm certain that I ( and others ) have posted about the existence of Wireshark in the past ..I think I even mentioned it before in relation to what Chrome gets upto..

I was following the other thread, and did think of mentioning it again in there..but didn't really have the time to get into the inevitable "no they don't" / "yes they do" arguments that I anticipated would follow..( it was a busy 10 days or so for me with other stuff )..folks here like to try to put one 100% into the G fanboi or G haters boxen..never thinking that there might be other boxen.. sometimes trying to spread a little light and get a little thinking going gets wearying :)..TMS and some others who don't post much nowadays knows what I'm talking about ..

Chrome was "given" to the world for just one reason..( detailed tracking of what people do on a page by page basis ) and the built in default browser on android mobile devices does very similar..( although it can't track mouse movements as they have no pointing device/ mouse )..this is why Chrome runs autonomous tabs..each tab open is a distinct interface..what you do within it and whether the pages call double click etc, or your mouse hovers over this area or that area, whether you scroll the page or not..all is available to be phoned home..

As each tab is separate it is easy to ID which is "in focus"..so open tabs that are not being viewed do not "screw" the data..

They launched chrome in 2008
[en.wikipedia.org...]

that is 4 years worth of data.( they "open sourced" "some" but not all ;) of the code later..the important parts are still not "open source"..


4 years of data..25 to 30% penetration as the browser on desktops ..multi platform..runs as a different version as default on all android phones and web capable tablets..

Is being constantly silently updated in android devices..

They have the means to gather all the data that they need,( without using Ganalytics) and they have done so, are doing so, and use/d it..( not in the "main algo"<= they were careful to say that..but there is no longer really a" main algo"..so that denial was semantic sidestepping ) ..How many android users change the default browser ?

Why did they make an addon for IE ?

What users do on pages is recorded with javascript by all the "independents" via overlays.3.

G developed and deployed what is a multiple tab separate javascript per tab browser..

Take a look at the source code of any G search page after you have pressed "search" and got the results back..see all that javascript ? ..who hosts the jquery libraries that most people who use jquery like to include calls to ?

G know more about how to use javascript and to get data about what you are doing in real time on a webpage than anyone..

[diberry]

*claps*

Part of why I started that thread was to show that it didn't seem possible Google could have enough info to make user metrics a factor WITHOUT using Chrome or something. What you're saying explains a lot. And as you point out, technically they're not lying.

It also explains a mild skew I see in Google, which is that it's very good on certain queries, but not on others, as if it "knows more" about those topics. The people likely to use Chrome are going to be a bit more tech savvy than the typical PC user. Topics that attract people who aren't tech savvy but use the web to learn, shop and network - gardeners, knitters, golfers, for example - seem a bit less robust in Google queries to me. These types of surfers aren't downloading Chrome - they're using what came with their computer, so unless they have Google toolbar installed or get an Android device and do significant browsing (rather than just downloading emails and videos of family and stuff), they're not sending Google much info on their browsing habits.

My own take on Google is not personal in any way. They aren't my friend or my enemy - they're a really huge corporation that's under huge pressure to make ever-increasing revenues. And the longer a company stays on top, the fewer ways there are for it to increase revenue - no matter how nice or honest they may be, eventually they HAVE to play dirty. That's the nature of largely unregulated capitalism, and Google didn't invent that. (And between you and me, even if all the worst things we suspect about Google are true, they are still more ethical than a lot of oil companies, pharma companies, etc. - and facebook, LOL.)

[gouri]

If you have Google Chrome on your computer, but aren't using it to surf the internet, can Google still see how you are viewing a page (e.g. time spent on page) and look at other user metrics that might then go into determining its rankings for keyword phrases?

[Leosghost]

No..Chrome browser only knows about what you are doing in Chrome browser..it isn't a trojan/keylogger/virus/malware..it "just"* ( without telling you, or allowing to switch off all such "reporting" ) talks to the mother ship a lot / sometimes about what it is doing,( and by logical extension you ) are doing with it..

*as Lucy24 would say :: for a given definition of "just" :: ..;)..

[Panthro]

Thanks Leosghost. Very interesting.

[Robert Charlton]

Google Chrome sends back all kinds of data.

G says no..

Wireshark says G are telling porky pies..;)..

In addition to diberry's Oct 8, 2012 thread...

What user metrics does Google use to determine rank?
http://www.webmasterworld.com/google/4505614.htm [webmasterworld.com]

...we have this dedicated Chrome discussion, from a months or so before, on whether Google's organic algo uses any Chrome data....

Matt Cutts: Organic Algo Does Not Use Any Chrome Data
Aug 23, 2012
http://www.webmasterworld.com/google/4487777.htm [webmasterworld.com]

The sources that ergophobe references in the Chrome discussion, including official Google sources cited, fully acknowledge that data is being sent back to Google. I don't think that's the issue. The question is what data might be privileged to Chrome vs any other browser, how to interpret what Google's saying in its privacy policies, and what, if anything, might be used by the organic algo.

I'll be kicking the latter thread up, hoping that Leosghost might be willing to share what he's seeing on Wireshark with various policies and theories discussed on the Chrome thread. Probably the best place for his comments is on the dedicated Chrome discussion.

[n00b1]

Some interesting discussion regarding dwell time and bounce backs as a possible signal for Google algorthmically - 49:20 onwards on John Mueller's 'Webmaster office-hours, 22nd October, 2012'. This video can be found on on YouTube or his G+ profile. He basically explains that these aren't signals Google could really use for its algorithms and that it isn't a good signal of 'bad quality' if a user bounces back after viewing a page and is often quite normal.

[zeus]

n00b1 - bounce rate is no good value, but still google use it, im sure, with Panda and time on site.

[gouri]

No..Chrome browser only knows about what you are doing in Chrome browser..it isn't a trojan/keylogger/virus/malware..it "just"* ( without telling you, or allowing to switch off all such "reporting" ) talks to the mother ship a lot / sometimes about what it is doing,( and by logical extension you ) are doing with it..

*as Lucy24 would say :: for a given definition of "just" :: ..;)..

@Leosghost,

Thanks.

I appreciate the explanation.

[n00b1]

Zeus, so are many paranoid webmasters. I don't see how you could be sure about that and the Google engineer seems to suggest otherwise.

[Vuffy]

Everywhere I read that Panda is either a site wide or sometimes sectional issue. My question is for anyone who is positive that they're affected by panda: if you've been tagged by Panda, are you still able to rank #1 for some keywords or is that an impossible thing with panda.

I'd like to know for sure if I'm affected or it's just the regular search algorithm. Although traffic is 30% lower than what I believe to be my pre-panda, I still have a lot of keywords ranking 1-10.

[claaarky]

Vuffy, Panda hit sites can still rank really strongly and hold top spots. The demotion varies across a site - some sections can be completely untouched. If you check your traffic graphs section by section you'll be able to see how much each section has been affected.

[Vuffy]

Thank you for the clarification.

[petehall]

@Vufffy you will find positions you do hold will decline with each update, I would resolve any issues ASAP.

[Vuffy]

I do see some declines, but some are actually climbing as well.