PANDA International - Data on Winners & Losers

[tedster]

This afternoon the SearchMetrics blog released their data on the international wave of the Panda Update - affecting English Language search. The article includes data for 100 losers and 20 winners, along with some dramatic graphs.

The data was collated between the 5th & 12th of April 2011 and we can definitely confirm that the update has hit the UK – in a big way.

Surprising is that ehow.co.uk and ehow.com has lost more than 50% in visibility and the alarm bells are probably going off at Qype.co.uk who’ve lost a whopping 96%! A lot of price comparison sites like ciao.co.uk (in a lawsuite with Google) and dooyoo.co.uk also lost nearly 90% of visibility. [blog.searchmetrics.com...]

This thread is for discussing the data - if you wish to share editorial opinion, positive or critical, please post in our other thread: Google PANDA rolls out WorldWide [webmasterworld.com]

[Dan01]

Still no ranking improvement, but I didn't drop any further on the second update Monday. Traffic is about the same.

Crobb, have you been printing your SERPS? I made a hard copy of some of my SERPS over the years. I punch holes in the paper and put them in a binder. That way a week, a month, a year or whatever from now, I can go back and we where I was.

Traffic can be fleeting. For instance, if you rank on cancer, there may be some new research on it and you might get more traffic because of that.

[Dan01]

For me, SEO is like a science. We need to look at the data and try to figure out what Google's code is. What their algorithm does. Obviously, producing good content helps, but if no one sees it, what good it is?

[walkman]

Google may lose some money from the Panda update or the wealth could be redistributed from one publisher with Adsense to another with Adsense.

But Google still makes most of its money off SE ads, so they have to take the risk of losing Adsense Network ad revenue. Last time I looked Google made about 80% from SE ads and 20% from Network ads.

More than a bit, since many more will remove them "just in case." And can't blame them

[Shatner]

So have any of you actually gone through the full list of losers for the UK rollout yet?

It's actually a little different from the Panda 1.0 list. The original Panda 1.0 list had a WIDE variety of sites and while some of them made a lot of sense and were clearly spammers/keywordstuffers/content farms, I'd say at least 30% of them didn't.

This list, nearly 100% make sense. Every site listed fits into one of these categories:
- Spammy coupon site/shopping site extreme low quality
- spammy IP, website tracking site
- Clear, obvious content farm
- User generated content

These are the only sites on the list that are an exception to this, and these are probably the ones we should be trying to analyze to see why they got hit, because it's not obvious:

techradar.com
reghardware.com
pcadvisor.co.uk
techwatch.co.uk
just-food.com
computerweekly.com

[Shatner]

For comparison, here's some of the sites from the original, Panda 1.0 rollout which never really seemed to make sense on the biggest losers list:

prnewswire.com
cinemablend.com
blogcritics.org
digitaltrends.com
technorati.com
daniweb.com
popcrunch.com

[walkman]

See here [google.com...] Shatner, many, many good sites being hit for nothing (other than probably too many Adsense blocks)

Those on the hit list have already made their millions, they can ride it out.

[koan]

other than probably too many Adsense blocks

If that's the case, it's easy to fix.

[pontifex]

in the top 20 list is one site full of hardcore #*$!... har har har, back to square one.... This is all so funny (and sad)!

[7_Driver]

I think the winners list is quite striking actually - even more so than the losers list: Not many small independent websites on it are there? It's almost entirely big brands and multi-million dollar corporations - which continues a "brand centric" theme that we've been seeing from Google recently.

I think they're now trying to produce "safe" search results - in the sense that they're good - but not necessarily great: Search results that are difficult to criticise. Showing more and more big brand sites is Google's version of the old mantra: "Nobody ever got fired for buying IBM".

[walkman]

Yp 7_driver,
Google is purposefully killing small businesses online, let's hope we kill or hurt them too in the process so we may have a chance later. The only thing you can control, content doesn't matter now, so we're screwed without million dollar budgets.

Go ahead and quit your job for an online business or hire an extra employee in today's Google climate

[darkyl]

The real problem I see with the quality of post-panda results is the harshness of the demotion of the sites that have been hit.

For any given search term there AREN'T hundreds of quality related pages available on the web. Pushing several websites down hundreds positions automatically makes unrelated, low quality websites go up in the serps.

Maybe that content farm article didn't deserve the top spot but was still more useful to the user than the less related pages that show up now.

This was already happening before Panda, but it got worse.

[pageoneresults]

So have any of you actually gone through the full list of losers for the UK rollout yet?

I did. I sampled more than a handful. I looked at things from a technical perspective as opposed to that which one may "see". You know, stuff behind the scenes that MANY SEOs fail to look at.

For example, I looked at EzineArticles first. Did you see that their home page has 754 links on it? And about 750 of those are internal? Do you think that might be a bit overboard? Taxonomy Turmoil.

I looked at more than a handful of others. You know what I found? Crawling nightmares! Poor server response times. Too many http requests, I found some with upwards of 300 requests per page, do you think that would hamper performance at all?

I also found a multitude of hard coded internal redirects. Lots of 301 > 200 and 301 > 301 > 200. Many of these folks have failed to maintain quality control internally. Google states this about internal redirects...

Never reference URLs in your pages that are known to redirect to other URLs.

Minimize Redirects
[Code.Google.com...]

I wonder how many of you have actually read (not skimmed) the documents Google have posted about site performance? They have quite an extensive library of help topics related to this. Not to mention all the tools in GWT for monitoring.

The most common thing I found amongst my sampling was performance issues. Did you know that redirects cause an additional RTT or two, or three, depending? You want to know what type of response I get from most Webmasters when I ask them to clean up their redirects?

You expect webmasters to constantly monitor old links to see if "someone else" changed them? I don't have time for that.

Yes, I expect Webmasters to constantly monitor their internal and external links. I look at hard coded 301s in this instance as link rot. Google also say...

Your application needs to have a way of updating URL references whenever resources change their location.

So, instead of looking at all of this from the SURFACE, how about digging a little deeper and take a look at the entire picture. I'll bet a majority on the loser's list present a very poor browsing experience for the user.

[browsee]

techradar.com
reghardware.com
pcadvisor.co.uk
techwatch.co.uk
just-food.com
computerweekly.com

As discussed here [webmasterworld.com...] 5 sites are using dark colors and 2 sites are using ads above the fold. I am very confident that the look and feel(not necessarily web 2.0) is playing a major role here as they are trying to implement user behavior as "quality".

[Shatner]

>>2 sites are using ads above the fold

Don't almost all sites used ads above the fold? Most advertisers require it. I'm not sure we can really consider that as a factor. Most of the sites on the "winners" list have that too.

[indyank]

where is the winners list?

[tedster]

Top 20 winners are toward the bottom of the SearchMetrics page [blog.searchmetrics.com], with a link at the end to download the top 100 winners.

[walkman]

The winners don't match the losers at all. Looks like Google is sending everyone to eBay and newspaper sites.

[econman]

Most intriguing to me was Amit Singhal's comment:

In addition, this change also goes deeper into the "long tail" of low-quality websites to return higher-quality results where the algorithm might not have been able to make an assessment before....

What is this kind of "long tail"?

I'm intrigued by this comment, as well. When I read it, the reference to "the "long tail" of low-quality websites" struck me as perhaps referring to all those millions of small sites that aren't operated by the content farms, the Fortune 1000, and the big media brands.

From Google's viewpoint, I suppose any site that isn't in the Alexa/Quantcast/Compete top 100K could easily be viewed as part of the "long tail" in a sense, (even though it might be one of the top 5 sites in its own narrow niche).

Obviously, quality varies widely within this huge swath of sites, which is awash in millions of pages of machine generated garbage, along with the lovingly hand-crafted pages published by small publishers, hobbyists, bloggers and so forth.

I wonder if Google is gaining confidence that their new system can help them separate the "good" documents from the "bad" ones within this portion of the internet?

If so, their confidence is probably misplaced, or they must be basing their judgement on where they think the machine learning system will be in a few months (as opposed to where it is right now).

[tedster]

Re-reading it now, it sounds like a kind of dumpster diving - taking another look at essentially low quality websites (according to Panda) to see if they have at least some pages that are still gems.

<added>
It sounds as if Panda Two might be applying an "exception list" flag to some pages, in cases where the whole site was demoted by the original Panda roll-out due to lots of low quality page scores. If that is what's happening, some of the small traffic gains that people are reporting recently might be attributable to this factor.

[crobb305]

It sounds as if Panda Two might be applying an "exception list" flag to some pages, in cases where the whole site was demoted by the original Panda roll-out due to lots of low quality page scores.

There does seem to be a big difference between Panda 1 and Panda 2. And I still see my site back to full pre-Panda rankings on the new international Panda. It leads me to believe this will be applied back to the U.S. at some point. I can't explain any other reason why I see such a huge difference in ranking between the U.S., and Google international (.es, .fr, .se, .de, etc).

[browsee]

@crobb305, how are your serps in .co.uk, .ca,.co.in and com.au?

Did they release panda in .es, .fr, .se, .de?

[AlexB77]

Reply to indyank

"Google Adsense" are a factor in this algo. I have removed adsense in one of the sites and it did improve.

the fact that you have removed your adsense only has one significant effect, which is your page loading speed, which would definitely be faster then when you have ad an adsense code, which explains why your site had an improvement in serps, but I have personally experienced 65% of a drop in US visitors on one of my sites mainly from 7th of march that has adsense links in 4 different places including top left corner and on 18th of march had a first ever phone call from Google Adsense team member, who said that I should use more link units in other parts of the pages as well as allow text and picture adds to improve CTR and overall revenue. Funny to say, but why on earth would google target any site that has adsense links that basically generates a huge profit for both google and publishers. Don't anyone think that this would be equal to cutting branch on which you are actually sitting?

[tedster]

I agree. Lowering rankings for SOME sites with Adsense is a side effect of Panda, not the target.

[walkman]

I agree. Lowering rankings for SOME sites with Adsense is a side effect of Panda, not the target.

OK, Tedster. You a Google engineer, you know data driven :)

Let's say stats show that 95% of spammy sites have Google adsense. What do you do with that obvious signal ;) ?

[tedster]

First, before you establish any signal, you'd have to look from the other direction. Out of all the Adsense sites what percentage are showing the "shallow content" signal?

When Google's machine learning program ranged over all the possible data points in order to generate what became Panda, I'm pretty sure that the presence or absence of Adsense couldn't have been chosen as a decision gate, because way more good sites also run Adsense. That would make it a bad signal.

[synthese]

@crobb35 Can you elaborate on this a bit?

And I still see my site back to full pre-Panda rankings on the new international Panda.

Are you saying that your US traffic is back to pre-Feb24?

[crobb305]

@crobb305, how are your serps in .co.uk, .ca,.co.in and com.au?

Did they release panda in .es, .fr, .se, .de?

Browsee,

My site has never ranked well on .co.uk, .ca, or .co.in. The Google index that I see for my search queries on those sites are completely different than what I see in the U.S., I don't see ANY of my competitors ranking there. I see Australian sites on .com.au, UK sites on .co.uk.

As far as .fr, .se, .es, .de, .co.nz, and some others, when I search on my normal queries, I see my site ranking as it did here in the U.S. before Panda (normal #1 rankings with sitelinks), but the more I look at those searches, the more it really does look like PRE-Panda. I see an EzineArticle ranking in the top 3. So maybe Panda isn't located on those sites yet. I don't know why not, if Panda has been rolled out worldwide.

Are you saying that your US traffic is back to pre-Feb24?

Synthese,

No, I have not seen any improvement at all in U.S. traffic or rankings. If anything, since Monday, I have dropped one or two spots for one of my main terms (page 3 for my main 3-word phrase, page 56 for main 2-word phrase). I just see my site ranking well on some of the international (English) Google sites. Probably doesn't mean anything the more I think about it.

What do you guys see when searching on those Google sites? Pre-Panda?

[walkman]

First, before you establish any signal, you'd have to look from the other direction. Out of all the Adsense sites what percentage are showing the "shallow content" signal?

When Google's machine learning program ranged over all the possible data points in order to generate what became Panda, I'm pretty sure that the presence or absence of Adsense couldn't have been chosen as a decision gate, because way more good sites also run Adsense. That would make it a bad signal.

Fine. Let's agree that if your site is borderline adsense can or should pandalize you. It is a signal and very clearly linked to MFA. Google might /should use it as a confirmation that if your site that meets x number of MFA criteria, it is made for Adsense if you have adsense all over your pages.

[chrisfurther]

The owner of discountvouchers.co.uk (one of the sites listed as being hit hard in the UK) is disputing the data and is aparently seeing no drops in traffic. Check out the comments here: [further.co.uk...]

[tedster]

Hello Chris - and welcome to the forums.

Thank you for that important bit of information. we normally have a very strict policy that members here may not lin to their own content, but for the first time in almost 12 years, I'm going to allow an exception for this case.

Confirming your report, Doug Scott, the owner of DiscountVouchers, made another comment on BlogStorm:

Doug Scott 13 Apr 2011 at 5:51 pm

http://www.discountvouchers.co.uk

We are listed as being hit hard by all the reports but we are seeing no effect in our logfiles.

[blogstorm.co.uk...]