New Panda Update - September 27

[BrodyDodes]

Google rolled out its most recent iteration of the Panda algorithm on Wednesday 9/28/11 (or Tuesday according to who you ask). Saw the first recovery of one of my punished sites since the first iteration back in February. Anyone else see recoveries?


[webpronews.com...]

When asked if an iteration of Panda was implemented this week, a Google spokesperson told us, “yes.” She also provided the following statement:

“We’re continuing to iterate on our Panda algorithm as part of our commitment to returning high-quality sites to Google users. This most recent update is one of the roughly 500 changes we make to our ranking algorithms each year.”

If you’ve followed the Google Panda update saga throughout the year, you may recall Dani Horowitz’s story. She runs an IT discussion community called Daniweb, and it was hit hard by the Panda update, but she made a lot of changes, and gradually started to build back some Google cred

[Hissingsid]

Oftentimes mis-matched traffic is the result of an ambiguous search term. In any case, a high bounce rate for one term logically shouldn't affect your rankings and traffic for other terms.

Not even if it affects your site's overall bounce rate?

What if Google is using some semantic web analysis as part of this?

The fact is that there is some indication certainly from the terms that are appearing in my logs that there is some mechanism that Google has introduced that put my site in a place that gets traffic for long tail terms that I wouldn't target. This has resulted in spikes in bounce rate and a higher average bounce rate.

It seems to me to be logical that these long tail terms are appearing either because of a change in semantic indexing or Google search suggestions that are ajaxed in below the search field or a mixture of both.

It also seems perfectly logical that Google may be looking at overall bounce rate for a site in comparison with other sites in a broadish semantic topic area.

[netmeg]

I don't understand the mismatched traffic thing. I mean, you have a title and meta description that adequately describes the URL, right? The traffic isn't being sent directly; the user has to click on something. And granted Google will rewrite elements, which is annoying, but I haven't seen a rewrite that's THAT off the mark. So how is it Google is suddenly sending mismatched traffic?

[walkman]

I don't understand the mismatched traffic thing. I mean, you have a title and meta description that adequately describes the URL, right? The traffic isn't being sent directly; the user has to click on something. And granted Google will rewrite elements, which is annoying, but I haven't seen a rewrite that's THAT off the mark. So how is it Google is suddenly sending mismatched traffic?

Netmeg, your sites aside (and I say that since you seem to be algo proof compared to us poor n00b souls), they are a lot of long tail key terms. The 'good sites' get the plum keywords, others are stuck with almost 100% long tail.

Say I have /nike-shoes and someone searches for "Nike shoes size 13 black color." I have Nike shoes but not size 13. Bounce. Maybe because we're still learning, but we haven't figured it out how to include all that in the title or description. You know, "don't click here if you want size 13 because I don't have it." If I do, then I have to leave something else out.

[edited by: walkman at 4:00 pm (utc) on Oct 5, 2011]

[Hissingsid]

Bing does is that way, but Panda is a sitewide penalty when certain criteria is reached, so the entire site suffers. I know it first hand. Maybe if sections are divided clearly, it's different. maybe.

Could that be why Matt Cutts suggested sub domains as a solution.

Could this be in part about the site as a semantic web?

If this is correct Panda may be half the site as a semantic web and comparing user experience data between similar (semantically) sites.

[chrism]

I don't understand the mismatched traffic thing. I mean, you have a title and meta description that adequately describes the URL, right? The traffic isn't being sent directly; the user has to click on something. And granted Google will rewrite elements, which is annoying, but I haven't seen a rewrite that's THAT off the mark. So how is it Google is suddenly sending mismatched traffic?

Sadly, some people will always click on result #1, even if it's clearly off topic. We currently have a number 1 ranking for a way off topic term and people are clicking it. There's also a few related but not targeted rankings we have since this last Panda run, and they're bouncing away too.

[walkman]

Could that be why Matt Cutts suggested sub domains as a solution.

When you have thousands and thousands of writers, when dividing them, simply by chance some will make it. Hubpages is no way near their pre-Panda levels.

The Panda Shamanism got a mortal blow when Daniweb and others that escaped in July were hit again. Al their work, apparently didn't do jack.

[edited by: walkman at 4:09 pm (utc) on Oct 5, 2011]

[netmeg]

Say I have /nike-shoes and someone searches for "Nike shoes size 13 black color." I have Nike shoes but not size 13. Bounce. Maybe because we're still learning, but we haven't figured it out how to include all that in the title or description. You know, "don't click here if you want size 13 because I don't have it." If I do, then I have to leave something else out.

And that's something you expect *Google* to be able to figure out?

[walkman]

And that's something you expect *Google* to be able to figure out?

I expect all search engines that ranked me for that word to 'solve it,' especially when they are a monopoly, or near it. If I don't have have what's in the title and description, then they can /should penalize me. How is it my fault that I mention the word 'size' and '13' in the page or said "sorry but we're out of size 13 shoes." and Google puts them together?

Like I said, it's hard for some to understand, but for us who haven't figured out everything yet, this is a major deal. Google holds 70% of the search market and over 90+% of people use search engines to find things.

[jinxed]

No they should not be expected to know that. But, they also shouldn't penalise an entire site because they have sent people for certain key phrases that provided a negative experience.

[walkman]

No they should not be expected to know that. But, they also shouldn't penalise an entire site because they have sent people for certain key phrases that provided a negative experience.

Ding, ding, ding! We have a winner, but that ruins Google's brand-promotion plans since brands will be outranked quite often for many keywords. This non-wholesale penalizing will result in a better experience for users but the money is in cleaning the 'cesspool.' Google engineers can't be super-smart and not know this, so which one is it? Are they dumb or... ?

[netmeg]

Sadly, some people will always click on result #1, even if it's clearly off topic. We currently have a number 1 ranking for a way off topic term and people are clicking it.

I get that, but I wouldn't just lay that off to Google. For years I've gotten a ton of UK traffic at certain times of the year because we have a lot of cities with the same name as prominent UK cities (Birmingham, Manchester, etc) People over there type in the same search strings (cityname + keyword) and my sites come up first, and UK people, apparently missing state name all over everything including the domain, click on the result. But that's a human thing, not a Google thing, and I don't think it's necessarily a Panda thing either; I've seen it since at least 2003.

If "mismatched traffic" is your problem, you could look for ways to make it an opportunity. Eventually, I put up some pages with some UK information for my UK traffic, plus a few outbound links to get them closer to what they want. So at best, they'll click on an ad on that page, or at worst, they'll follow MY links out, and not use the back button to search in Google again.

[dataguy]

The Panda Shamanism got a mortal blow when Daniweb and others that escaped in July were hit again. Al their work, apparently didn't do jack.

Um, Daniweb has recovered, again: [webpronews.com...]

[netmeg]

No they should not be expected to know that. But, they also shouldn't penalise an entire site because they have sent people for certain key phrases that provided a negative experience.

True, but we don't know for sure that they're doing that; it can only be speculation. And in fact, in most of the Panda cases that have been shown to me, it's not the entire site that's been brought down.

It would be interesting to know if, on URLs that have been severely affected by Panda, they are suddenly ranking for long tail, not-well-matched keywords (that they WEREN'T ranking for pre-Panda). I'd expect that information could probably be found in Analytics/GWT.

[Zivush]

Google's brand-promotion plans

If Google starts plan B (BRAND PROMOTION), we as web masters must work differently.
One has to completely diversify the traffic in order to get the recognition deserved..Otherwise our sites get to nowhere.

[Hissingsid]

Brand promotion is just a side effect of the Panda medicine. Brand websites have something that Panda wants but Panda doesn't want brands so go figure what it is that brand websites have that makes Panda boost their rankings and do that.

[alika]

Um, Daniweb has recovered, again

Seems like my main site is following Daniweb's patterns.

We recovered in July, and got creamed again September 27. Some of our keywords lost as much as 70% traffic. Our GA was all red after September 27. Worse, our Adsense went down the drain as well as impressions went down and RPM went down.

Yesterday, everything seemed to recover. Even our Adsense RPM is almost at the pre Sept 27 levels. Comparing our Tuesday yesterday with Tuesday 2 weeks ago before we got hit again, we're only down 3%

Hope it sticks, but it seems like you never know ...

[proboscis]

My pandalized site appears to be following Daniweb's recent patterns in the opposite direction.

Small recovery on Sept 27 (first since Feb 24) then yesterday...it's all gone. Worse than ever.

[dunivan]

I think this panda update hit a lot of false positives again, and is the reason why Matt gave the weather report he gave yesterday - to be prepared for changes.

[walkman]

Seems like my main site is following Daniweb's patterns.

We recovered in July, and got creamed again September 27. Some of our keywords lost as much as 70% traffic. Our GA was all red after September 27. Worse, our Adsense went down the drain as well as impressions went down and RPM went down.

Google discovered some new 'data' as soon as Daniweb hit the twitterverse causing Google to look stupid and showing how it's near impossible to come out of Panda. Amazing, maybe the 'data' was under someone's desk and Google used their "scientific process" to change the SERPs. Your site will be fine as long as it fits the exception they made for Daniweb. You do realize why you came out in July, right?

My pandalized site appears to be following Daniweb's recent patterns in the opposite direction.

Small recovery on Sept 27 (first since Feb 24) then yesterday...it's all gone. Worse than ever.

I lost some ranking too after the changes they made for Dani, but it's a small price for me to pay so some can be manually expected. I couldn't the bad press Google got so I am happy it's solved now, even if means even less money for me and tens of thousands of not-famous sites.

I think this panda update hit a lot of false positives again, and is the reason why Matt gave the weather report he gave yesterday - to be prepared for changes.

They don't care about false positives, they care about famous false-positives that can give them bad press.

Brand promotion is just a side effect of the Panda medicine. Brand websites have something that Panda wants but Panda doesn't want brands so go figure what it is that brand websites have that makes Panda boost their rankings and do that.

Use adwords to advertise and see the difference in bounce rates between long tail and that. Bingo, I just gave away the secret!

[ACFinLA]

Our site has followed the same pattern as DaniWeb. To summarize, 50% drop in Febraury, 100%+ recovery in July, 43% drop on September 27th and now what appears to be a full recovery starting October 4th.

One thing I did do was submit a reconsideration request to Google last week, before I read anything about Panda hitting, because I thought we may have been penalized for a developer bug that caused a lot of UGC to be hidden from users but not from the engines. That bug was published roughly the same day Panda hit, so it's difficult to tell the true cause of the drop. However, seeing what happened with DaniWeb's recent recover, and what Alika said above, I'm assuming the developer bug had nothing to do with it.

[dunivan]

They don't care about false positives, they care about famous false-positives that can give them bad press.

I doubt they care about famous false-positive damage, hence the touch and go nature of the update itself, if they cared they would test it to death, and results would actually be better. One of my clients actually has an exact match domain, with decent PR, and thousands of good links being out ranked by an e-zine article.

edited: word omitted

[Lenny2]

I'm seeing the usual SPIKE in Google crawler activity that comes right before a panda update... anybody else seeing anything? Seems kind of odd since we just had a panda update last week...

[johnhh]

the usual SPIKE in Google crawler

normally kiss of death ... by a cuddly Panda or.. beware wolves in sheeps clothing

[dataguy]

I usually see the spike AFTER an update.

[walkman]

I'm seeing the usual SPIKE in Google crawler activity that comes right before a panda update... anybody else seeing anything? Seems kind of odd since we just had a panda update last week...

Google is probably going to reverse a lot of these these Panda things, their lies have been exposed now. Looks like not even Danny Sullivan bought them after 7 months:

Danny Sullivan has been pressuring Google behind the scenes to give us more on what is going on with Panda. I think, not certain, it lead to Google's Matt Cutts to tweet a "weather report" (Yahoo first named search updates weather reports back in the day), on Panda:

http://www.seroundtable.com/google-panda-25-tweaks-14127.html
I see the tweet as "don't talk about our lies or Daniweb, changes are coming, talk about that..."

[seoskunk]

As I said before I am the only guilty person in Panda so I can't moan about it.

What I have seen is some very thin pages rise to a half decent rank though. By thin I me a wordpress blog with only the default text. These sites are doing well off keyword in the domain. I do think when sites such as Daniweb get penalised its not the site with the problem its google. Maybe we should all just get pissed and party as these updates seem flawed at present and need some refinement. It seems a better idea than stressing over something that frankly is not working correctly.

[netmeg]

Um, I read that as Danny pushing for more transparency, nothing about exposed lies.

[Whitey]

I suggest more focus is needed on medium to long term site solutions during this period of instability and distraction.

Any continuing inputs on core fixes from observations and first hand experiences out there in preparation for the settling down period that Matt Cutts is predicting?

btw - I support the view that much more more transparency for webmasters is required. Thanks for raising it netmeg.

[edgeman]

Our site has followed the same pattern as DaniWeb. To summarize, 50% drop in Febraury, 100%+ recovery in July, 43% drop on September 27th and now what appears to be a full recovery starting October 4th.

I posted earlier in this thread indicating the same pattern: Hit in late February, 100%+ recovery in July, 45% drop on September 27th and 110% recovery on October 4th (actually, recovery began around 6pm PST on October 3rd).

One thing I did do was submit a reconsideration request to Google last week, before I read anything about Panda hitting, because I thought we may have been penalized for a developer bug that caused a lot of UGC to be hidden from users but not from the engines.

Interestingly enough, we have some similarities here. We had a developer bug go into production that Google picked up on the weekend before September 27th. They sent us an automated notification in Google Webmaster Tools telling us they had discovered an extremely high number of pages. GWT reports they crawled greater than 1.3 million pages over the weekend as a result of the bug. Our average is 280k. There's no way we wanted these pages indexed as they serve a purpose on our site, but they are, by definition, "shallow" content we still intend to improve in the near future before allowing them to be indexed. We fixed the problem immediately, submitted a reconsideration request fearing a penalty, submitted a directory removal request through GWT to remove the content (completed within 24 hours) ... and then we waited.

However, seeing what happened with DaniWeb's recent recover, and what Alika said above, I'm assuming the developer bug had nothing to do with it.

I'm considering the same. But I don't want to be quick to ignore what happened in our case. I would not be surprised whatsoever to hear that a site gets nailed by Panda after adding 1 million pages of less than ideal content.

[Whitey]

I would not be surprised whatsoever to hear that a site gets nailed by Panda after adding 1 million pages of less than ideal content

.... almost a certainty. Where do you get so many pages of unique and compelling content from?

Managing large organic sites is likely impossible for sustainability , unless you scale them from solid foundations with good techniques and solid content differentiators. Same subject , different words is not good enough. Google wants a different and compelling user experience to compete around what it can do itself or has plenty of.

So if you're investing heavily in rebuilding content with different words, think again. There's much more to the strategic direction of Google than shifting the " deckchairs on the Titanic".

The mentality for copying or building similar sites to compete with what the next guy/gal/business is doing needs to be radically adjusted to embrace these changes.

Support your niche with things that Google and competitors cannot do as well as you. Combine information, usability and relationships in a captive combination.

Again, more talk around this type of thing and sharing will help elevate folks to new levels of creativity and solutions.

Lurkers ... come in from the cold and be bold :)

[edited by: Whitey at 11:30 pm (utc) on Oct 5, 2011]