Panda fix - is this a time penalty / trust thing ?

[Whitey]

I may have missed something but I haven't read one report of a sustained return from the "Panda Slap" . Presumably, folks have been busy writing new content , taking down "thin" pages , boosting links ... etc etc ...

... and Google doesn't comment on penalties.

I think this is more than a slap. Is there a time / trust element in the release from the Panda's grip?

[Whitey]

Walkman , Now that's positive. Was it more or less instantaneous ?

It would be good if there were 100's of similar reports that held. ie not to return and then only to drop again as some reports suggest.

[edited by: Whitey at 11:23 pm (utc) on May 24, 2011]

[walkman]

Brinked, I don't have 'articles,' I have descriptions and very well written ones. Not $1 each, that's for sure and I still improve them during the day as I see them.

And plenty of sites like mine that are doing well do. And I don't think there's a lose 80% traffic key, it depends on the seriousness of the violation. I have removed search pages /tags and that alone should have given me a boost if Panda didn't like them.

Farming content alone is nonsense, sites that sell shoes with 3 lines of description or very good sites with too many ads have been pandalized as well.


Whitey: right AFTER panda, not before Panda. Huge difference ;)

[Shatner]

@walkman It seems like I have returned to my "right after Panda 1.0" levels too.

For anyone who is confused, what that means is this....

After Panda 1.0 I took an instant 50% hit in my Google traffic.

After Panda 2.0 I took another 15% hit in my Google traffic bringing me to 65% total.

After Panda 2.1 I took another 5% hit in my Google traffic bringing me to 70% total.

Now I have suddenly gone back to my post Panda 1.0 levels, which is to say my Google traffic is now down 50%.

Any idea what this means? Has Google rolled back Panda 2.0 & 2.1 and kept 1.0?

[Whitey]

@shatner - I think what I was questioning was whether a full returns are being reported. There may be others I've missed , but Walkman's is the first I've seen or heard about directly. Tedster's is the other.

Certainly if there were 100's of reports we wouldn't be talking.

Have you been furiously rewriting content to the same extent as Walkman?

[walkman]

@Whitey
I wish. I said I rank for the same pages like /page1.html was doing well after panda and is still today. I didn't count if all do the same or not, but I recognize when I see the referrals. I have deleted many pages so my post-Panda stats are a mess. I cannot tell you how much traffic I lost from the pages I deleted, I deleted them on a gut feeling.

I am NOT back.

[Whitey]

@Walkman - thanks for the clarification

So we're line ball again, back to Matt's remarks and waiting for reports Y/N ? ..... i got a sniff of excitement.

$1M's of new collective content writing investment from Pandalized site owners out there, in a log jam living and breathing for a break. Patiently waiting , waiting and waiting ....

[brinked]

panda 2 is not entirely panda. I can tell you that whatever google rolled out on the day of "panda 2" is still out there and has not been reversed, maybe improved...but not reversed.

[Broadway]

I benefited from Panda 1.0. I was hit extremely hard by Panda 2.0.

I was under the impression that eHow did ok with Panda 1.0 and got hit with Panda 2.0. So how am I similar to them?

WebmasterCentral says this about Panda 2.0:
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.

I was thinking that Panda 2.0 might have something to do with evaluating a site's deep pages, in the sense of cross referencing same-site pages to see if too many of them were too similar (ie. content farm).

The page isn't just evaluated to see how it ranks for the search terms, such as "drink pepsie."

1) but also how other pages from the same site rank for these same search terms. (How to drink Pepsie, How to drink Pepsie light.)

2 Or how similar the content of other pages is, even though related to a different search terms. (How to drink Pepsie, How to drink tea.)

If there is too much overlap between pages on the site (too similar in nature), then Panda 2.0 snags it.

I can see how my pages, comparing multiple brands/models of products would get snagged by this.

To be a value to the reader, you have to have similar headers and discuss similar keywords. That's the idea of directly comparing things.

But to Panda 2.0 it might just look like a beefed up version of cookie-cutter template pages just made to get long tail traffic.

[DirigoDev]

Lost ~60% on Panda 2. First move was last Monday 5/16. Moved up 15% after many changes. Mainly, a breakup of large pages, reorganization of large silos, unhyperlinking reference page external links, new server and 50 hours tuning performance (CDN, sprites, minification, all new JS and CSS), and tons of removing internal links. Also,our blog seeder is now throwing away ~20% of their posts with links to "click here" or other stupid anchor text to look more organic. I have no clue why we moved up in ranking. And this is exactly what G wants. This thing is a huge black box.

[Whitey]

First move was last Monday 5/16. Moved up 15% after many changes

That's encouraging - let's hope it holds and builds. But you took some radical steps and risks in the process. Did you add any new content?

[DirigoDev]

We add 2 new pages a week with up to 2 additional supporting pages. Main articles are ~2500 words per page. In addition, about 4 pages a week get reworked. Four 4 time writers. This is the same pace as we've had since mid-2006. Yes, a little new content. Nothing special. The ranking move was unmistakable across the board. We continue to make big changes.

I'm now starting to see Panda 2 as a good think. It was a swift kick in the ass. The pace of change is huge and we're really focused on making our site(s) better for the users. Sales are still down > 20%.

[Whitey]

@DirigoDev It's hard to imagine Panda was your issue with that amount of unique content being regularily written if you have a relatively small site - say under 1000 pages.

Are you sure it wasn't something else and the dates just corresponded? Unless some of the things you mention tie into it in some way.

[balibones]

I'm new to the forums (although I had a WebmasterWorld account from circa 2006 that seems to be gone?) and hate to break into such a big discussion, but I took Netmeg's advice and registered / bought an account so I can join and hopefully learn some things. Unfortunately it sounds like most of us are in the same boat.

For what it's worth, here's what I think is going on...

New factors are being computed that have the power to override the factors we've all been used to for years. Old = Links, on-page factors, technical issues... New = User-feedback signals of "quality" like back-clicks, bounces, blocking sites (for negatives) or bookmarks, RSS subscribes, UGC participation (for positives). Those are just guesses, but the point is that new things are being taken into account in order to assess "quality" whatever the F#$K that means.

Between each "iteration" of Panda (as Matt called it today in his live-chat where he did his best to avoid all Panda-related questions) enough data has to be collected about these signals before another iteration can be ran (that would affect your site). This has to do with statistics 101 - you need to reach a certain level of data before it has any statistical significance or the margin of error is so wide that the "conclusion" is pretty much invalidated. It would be like saying 54% of all Twitter Users are female (margin of error 5%), in which case "most" Twitter uses could be Male, or "most" could be female. You won't know until you collect enough data to make the margin of error smaller by growing the sample size.

Let me explain why I think this and then explain what the ramifications "could be":

It would explain why low traffic sites weren't hit (as badly?), including low traffic scraper sites - but they'll get hit next time since they took your high traffic ;-) - and why low traffic keywords aren't as affected (as in kd454's comment earlier). There just wasn't enough data/visitors to determine if a site/page was "quality" or not.

It would explain Matt's comment to Tom Critchlow too. He basically said they need to wait until they get the data and then run Panda again. This statistics 101 stuff seems like the exact boring kind of crap NPR would leave out of an interview. Then again, NPR is sort of boring anyway so...

And, above all, it would explain why we don't have any changes in rankings yet after working on a ton of quality issues that Google has mentioned, from removing anything remotely "thin" to putting Adsense further down, or removing Adsense, to putting better/more content higher on our pages, to speeding up the site, to reducing bounce rate, increasing time spent on site, increasing pages per visit, etcetera, etcetera...

What This Means if it is correct. First of all, I think it's "somewhat" correct or I wouldn't be saying it out in public. But nothing any of us can deduce from outside the GooglePlex is going to be 100% correct. But assuming there are some kernels of truth and common sense in this, it means:

#1 There is a catch 22 at work. Without your old traffic levels how can you get enough traffic to build enough data to make a statistically significant statement to the algorithm that you are now a "quality" website? Some answers include: Build more brand traffic (didn't Google say to do this?) and find other channels (I'm with whoever said earlier "F*&k 'em, I'll get my traffic from elsewhere" if they keep it up.) from which to get your traffic, including mobile apps, social media, advertising, and little bitty Bing.

#2 It isn't just about making your site higher quality, as much as making your visitors TELL Google in some way (via some user feedback signal as discussed earlier and listed on my blog post) that your site is high quality. Get them to subscribe and click through newsletters; get them to bookmark the page; Plus ones, Tweets, FB Mentions, posting UGC on the site (which is the only thing that I've found that gets our rankings back temporarily so far), spending more time on your site, visiting more pages, etc...

#3 With a big enough virtual army in enough locations (Mechanical Turk?) you can seriously screw your competitors.

#4 God, I hope this one is true... The next time they run Panda with a fresh crawl of your site and all of the user-feedback data they need, assuming you've improved your "quality" (however that is determined, whether by the metrics I mentioned or others) we should start to see some sites get some of their rankings back and some scrapers get killed off again.

But I guess if that doesn't happen we can all go get jobs at Starbucks or become social media consultants. J/K SMM folks, J/K.

[walkman]

[youtube.com...]

Thanks Balibones,
this explains a lot of things and turns out everyone was right. Panda has been run, content changes have not been considered for some so Panda hasn't re-run for them. The worst part is that Google now looks much more for 'signals' so unless you're discovered or seen around you are in Panda Land. Something not seen or nto seen as much might not be 'quality' and there's a thread for that in the front page.

There's another major Paradox: give all the info in one page or? If page-views /user or time on site is a major factor there's an interesting incentive to make people go in just enough circles not to annoy them too much but circles, none the less.

And, above all, it would explain why we don't have any changes in rankings yet after working on a ton of quality issues that Google has mentioned, from removing anything remotely "thin" to putting Adsense further down, or removing Adsense, to putting better/more content higher on our pages, to speeding up the site, to reducing bounce rate, increasing time spent on site, increasing pages per visit, etcetera, etcetera...

Edit: I am starting to think that maybe Google is measuring these INDIRECTLY. Fix them all you want but that may not matter unless they get the 'singals' from users. This could also explain why sites with 200,000 really thin pages gained on Panda 2.0 and 2.1, social media in certain sections. So Google is saying (IMO) do these things because we think you *could* get better user signals and rank higher not becuase we're counting your word count or whatever.

And delete pages and after Panda and present enough users with a 404 and you will go down a lot more

[balibones]

I am starting to think that maybe Google is measuring these INDIRECTLY. Fix them all you want but that may not matter unless they get the 'singals' from users.

Waklman, my point exactly. Glad someone else is on the same page as me. I was beginning to think I was crazy. ;-)

Measuring "thin" content directly was easy for Google. It's called Word Count. Sure, that gets more difficult with non-text content, but still. They've measured it directly for quite some time.

Measuring it indirectly via user feedback signals is TOTALLY different because it requires lots of users to perform an action that sends Google a message. Unlike Direct Measurements, they can't pull this one out of their database whenever they want. So we wait. And wait. But I think the more traffic we can get to our sites from other channels, and from other searches (like more Brand) the better off we'll be when the time comes.... assuming we've fixed our "quality" issues.

[whatson]

And delete pages and after Panda and present enough users with a 404 and you will go down a lot more

Are you saying if you delete the pages of thin content, then your rankings will suffer in further?

So I guess the big question/s is what algorithmic measurements can determine good quality content.

[johnhh]

The problem is how is Google collecting this data - how can they know my newsletter signups - how can they know how many phone calls I get for further information ?

Years ago when I did advanced statistics at university someone proved a direct correlation between the penguin population change and the number of people who voted Republican !

[Whitey]

Matt Cutts: [youtube.com...]

Brief summary:

Users need to send Google signals about your content, not sure how much signals and content each weigh but social could mean life or death for some sites. If you improve, you'll move up the next time. But you need enough users /visitors for that first.

Panda has been re-run.
[webmasterworld.com ]

.... so it's when they choose to manually flick the switch ( sorry i was unclear on previous communications on this )

But again what and when will folks spending millions of collective dollars and laying off staff be given a clearer fix on exactly what to focus on. The directives are too vague causing folks to guess wildly. This is critical to the time expectancies of every good person involved who is trying to respond and adapt.

I'm just not hearing success consistancy in the stories back from the webmaster / siteowner community to back up these directives integrity.

[DirigoDev]

It's hard to imagine Panda was your issue

I'm sure that Panda 2 is our issue. We've been SEO optimizing our site since 2003. We have 1,300 highly tuned pages. Too highly tuned. Sales funnels and links are all tuned for best performance. Sales funnels all lead to the same place. Too perfect. Not natural enough. This is why we lost 60% of our Google traffic on 4/11. I was in denial but now see the light.

We have too much content that overlaps. Since the beginning our model has been fairly simple. Get 3K new e-mail opt-ins a day (there is some magic to this – clean with a superb SenderScore and low FBL). Build a super quality newsletter every two weeks. Build the newsletter landing page as a content page. Make it evergreen by updating the page at least once per year keeping it current. Point PPC campaigns at the new page with no budgetary limits – just manage to an acceptable cost-per-order. Let the SEO team optimize the page. Use reputable in-house blogger (members of their blogging community) to write valuable content about our new article on a high quality blogs (e.g. nice inbound links). Make $100-$500 per week from each page by soft-selling products (very soft). And from 1999 to 2011 this worked exceptionally well.

The problems: because newsletters were the genesis of content and profit was the end game, we wound up building lots of content around super high traffic/highly sales converting keywords. Many of our pages have lots of overlap. Every year we launched a newsletter about XYZ. We did so because XYZ sold 500 widgets. Naturally, we wrote an entirely original piece (KW set, draft, research, authorship…). Do this five years in a row and you get 5 content pages about XYZ with tons of overlap. Each of these articles was linked in the left navigation. We did our best to make the articles from a different point of view/different slant… Everything got internally linked by the SEO team to exploit the content. The pages are super well written and authored by people with authority. Our visitors love the content. Some of the pages have 4,000 FB likes. But at the end of the day, the entire approach does sort of look like a farm. Look at the KW densities on these pages and they look very similar. We figured out a system for growing $$.

We’re no Demand Media because our users adore the content. And I still don’t believe that we were Google's Panda target. But we wound up farming (with a small f). Let's be honest about things. We all have business models and we all do our best to make money from our craft. We're all farmers and the lines are very blurry - event for us white hat folks.

Our new path is to separate the newsletter team from the content team. This will provide a more rational editorial calendar. The content team will fills holes in our content offering. The newsletter/conversion team will build highly converting newsletter content all of which will not be noindex-nofollow. Many of our pages are going to be unlinked from navigation and live in an archive accessible from a sitemap and our Google Mini. We still have lots of traffic and are hard at work fixing the farm-like issues.

I’ve dropped FB comments into several silos and we’ve picked up as many as 20 comments on some pages in just 4 days. We're even building content now for sites like the Huntington Post. If G wants social signals I’ll give them what they want.

The whole Panda thing has reinvigorated the marketing and I.T. groups. We were once innovators (99-05). We became farmers (06-11). And now we’re innovating again (11+). This is not going to be a good $ year. Notwithstanding, I’m excited about the future.

We've gotten back ~15% of our traffic and I'm confident we'll get a full return within a year.

[edited by: DirigoDev at 11:03 pm (utc) on May 25, 2011]

[balibones]

Whitey,

That's the thing. They can't GIVE you an exact set of things to focus on because it's up to the users. Google could tell you to remove all thin content and any Adsense above the first paragraph, but if there's something else on your site that bugs users (pop up ads, hard-to-read font, fuzzy images, you name it...) then those users will send feedback (how? What feedback? Again, you name it. Google has Gmail, YouTube, Toolbars, Analytics, access to Firefox data, and millions of people searching and surfing while logged in...) to Google that they don't like your site and think it is "low quality". So how can they tell you exactly what to fix when it's all up to the users?

Google did NOT come out and say that. And Walkman was clear in that other thread that those were his words, not Matt's (although what Matt said seems to support it). BUT... it is what makes sense to me right now. I have no other way to make sense of it all. I've tried and just can't see it. This has to be based on user-feedback signals from somewhere.

So yea, social and other traffic channels besides Google could be life or death for any Panda-smacked site. Good luck and may God help us all. ;-)

"Make the users happy" has always been the holy grail for Google. It just makes sense to me that this would be the huge project that the Search Quality team would have been working on.

The trick for us from now on (in addition to building links and all that we usually do) will be to find out what is making our users unhappy and to fix it. Thus, we have to give Google what they've always wanted. Happy users.

SEOs lost the battle but not the fight. I'm sure people will find a way to game it (Mechanical Turk armies?) but until then we have to get and maintain lots of happy users and hope - for crying out loud - that some of this stuff works... SOON.

[whatson]

There is definitely more to it than just users, I am sure sites are getting pandalized for duplicate/thing content also, that is somewhat algo related.

[walkman]

The trick for us from now on (in addition to building links and all that we usually do) will be to find out what is making our users unhappy and to fix it. Thus, we have to give Google what they've always wanted. Happy users.

SEOs lost the battle but not the fight. I'm sure people will find a way to game it (Mechanical Turk armies?) but until then we have to get and maintain lots of happy users and hope - for crying out loud - that some of this stuff works... SOON.

My friend you got it totally wrong, or partially wrong, not everyone tries to game Google to the extent of not having happy users. We all add a keyword more here and there but there's a difference. Until a month or two ago there was no way to tell Google +1 or -1 (by blocking) and that happened after Panda so it's all guesses on their part. So having a page that is straight to the point can make the users extremely happy but if Google says that users who spend more time there, even if it's to scroll to reach the info, are happy then you are pandalized. I am oversimplifying it but you get the message. That's the same debate as with bounce rates.

So I can't play that game, and especially not with my bread and butter site.

There is definitely more to it than just users, I am sure sites are getting pandalized for duplicate/thing content also, that is somewhat algo related.

No doubt but say you get a 10% demotion and other sites get a 10% promotion...you may be on page 3.

[Andem]

>> There is definitely more to it than just users, I am sure sites are getting pandalized for duplicate/thing content also, that is somewhat algo related.

I agree completely with you whatson. I read recently on other threads in this forum by senior WebmasterWorld members that sites with high bounce rates saw increased SERP performance compared to their other sites with low bounce rates which were hammered by Google. I've also experienced a net increase in G referrals to pages with extremely high bounce rates (85%+).

[whatson]

I have 4 sites in one particular "keyword industry", 2 of them had content that came from other sites, the other 2 didn't. Only 2 of these sites suffered from Panda 2. I bet you can guess which ones they were.

However, prior to Panda 2, I saw about a 35% increase in traffic to the sites that got hit from Panda 1. So maybe they are actually two different stages of the same algo.

[SEOPTI]

I believe the first person mentioning they use user metrics along with Panda was Rand Fishkin from seomoz and Rand has been at the Plex a few times. Look at his face when he says "Panda" ;)

[youtube.com...]

"All those signals are now being used in this Panda..." he knows what he is talking about.

[Kenneth2]

The top 2 bloggers of our country got pummeled by Panda 2.0. They run their blogs independently under own domain names. From the blog comments it appears that those who run their blogs on wordpress or especially blogspot see traffic boost after panda 2 regardless of content quality. So there is another quality metric small publishers/bloggers would have to overcome. The existing Domain/Brand Power! Shall we just move our sites to blogspot? LOL

[koan]

it appears that those who run their blogs on wordpress or especially blogspot see traffic boost after panda 2

I'm surprised to learn there are legitimate sites on blogspot.

[Kenneth2]

I'm surprised to learn there are legitimate sites on blogspot.

My wild speculation on G algo (tongue in cheek style).

If ($overallcontentscore > 80)
{ SEOpoints = 1000 }

If ($pagerank == 6)
{ SEOpoints = 10000 }

If ($Googleproperties == 'yes' )
{ SEOpoints = 100000 }

[tangor]

If ($seopoints > 0)
[Panda = 1000000000000)

[tranquilito]

I believe the first person mentioning they use user metrics along with Panda was Rand Fishkin from seomoz and Rand has been at the Plex a few times. Look at his face when he says "Panda" ;)

[searchengineland.com ]

_{"Cutts said the Panda update will eventually roll out internationally, but he’s not sure how soon it’ll happen. Right now, Panda has rolled out beyond the U.S., but still only impacts English-language queries worldwide. "} -

this has nothing to do with users metrics - user metrics are not linked with site/user language

_{“There were some characteristics that were more applicable to English-language sites,” Cutts said. The original question came from a viewer in Poland, and Cutts explained that “the link stricture of websites in Poland is a lot different” from the link structure of sites in other countries.} -

the question is: internal links structure or backlinks profile ?