| This 49 message thread spans 2 pages: 49 (  2 ) > > || |
|Panda Recovery - A 1 Year Case Study|
We normally don't post links to blog articles unless they are quite authoritative, usually including information from a search engine employee. I'm making an exception here (there's not Google rep involved) for several reasons:
1. This article was printed on Search Engine Journal - a well-edited and professional source
2. The content of this article was also presented at the SMX Advanced conference
3. The detail offered here, and especially the information about the judgment calls involved, makes for some excellent reading and food for thought.
Yes, this is a long article. It's worth the time:
|The results that have occurred over the course of the past year have been truly stunning to say the least. I knew the site would see big improvements in organic data. I just had no idea it would be on this scale... the client made all the changes I recommended, and, as a result, a year later they've increased organic traffic by 334% |
Here's one observation that knocked my socks off:
|The data you see above is from 2010. THAT major drop was from Google's MayDay update. As you can see, daily organic visits plummeted as a result of the MayDay update. That's something I see in about 40% of the Panda audits I've done. |
What is that correlation telling us?
|Martin Ice Web|
thank you for providing this very good article.
When i summarize this excellent post:
-his efforts for recovery are mainly going through "Deep Content"
-don´t mix different Content, try to make "Content Sections"
-do not crosslink this sections
-Unfortunately, he didn´t say anything about coding, nofollow, noindex ....
-if you face duplicate content, make refine sections with refine content
You think Mayday and Panda are the same algorithm, only tightend?
Thanks for sharing. The real question for me here is: has this site recovered due to the fact that the many on-site changes were positively rated by Google's algorithms (internally), or did the on-site changes simply have a positive influence on usage data (externally), and the site recovered as a result of that?
If it's mostly real-world usage data, it seems to me that this would be hard to pull off on a low-traffic site (at least it would take much longer). They had that advantage.
My site also got hit in april, like hes, so it was some interesting read, but i dont think it helps me that much. When I look at external data rates looks a little better then hes new version, content, well its a image site so i can not compare. Lets see what happens on the next panda, be cause I would say im done now with changes.
Thanks for sharing this Tedster.
|Rewrote page titles |
Modified URLs and applied 301s to old version
Any idea on why these two were done? What necessitated the title changes and were the title changes done for many pages? What will be its percentage to the total no.of pages on the site? Was this done en masse? What kind of title changes is he talking about?
The next one is even more interesting. Any reason to modify URLs and apply 301s? Was this also done to a number of pages? What will be its percentage to the total no.of pages on the site?
|Moved 30% of sub-category pages into more refined categories? |
I am not understanding why this was done. Was it because the sub-category and the main category weren't relevant to each other or loosely coupled?
|Eliminated “node” page indexing |
What does he mean by a "node" page? Why were they de-indexed? How was this done? i.e. Was it through "noindex, nofollow" or "noindex, follow" or through some other method?
|Inbound Link Efforts |
Called for establishing a much larger base of inbound links
Called for greater diversity of link anchors (brand, keyword and generic
Called for more links pointing to point to deeper content
I am not even sure how this could be achieved all of a sudden without any control on the incoming links. :) But I am sure that this has helped. :)
Lastly, he isn't talking about the use of user metrics at all in the whole process!
It would have be great and very useful if the author had written about the above in detail. Is he a webmasterworld member? If yes, I would request him to throw some light on the above queries here. But Tedster, do you have any ideas?
no, he's no longer a WebmasterWorld member but you can usually find him on twitter.
Since he's a professional, I wouldn't expect him to give you too many details, as 1) it's what he does for a living and gets paid for 2) some amount of client confidentiality.
A major focus of all these changes was tightening up the number of links (300 per page was outrageous!) and thereby helping Google understand the sites themes much better. All that improves organic traffic. That explains a lot of the above questions about reasons for various actions. Note that the whole of this report was NOT second-guessing Panda itself, but rather aligning the site to best practices, period.
He did say he went right for the analytics reports - I'm sure user behaviors were in his mind st least as he made his evaluations. They just didn't make this article. If I were working on this site and trying to loose 300 links from the page templates, I KNOW I'd want to look at user data - wouldn't you?
With regard to the Mayday/Panda correlation, The timing of Mayday is suggestive. It was released after the Panda project was under development, so there's every chance it might have been "Panda Junior" - since the huge data mining challenge project that Baswindath Panda finally solved was still an obstacle.
For anyone who wants to research further, here's our Mayday thread: Mayday Algorithm Update - with video from Matt Cutts [webmasterworld.com].
|Refined keyword topical focus across top 100 pages |
Rewrote page titles
Modified URLs and applied 301s to old version
Rewrote content across all 100 pages to reflect refined focus
Reorganized content by category type
Moved 30% of sub-category pages into more refined categories
Eliminated “node” page indexing
Some major fixings right there... I wonder if Google uses an algorithm to detect if a page has changed "significantly" enough to merit serp movement. Webmasters are constantly making small changes, I wonder what the noise to action level is(assuming Google has a detect change meter. It would make sense, if for nothing else but to reset the user metrics data being gathered on the "new" version of a page.
Read the entire article. One of the graphs suggests that the recovery started happening as soon as they started making the changes. Other threads on webmasterworld suggest that google needs time to reindex the changes, collect that data over a period of time, churn it up with it's super secret recepie and then finally let the new rankings out. Any clues?
The graph I'm looking at (Organic Search Traffic Jan 1 to Jun 30) shows an improvement a few weeks after the fix started. Since there was a Panda update on June 16, 2011, that would line up pretty well, I think.
Of course, the improvements did a lot more that just address Panda - and I think that's a good approach anyway. Since we don't know exactly how Panda works, we should address the entire site and not think of Panda like a traditional "penalty" that we're trying to address. It is a new algorithm addition, after all, and that's not the same as a penalty, even though it can mean lowered rankings.
Google even stopped using the word "penalty", even for "manual actions" - did you notice?
I don't know if all the traffic increase on that site was due to fixing Panda issues or just general SEO and UX improvements as it looked pretty bad to begin with. When a site is already SEO'd and UX'd and then it's hit by Panda it's harder to figure what caused it. I'd like to see a case study on that type of scenario.
|Google even stopped using the word "penalty", even for "manual actions" - did you notice? |
This is an interesting point. I just had a site that got its manual "action" removed on it.
@Tedster - so if this was Panda Junior , do we have any strong indication of the cause? I mean you've speculated on the suggestion, but I'm digging deeper into your thoughts. By isolation of this does it reveal anything very specific.
Sorry for being repetitive, I just think there's something to be discovered here.
What this report cries out for from leadIng SEO's is more top quality reports in order to qualify and prioritize remedial actions. The thing that stands out to me is that this was not a quick fix. And it's not made easier by the delay and magnitude of lack of data.
Hats off to those that are sharing, and encouragement to those with knowledge that could be generous enough to circulate further case studies. Since this is not a one size fits all challenge.
Good stuff Tedster, thanks for openIng this post.
I watched both the video from Matt Cutts and reviewed our member comments. I think the "Panda junior" connection is pretty clear. In addition to the timing, Matt talks about a site quality algorithm (so was Panda), and members here also were talking about a link component and thin content component (but not yet shallow content).
That alone makes it a kind of preliminary shot at what Panda took a big shot at. I'd be very interested in hearing from other Panda sufferers if they also saw a Mayday hit that only took their long tail terms. Panda then took both the long tail and the lead terms.
The case study specifically mentions the correlation between Panda 2.0 and Mayday. I was originally hit by Panda 1.0 and NOT Mayday. I don't know what that indicates, if anything.
I tend to listen to what Google reps say the intention was - and that's where I see the connection. It doesn't surprise me that the version of Panda that correlates to Mayday could vary - the technical set-up was definitely tweaked between the two versions of Panda, but they are all targeting a similar end result, through a machine learning process.
This is a classic example of why I don't tune out what Google reps say. I realize there is a particular spin to public statements (only natural when they are talking all levels of skill and comprehension) but under the spin I always find value. Don't tune them out as "all lies" just becuase you've been hurt. I've been hurt by Google changes. over and over for more than a decade - but I do love the free traffic!
|I'd be very interested in hearing from other Panda sufferers if they also saw a Mayday hit that only took their long tail terms. Panda then took both the long tail and the lead terms. |
For sites I know, May Day update increased the traffic significantly until Panda 1.0 attacked them. So I don't see a connection between Panda 1.0 and May Day. I do not know about May day and Panda 2.0 connection. But what you said in the last statement holds good for my Panda 1.0 hit cases as well. These sites suffered from traffic loss on long tail (Panda 1.0 almost wiped out the traffic for many long tail keywords) and a significant ranking loss for lead terms.
I managed to find the site he is talking about and yes, it recovered from Panda.
These are the first things you saw on the site as a webmaster.
The site is a ad heavy website, even after he reduced # of ads.
About 7 ads + in-text context ads of infolinks (too many - more than 8 in-text links). I thought my sites are somewhat ad heavy. Well, they aren't as compared to this site.
Because of the ads, the site speed is not that good (20seconds or so), but he uses lazy loading by placing ad codes on the the footer, so the page content loads first and then after the ads.
Related content (user metrics?):
In some pages he puts 10 related posts on the right sidebar and 10 more on the footer of the articles. Plus, Read More links on the top of the page..
Plus popular keywords and popular destination (20 more links).
Some good job has done to make the visitor stay on the site. All of this is done using java scripts. Good!
I'd call it a link heavy website..
But it makes it easy to navigate. The header toolbar is inviting and very interactive. Many category links on the footer. Also, categories are listed on the left sidebar in some sections of the site, .
To my surprise the tags haven't got no-index or no-follow atr.
I assume he lowered the # of categories, moved articles to the right ones and deleted the sub categories.
I didn't like that the font Pixel/color are so small and grey. But this is subjective.
I wasn't impressed of the site, firstly because I am not interested in the topic of the site so I haven't red the content, but also because I have seen (and have) better designed websites.
A classical pharma spam site.
I am sure they were making a killing on it, because pre-panda was clearly designed to get a user to a content page and then click on the ad.
Nothing but MFA(dvertising).
I'm sure content is shallow too.
Clearly a major redesign was due, and judging by the niche and traffic owners could easily afford a top SEO and a full redesign.
Actually, after looking at the site,
I find that it appears to be an outstanding site, managed and edited by people who appear to be properly qualified to write about what they put out
On the face of it, top site.
scooterdude, I didn't look in the details because I didn't find a site, so possble. But looking at the front page:
"Seasonal passes on sale" - some theme park ads
"Fertility, pregnancy etc.", medical news including allergies - top middle of the page , lots of links, categories and articles medical-related (i.e. where my "pharma MFA" came from)
"$1000 please take a survey" huge ad - is this a survey site?
"Black friday forever" huge ad block - laptops, TVs , printers etc. is this technology site?
If I got to the main page, I'd be confused what this site is about. So they clearly didn't care about people getting on the main page , registering etc. they drove people to content pages, they obviously did this so content pages ranked in Google. And blasted lots of ads around the content, I am sure for charity.
What this guy doesn't say is that after redesign you can clearly say what this site is about - pregnancy. Before that from the main page it looked like non-themed MFA.
Perhaps we are all familiar with ad agencies supplying missmatched adds
Plus, that was before the redesign
I presume you forgot the <sarc> tags..scooterdude..
re the SEO work done..IMO all obvious, nothing insightful or astute or inspired..given that he started with a "sows ear"..almost anything would have been an improvement ..but after some basic work ( and doubtless a heavy fee ) ..it is still a "sows ear" just a slightly less garish one..
I agree with aleksl2, IMO it is the kind of site that proves that panda needs to be tightened and rerun many times, as apparently they are still missing an awful lot of the crap..and a bit of window dressing and a less loud shade of lipstick on the pigs, is letting them back in, in some niches at least...
I've seen more actual "content" compared to ads, on big box store summer promotion flyers and shopping channels..
My traffic was normal on MayDay 2010. I wasn't hit till Panda 2.3 July 2011 when I lost about 1/3 traffic (I suspect this was partly due to a stupid experiment on my part and also others copying my content). Then I lost more traffic with later Pandas. I cleaned up the experiment problem but not the copied content which I'm still working on. I've also found thousands of site wide links put up by competitors which I'm getting removed. I hope google comes up with that tool to disallow bad links soon!
I have found the site, and the first page I landed straight on from Google search tried to open a pop-up ad on me (thanks to FF pop-up blocker for saving that from me).
The post is a nice read, but seriously, that's exactly the kind of site that should be hit by Panda, and if there is anything weird and unjust here, it's the quick recovery that this site simply does not deserve.
|I find that it appears to be an outstanding site, managed and edited by people who appear to be properly qualified to write about what they put out |
scooterdude, what are you talking about? So people educating the public on amenorrhea and gynecologic cancer do not need to hold an M.D degree anymore? Or may be they are doctors after all, just too bad the authors' names do not appear under the articles for us to see.
The fact that the articles are followed by multiple "references",( none of which are followed or linked to ) would lead one to presume that the site's "content" model is article and scientific paper scraping, and re-spinning via human writers..pure MFA ..a la ehow..the site owner is a doctor..with a huge MFA med site empire, spanning two continents..the about page / bio is "enlightening"..much more so than the "site content"..
[edited by: Leosghost at 3:42 pm (utc) on Jul 5, 2012]
I do wonder whether we are all looking at the same site, otherwise i'd wonder at something else
It's true the site was awful to begin with, and most of his improvements were just "best practice", and the site is still arguably a content farm.
But there are two things that IMO must not be missed here:
--The possibility of a Mayday/Panda connection (which some posters here are showing may not be as straightforward as the author thought, but I still think it's very interesting)
--This site, like Ehow, was a content farm, got hit by Panda, made improvements but remained a content farm, and is now okay with Panda.
I can't really add anything about Mayday/Panda, since I appear to have been hit by neither of them on any of my sites.
But as for the "content farm that beat Panda", Panda is clearly looking at a complicated set of metrics, and it can't always get it right, or even come as close as a group of humans who have slightly different ideas about what's thin and what's valuable. Is it that Panda overlooks your content farminess if they get enough indications people like what they're reading on your site? Or is it just that if your content farm makes big improvements like this one did (and I think ehow deleted some dupe articles and changed their layout), Panda sort of assumes you're good to go now (but in that case I'd expect it to hit these sites later, as it tightens the noose).
(What I REALLY still want to know is how Ehow found out just what they needed to fix in order to get back in with Panda so quickly. Did they just have nearly psychic SEOs, or did they know somebody who knew somebody?)
| This 49 message thread spans 2 pages: 49 (  2 ) > > |