Recovered from Panda - and how I did it

[brinked]

Alright so one of my sites just recovered from panda. Its a site I bought a few months ago and have mostly neglected but it earned a decent income (all adsense). Site had 3 adsense ads on every page, I removed them all, since the panda update it was only making about $5 a day anyway.

Site has about 80 pages, all with unique content, its an old site, really should have never been hit IMO. Here is what I did:

- removed all adsense ads from every page

- Updated the homepage text a little bit, added some and tweaked some old text.

- removed duplicate meta descriptions (there were 6 pages that had identical meta descriptions)

That is all. I am encouraged that it only took about a week for the panda to gets its paws off. I wish my other site was recovered but its much bigger and many more pages. I am not sure if it was one thing or a series of things...to be more clear, there was one adsense ad at the top posing as a top navigation and then two of the smaller banners 1 above content and one below it.

[DirigoDev]

assuming the Googlebot can read javascript

I can confirm that as of 4/11 Google can read JavaScript. Just after the 4/11 update we started to see our original publication dates show in the SERPs. Previously the SERPs saw our last modified dates.

Here's the kicker, the original publication dates are in a JavaScript document.write script. This means that Google is reading the scripts. I cannot pinpoint a date certain. The first time that we noticed the wrong dates in the SERPs was 4/11. I'm pretty sure it was rolled out with the Panda update.

Anyone?

[Leosghost]

Anyone?

?

You mean as in anyone confirm ? ..I thought I just did :)

The game changer will be when they get action script parsing sorted..there are signs on some serps that they are getting there..or they may be even further advanced than it seems along that road and are holding off using what they have until they think they have it accurate enough..combine with xml ..and the web gets much more "immersive" and less text driven.

Javascript was being accessed for a long time now ..just not being used in such an obvious manner in serp contsruction..

[crobb305]

I am seeing phrase-specific recovery. My rankings have been restored (with sitelinks) for some key phrases, but things as small as adding "for" or "with" in the search query completely changes the result.

Example:
big widgets blue houses (#1 with site links)
big widgets for blue houses (page 2)

Very odd. But, my traffic is up 40% to 50% from the previous 4 weeks. If it holds, I'll post more details. This is the biggest glimmer I have seen.

[edited by: crobb305 at 3:40 pm (utc) on Apr 26, 2011]

[enigma1]

I can confirm that as of 4/11 Google can read JavaScript

I don't think so. It may read some inline js via onclick onchange etc handlers, but newer DOM/js based I don't think it will ever do. I have quite a few jquery scripts exposed and not even once were accessed by any SE. They don't even know they're there.

[DirigoDev]

I don't think so.

Okay. Let me restate. They're reading and acting on simple inline js (e.g. a document.write statement).

[browsee]

Very odd. But, my traffic is up 40% to 50% from the previous 4 weeks. If it holds, I'll post more details. This is the biggest glimmer I have seen.

Did you reach pre Panda level?

[Shaddows]

search query completely changes the result.

Example:
big widgets blue houses (#1 with site links)
big widgets for blue houses (page 2)

Are you the only major change between the queries, or are the SERPs totally different?

Where's the #1 that you displace?
Whats in "your" spot on page 2 when you move?
What constitutes the top 3 in each case (as in, types of site and do they match)?

[freejung]

I have quite a few jquery scripts exposed and not even once were accessed by any SE. They don't even know they're there.

Google is definitely using my javascript to generate my site previews, though. I have a jquery fade animation that is usually displayed mid-fade, that is, with two pictures partially opaque in the same spot. That's not possible without reading the jquery animation script. With JS disabled, the pictures don't even display at that location, let alone partially opaque. So on some level they are reading JS, including jquery.

[enigma1]

That's not possible without reading the jquery animation script.

You can use js in many ways but unless you somehow expose directly the js links or js resources in the html part, don't expect the SEs be able to figure out how the code works or when it runs.

Also it may cause problems especially if they start doing ajax calls since they have no clue what the code does. What they parse is really basic and if you have the html clean of js using the DOM for jquery instead of inline they don't know its there.

PS: about the site previews is quite possible your browser simply run the js part.

[edited by: enigma1 at 4:36 pm (utc) on Apr 26, 2011]

[crobb305]

Did you reach pre Panda level?

Browsee, I am seeing about 45% recovery from each of the past 4 Tuesdays. Panda killed 60% of my traffic, so about 45% of that has been recovered today. My traffic is at its highest point in 60 days, but not quite to pre-Panda levels. It is coming back incrementally (phrase specific), it seems. I noticed this yesterday when I had a 10% recovery in traffic. I started to see some good rankings for certain phrases. That has increased dramatically today.

Are you the only major change between the queries, or are the SERPs totally different?

Shaddows, I just checked some phrases that I knew I wasn't ranking well for earlier (mentioned above). Initially, I didn't see my site listed; but upon refresh my site pop into the #1 spot. The other 10 moved down, but stayed in same relative position with each other. So it seems to be a re-ranking for my site, rather than a complete SERP shuffle.

[jinxed]

@crobb305

Good to hear!

[Bewenched]

I've been seeing alot of scrapers ranking higher than us for our own unique content. What I'm noticing is that they have a fresher/more recent publication date. Prior to January 1, 2011 we had not been serving E-tags, we implemented them because i thought it might help tell Google who the original publusher was.

I'm wondering if other sites that have been affected are using e-tags and last modified dates.

Are there any sites affected by the 2/24 Panda that are not using these?

[crobb305]

Bewenched,

Walkman has posted some examples of how Matt Cutts' content is being outranked by scrapers (for snippets), even some of his recent blog posts. It's definitely a frustrating issue right now.

[Bewenched]

I'm just thinking that there is a "freshness date" that is playing a part of it because the scraper sites I'm seeing are junk.... I'm talking polish, india and chinese scraper sites with bad jibberish english,... a few scraped titles from sites and adwords, however their modified dates and etags are newer than my content.

I'm going to run a few tests on some of our static pages that are suffering. Some I'm going to remove header date information and etags and the others I'm going to use pfrank to set the create date and modified date of the files to today and a few others to modifed date of today and leave the create date alone.

[freejung]

about the site previews is quite possible your browser simply run the js part.

Actually, I think you're right about that -- it's being run locally on my browser, not by Google itself. So the preview doesn't really indicate anything about the search algo's capabilities.

[TheMadScientist]

...etags are newer...

Just to be sure there's clarification on what an ETag is.

It's simply a 'unique identifier', which could be a time-stamp, date or hash or anything else 'unique', so an ETag, unlike date, is actually evaluated for 'different' rather than 'newer' or 'older' as a date could be.

ETags can be anything EG:
1
page_version: A
1sryaovart43253nfa

Anyway, just wanted to make sure people know an ETag is not necessarily 'date' oriented, even though date may be used as an ETag.

[crobb305]

Regarding my improvements that I posted about above, I am making another observation that is interesting...

One of my heavily Pandalized interior pages is now ranking on page 2 and my homepage is #1 with sitelinks for one of my big phrases. However, despite the fact that the phrase I am ranking well for is a big phrase, it isn't the target phrase. The target phrase has the words in reverse order. Last week, I added the target phrase to the title tag of the Pandalized interior page. Now the page is unpandalized, but ranking for the target phrase with words in reverse order. The improved relevance seems to be bleeding back to the homepage, thus pushing the homepage back to #1. I hope this makes sense. Let me give an example.

For Example:
I added "big blue widgets" to the title tag of Pandalized interior page (to improve relevance for both the target page and the homepage which links to it heavily).

Ranking:
1) "widgets big blue" -- homepage now ranks #1 (was page 2 last week), interior page ranks now on page 2 at #15 (was page 50 last week)

2) "big blue widgets" -- Still on page 4.

Definitely an improvement overall, especially since traffic is up 45% (which is the real measure of improvement) and given that the Pandalized page is no longer pandalized; but why it is ranking so well for the phrase in reverse order is odd to me. This is their phrase-based document scoring at work.

[Bewenched]

Thanks MadScientist for that clarification.

I'm especially interested in the header date since almost all of our dynamic pages do not have this set currently
..
example:
Last-Modified: Tue, 22 Sep 2009 20:20:09 GMT
and
If-Modified-Since: Tue, 22 Apr 2009 20:20:09 GMT

I've now set it dynamically to pull from the last time a product was modified. This should be interesting.

[TheMadScientist]

It will be interesting, and I think for 'doing things the right way' it's something good to do, but personally wouldn't be too concerned with it for 'origination' DIRECTLY, because it's too easy to spoof, so it's fairly unreliable as a source of 'who published first'.

I've personally run Last-Modified and ETags for dynamic pages for years, just for the bandwidth and crawl savings.

They could, imo, have an indirect effect: EG If you serve Last-Mods and proper 304s without delivering content when an unmodified page is requested it would seem you would cut down on your 'crawl budget' by cutting the bandwidth and processing necessary for unmodified pages, which could free up some crawl cycles for other more recently modified pages and THAT could translate into a new page being crawled and processed sooner, which COULD help with determining origination.

[enigma1]

I'm just thinking that there is a "freshness date" that is playing a part of it because the scraper sites I'm seeing are junk.

Yes that's pretty much my view too.

But I don't think the etag is going to play a role vs the content. I still believe what I mentioned in a couple of other threads that domains are way undervalued and so building a domain farm doesn't draw significant resources, amplifies the scrap problem a lot and at the same time makes things very complicated for SEs.

ICANN maybe able to change it for good if they wanted to by driving the domain value way higher than it is right now.

[c41lum]

Have your pages recovered worldwide or just US?

[kidder]

We run a handfull of adsense sites, we try to keep the quality decent and so far they are all fine and we run 3 blocks of ads right where they can be seen. I'm not buying into it. What I can tell you is that "networks" of adsense sites can be connected and dealt with harshly, think of the user experience. We are always looking at the user behavior and the page views, if they are sticking on the site and looking around it might just go in your favor even if your running adsense. Just a theory...

[Jane_Doe]

I don't know if freshness is necessarily important. I have a number sites that haven't been updated in years suddenly start to get extra traffic with Panda.

[kd454]

I removed Adsense all together on one of my sites and in 4 days it jumped from 3rd to 1st page #10, had 2 nasty Adsense blocks between title and content removed = improve user experience.

Again I don't think Adsense was the issue but user experience sucked.

I did a case study with my wife (she uses yahoo search still average user), on pages with my Pre-Panda Adsense placement and new placement.

I asked her to tell me what she thought about Pre Panda placement 2 blocks between title and content, she said I would get off that page and click back button.

Then I showed here new Adsense Placement with content up top and Adsense under the 3rd paragraph, she said "yes I like that much better and would read it".

The sites I have moved the units to new placement are already converting better higher CTR, bounce rate gone down to low 20's from 40's, page's viewed jumped to 5 avg from 3, time spent on site up from 3 to 5 min's. This is what Google is looking at. Now back to work on the other 1000's of pages, recovery to all my sites is just a few thousand clicks away and how many days?

Google is giving the internet a Facelift, instead of Panda should be Googlift Algo.

[DirigoDev]

I removed AdSense all together on one of my sites...

Back to my theory about the ratio of internal to external hyperlinks. Why would Google penalize sites for using AdSense? Makes no sense. I know that some have said that the presence of AdSense is an signal for low quality sites – and I agree in many cases. It is not specifically AdSense.

I’m guessing that by removing AdSense you’ve removed ~4 external links – possibly more. This brings the ratio of internal to external below ~45% which tells the algo that you’re not a farm. To test this, why don’t you put a few external links back on your pages to see if you’re penalized again?

I’m seeing very uneven linking ratios in the Panda losers.

[TheMadScientist]

Hmmm ... Not sure how a one page site would rank if your theory is correct? The internal to external link ratio would be way off from the threshold you have if the site had even one external link on it.

[DirigoDev]

Not sure how a one page site would rank if your theory is correct?

Perhaps one page websites have a different rule. Was your site a single page?

[Shatner]

@crobb Any insight into what you might have done to cause the partial recovery? Are the pages that have recovered only ones you have done specific work on? How many pages overall is your site?

On Monday, for the first time since Panda, I saw recovery on ONE specific keyword phrase... but it was only for one, single page which I had completely rewritten and added tons more content too. And even then... before Panda it was #1, post Panda #500 then Monday #10.

I was happy about that even though it wasn't much.

Then woke up this morning and checked again and it had dropped to #12.

Now it's down to #14.

Ugh.

[Andem]

I'm just thinking that there is a "freshness date" that is playing a part of it because the scraper sites I'm seeing are junk

There must be. An artcile published recently having to do with *North American* current events, by a native English speaker on a native English-speaking web site being outranked in the organic SERPS for all relevant keywords by a site registered in India for the #1 keyword (2 paragraphs), the second keyword for an obvious scraper (a full copy), fake US-Registered, Asian-IP site and the third and forth for domains/IPs from Asia (both more-or-less a full copy). Strangely, each link back to the original article without nofollow.

It's an F up'd algorithm.

(Please excuse the foul language)

Edit: Just to add: We are located where (and especially, in the same country) as the above example took place! Even though it's an article with regional relevance, scrapers from third world countries still rank higher.

[supercyberbob]

Andem

That's pretty funny, but not.

So many reports of scrapers outranking originals, so little time.

Is Google broken? Only YOU can decide. Next on FOX news.