Lessons from a Confirmed Penguin 3.0 Recovery

[adder]

Hi,

I've got another confirmed Penguin 3.0 recovery and I thought I'd share some details with you. Who knows, maybe somebody finds this useful.

[imgur.com...]

Attached is the weekly sessions report for Google organic source and there's been an average of 31% increase in organic traffic.

It's a medium-size website in a specific healthcare niche.

It has never had any manual actions and it has been hit by each successive Penguin update starting from April 24, 2012. Except for this last one, of course.

The client's webmaster has been updating disavow files ever since Google opened the tool to general public. And I think the main problem was with the disavow file.

I started by removing as many links as possible manually. It wasn't a big success because the majority of the links were controlled by a network owned by one of the client's previous SEO companies. BTW, they're still operating and have taken on new staff, so who says spam doesn't pay?

The amount of links removed was negligible, so there's no way this could have lead to recovery in isolation.

Then the disavow file. There were a few issues with it and I'll try to be as detailed as possible to hopefully help someone not to repeat these:

1. It had a lot of false positives - sites that you don't have to disavow. For example, a respectable infographic site that has nofollowed ALL its external links. And it's been nofollow since the dinosaur ages, so no need to disavow that. Also, a lot of resource list links. I never disavow links only because they're on a resource list or a page that's got "links.html" in its url. There are good resource pages and bad ones.

2. Typos. The file had a few commas and it also had this line in the middle of the list:
"domain:example.com,"
quotes and a comma. I don't know how Google handles typos, so it's better to make sure you don't have any.

3. The most popular Article directories were missed. Even the baddie daddy of the article directories was left out.

4. The original file had lots of blank rows. I have no indication to believe blank rows can cause problems but since it only took a few seconds I got rid of them.

5. It only contained domains downloaded from GWT. I used additional two link crawler tools and combined the results ending up with a much longer list.

In terms of recovery, it is nowhere near the pre April 24, 2012 traffic level but this is understandable because the site has now lost the majority of links that helped it rank high back then.

Normally, when you're confident, you do branding/link building alongside the recovery effort so that you get more traffic when the recovery kicks in. The client had no confidence in this site so he didn't allow me to build any fresh links. Now that the phones are ringing once again, he wants more links but, I tell you, he's not getting many - you need to read this post on Google Historical Data Patent to see what I mean:

[webmasterworld.com...]

A site that has had no link building history for the last 12 months can't just start adding dozens of links without triggering another penalty.

I uploaded the disavow file in early September and there was no movement whatsoever until the Penguin update happened. So, a disavow file has to be "activated" in order for it to start working. On this occasion, it was the update itself that activated the file.

[Shepherd]

Thanks for sharing Adder, seems like some good solid stuff in there.

Have you had a chance to figure out where (keywords) the organic increase is coming from?

Do you know is the site was hit for specific keywords during the first penguin run (2012)?

[Shai]

Great feedback Adder. Here are some comments on the specific points made regarding the Disavow file:

1. I agree, having a links page does not necessarily make the links bad. We also keep some of these unless the page is very bad or has more than one page of links (link-1.htm, links-2.htm... links-142.htm etc)
2. Google normally reports any errors on the page just after hitting the submit button. The indication is that the domains that are found with no errors are being processed. In the example you provided with the comma after the domain, it would be ignored as suspected.
3. This (along with point 5), I suspect is the main reason the previous attempts failed.
4. As you suspected, blank rows are ignored.
5. Another big reason why I suspect previous attempts did not work. We have had Google reject reconsideration requests in the past citing examples of links that have not been removed. Those links were neither reported on Google WMT or even on Majestic. We now use a combination of WMT, Majestic, Ahrefs and Spybot to harvest links before going through them (manually)

I uploaded the disavow file in early September and there was no movement whatsoever until the Penguin update happened. So, a disavow file has to be "activated" in order for it to start working. On this occasion, it was the update itself that activated the file.

This is a wrong description in my opinion. The reason you did not any movement until the Penguin refresh is purely because that's how Penguin works. The Disavow file does not need to be activated. It works as a filter between the outside world and your site. Every time a link is followed it goes through this filter and if the domain/url is on the list, it get neutralised by turning the link to a NOFOLLOW. So, in essence, its a gradual process as some links can take weeks before they are crawled again while others are checked very regularly. Just as an example, when we are dealing with a manual penalty, we give the file approx. 4 weeks before we submit a reconsideration request. In a couple occasions, we have had one RR rejected and then, 2 weeks later accepted with the same file in place.

Good job and good luck with the site. my only suggestion is to do another check for links in a month's time to keep on top of the disavow file.

[Planet13]

Thanks for the detailed post.

Question: Is there a place in webmaster tools where one can view / edit / delete the previous disavow file? Can't seem to find it...

[martinibuster]

Congrats on the recovery! Well done! ;)

A site that has had no link building history for the last 12 months can't just start adding dozens of links without triggering another penalty.

Not my experience. Of course, for any assertion made about Google you can guarantee someone will crawl out of the woodwork with a contradictory experience.

Here's the issue. It might not be the dozens of links that are re-triggering the penalty. It could be the links are poisoned, could be certain link building ratios are all wrong. It could be the method used to gain the links is poisoned. That's just for starters. I can go all night listing all the wondeful ways a link can wreck your day.

[Shai]

Question: Is there a place in webmaster tools where one can view / edit / delete the previous disavow file? Can't seem to find it...

You can view the currently loaded file only.

[Shai]

@martin - Yep, I agree with that. Why would there be such a trigger. I think its actually more natural for sites to receive waves/spikes of link rather than a gradual trickle. It all depends on the situation of course but logically speaking, I cant see why a site will be penalised or trigger some sort of review if it has a spike in links for publishing an exclusive or some other sort of link bait. Just one of those old myths that was repeated enough times to make it true.

[mirrornl]

Question: Is there a place in webmaster tools where one can view / edit / delete the previous disavow file? Can't seem to find it...


You can view the currently loaded file only.

In other words: it is overwritten when you upload a new one

[martinibuster]

Re triggering penalties because of historical data, here's the original thread, started by Tedster. Please take a moment to read it because it's a good one. ;)

http://www.webmasterworld.com/google/3747164.htm [webmasterworld.com]

What stands out is that the patent covers ordinary situations like a domain drop accompanied by a change in content. The part about links acknowledges the possibilities for false positives, for example when a site becomes topically relevant or because of the seasons.

Obviously, there are many other legitimate reasons why a site may experience a rise in links. Historical data for links (and link velocity) are both subject to false positives in the context of link spam detection. That's why, as I understand it from reading scientific papers on the topic of link spam detection, that other methods appear to be preferred, because those other methods are more accurate. I have read many papers on the topic of link spam from Microsoft, Google, Yahoo and other organizations. Historical data in the context of link spam isn't really discussed in the research papers. I suspect that is because it's not a significant factor for link spam detection and other more reliable and accurate methods are preferred. But it may play a role in other contexts like dropped domains and in the process of depreciating the citation value of a link, which is part of the overall calculation of PageRank, relevance, authority, etc.

[Planet13]

You can view the currently loaded file only.

So this means that you CAN see the current disavow that was uploaded previously?

[Shai]

Correct.

[adder]

Do you know is the site was hit for specific keywords during the first penguin run (2012)?

@Shepherd, yes, we have two groups of keywords "money keywords" and "browsing traffic keywords" - the money keywords were most affected during the penalties. Some browsing keywords were not affected at all. Now we're seeing some positive movement around the money keywords.

This is a wrong description in my opinion. The reason you did not any movement until the Penguin refresh is purely because that's how Penguin works.

@Shai, I refer to the speculation that some people have seen Penguin recovery between the updates.

Question: Is there a place in webmaster tools where one can view / edit / delete the previous disavow file?

@Planet13, When you go to this page [google.com...] it lets you view the old report and when you upload a new one, the previous disavow file gets overwritten.

I can go all night listing all the wondeful ways a link can wreck your day.

@martinibuster, thank you. Yes, you're right. Maybe I'm getting overcautious.

[martinibuster]

Maybe I'm getting overcautious.

Better overcautious than under. You're an asset to your client. ;)

[Awarn]

I take it this was from a site that received a notice stating there was a backlink issue. I would like to know of cases where an algorithmic penalty was resolved. Manual penalties seem to be able to be fixed.

[xelaetaks]

Something I wonder is if people can see recoveries improve over updates. For example maybe a penguin upate processed a percentage of the disavow file so there is a boost from an update, then by the time another penguin update comes around maybe all or mostof the disavowed links has been crawled by Google so a site can see an even bigger boost.

[Shai]

@awarn No, Adder is describing an algorithic penalty (in this case Penguin) so no manual penalty notice received. With regards to cases of resolved algorithmic penalties, there has been quite a few reported over the last refresh including a few from sites we look after. Have a look at the penguin thread for more info.

[adder]

@martinibuster, thank you, so kind! :)

@Shai, @Awarn, yes, exactly - this site has never had a manual action. The whole OP was based on experience with steering out of Penguin.

It has never had any manual actions and it has been hit by each successive Penguin update starting from April 24, 2012.

@xelaetaks, from personal experience with disavowing and link removal, I'd guess that the disavow file gets processed in its entirety (provided it's been submitted at least 4-6 weeks prior to the beginning of rollout of the update) because looking at other successful recoveries I've never seen a further traffic "jump" after a consecutive update. Only the initial recovery plus a gradual increase in traffic as you carry on with brand building and link building. That's my personal view. I haven't got scientific data to prove this.