Central Perk: How WPMU.org Recovered From The Penguin Update

marți, 29 mai 2012

How WPMU.org Recovered From The Penguin Update

Posted: 28 May 2012 02:53 PM PDT

This post was originally in YouMoz, and was promoted to the main blog because it provides great value and interest to our community. The author's views are entirely his or her own and may not reflect the views of SEOmoz, Inc.

Last Friday, Google pushed out the first refresh of their infamous Penguin update, bringing many webmasters to stumble towards their analytics, SERPs and the like in hopes of signs of a recovery from previous ranking drops, and/or to hope that they had not seen a precipitous drop in rankings. For most, the algo refresh seemed to do little, and for good reason - Cutts informed us that it apparently only impacted .01% of search results. Not many SERPs changed, and most of the voices heard over the weekend only seemed to be coming from people that were newly crushed by Google's pet Penguin.

Soon after the algorithm hit, we learned that Penguin refreshes, as it did last weekend. This makes it unique and similar to Panda in formation. For most, there was a lull period where these Penguin-impacted webmasters would sit around, gather the facts about the update, and then take action towards recovery where there were negative impacts in a week or two after the event. Many removed and/or edited links - others simply moved completely away from their manipulative linking strategies. However, because of this time lull between implementation and action, it's possible that moves towards recovery were not rapid enough to see full changes, as links take time to crawl, new actions take much time to implement, and the new refresh took only a month to once again take effect.

In that respect, it may leave some to think "is a Penguin recovery even possible?", or "should I just start over with a new domain?". These are real questions that will come with edits to links and strategy, a refresh and no changes to our rankings. To that end, I don't have absolute answers, nobody does - just strong suggestions about the data points we know about what survived, what didn't, and how Google has treated penalties in the past. What I do now know, though, is that a Penguin recovery is possible, and possible in a short amount of time - because I've seen a big, seemingly complete recovery from the update at the first refresh. This recovery came for a website that felt a previous, critical impact from Penguin at the first iteration - that website being popular Wordpress portal, WPMU.org.

The WPMU Story

On April 24th, 2012, WPMU.org was hit by the Penguin Update. Traffic from Google dropped over 81% week over week, causing a real, massive hit in revenue for the business over night. This was not the "three or four spots" Google Penguin drop, this was the "almost disappear completely" type Penguin hit that was among the worst kind of impact most websites felt - and for the owner, James Farmer, this came as a real, completely unexpected shock.

WPMU is a Wordpress information hub with resources, plugins and more - but the most important of its resource portfolio is its themes. WPMU's themes function like many Wordpress installs usually do - they create citation footer links to declare the theme type being used, so when its popular theme packs get installed, they generate a "Powered by X" link in the footer of the site back to the theme page.

Although it made sense in the context of these blogs, and for these types of themes, it also generated a high volume of sitewide links on low quality sites. It also, in its majority iteration, used the anchor "Wordpress Mu" - which is a somewhat valid iteration of "WPMU" - but to Google, it was likely seen as an attempt to get commercial anchor text pointing at the site.

WPMU Anchors

To WPMU founder James Farmer (as well as others), this was extremely frustrating. Wordpress and web design installs are a unique use case that might have been caught in the crossfire of this update. It simply makes sense for these sites to have a link in the footer back to the theme and/or designer - this is definitely what users expect, and is good from a usability perspective overall. However, when looked at purely from what we normally consider "clean" link profile characteristics, its raw numbers fell outside those "good" ratios - and surely, the nature of Wordpress themes and the majority of people who select them mean that a good amount will be low quality and/or spam.

However, WPMU clearly had many other things going for it. 10,700+ Facebook likes, 15,600+ Twitter followers, more than 2,500 +1s, and over 4,250 people subscribed to Feedburner in total. Its backlink profile includes links from Technorati, Ars Technica, Wired, Huffington Post, SEOBook, Business Insider, Boing Boing and more. Surely, this isn't a site that deserves to get penalized, right? Well, apparently Google thought differently.

Post penalty, Farmer reached out to the Sydney Morning Herald, the biggest news site in Australia, in hopes to get coverage of the events. He got what he asked for, and the Herald, according to Farmer, got his site in front of Cutts to ask why a domain like WPMU would get hit by Penguin. Cutts replied, pointing out links that in particular lead to the penalty - for example, the below pages. Copy and paste to view.

http://baydownloads.info/11580-Wordpress-Membership-Plugin-Wordpress-PayPal-R-Plugin-show-5starserve.htm - A site pirating WPMU software.
http://computerofficechair.blogdetik.com/category/tak-berkategori/ - A splog using an old theme pack with a link to WPMU in the footer - with said potentially "commercial" anchor (my words, not Matt's).
http://computerchairs.blogdetik.com/ - Same splog.

Cutts said, according to Farmer, "that we should consider the fact that we were possibly damaged by the removal of credit from links such as these". Sure, based on what we now know or assume about the update, this makes sense. Low quality links, and also spammy, rarely-clicked footer links with over-optimized anchor text. Right.

Although this information was helpful to Farmer, what also came from it was Google awareness of a site that potentially might not have really "fit" within what they were hoping to accomplish with this update. On top of Cutts now knowing about the changes, Farmer then went on to blog the details of the penalty on WPMU, leading to more coverage and links, tweets from Rand, and also, according to Farmer, Danny Sullivan of Search Engine Land putting the site in front of Google once again.

With the burst of awareness this created in the SEO community, many people, such as myself, ended up commenting on Farmer's post on WPMU. The community was gracious in offering advice, suggestions, and other reasons why the site may have been penalized - and from there, what Farmer might do to recover. Based on my comments and tweets at WPMU about the subject, Farmer reached out to me about taking next steps in undoing the impacts of Penguin. I obliged, and work began.

We had two choices for WPMU - get the nofollow attribute added to the links, or simply remove them completely. The first goal was to cut down on as many of the sitewide, "Wordpress MU" anchor text links as possible. I initially thought nofollowing would be the best solution because of the potential for these links to drive leads for Farmer and WPMU, but Farmer thought, to make it easy to change and correct for bloggers, the best solution was to simply ask for removal.

The EDUBlogs.org Removal

The most perilous piece of WPMU's link profile came from one site - EDUblogs.org. EDU Blogs is a blogging service for people in the education space, allowing them to easily set up a subdomain blog on EDUblogs for their school-focused site - in a similar fashion to Blogspot, Typepad, or Tumblr, meaning that each subdomain is treated as a unique site in Google's eyes. Coincidentally, this site is owned by WPMU and Farmer, and every blog on the service leverages WPMU theme packs. Each of these blogs had the "Wordpress MU" anchor text in the footer, which meant a high volume of subdomains considered unique by Google all had sitewide "Wordpress MU" anchor text. In what might have been a lucky moment for WPMU, this portion of their external link profile was still completely in their control because of WPMU ownership.

In what I believe is the most critical reason why WPMU made a large recovery and also did it faster than almost everyone else, Farmer instantly shut off almost 15,000 'iffy' sitewide, footer LRDs to their profile, dramatically improving their anchor text ratios, sitewide link volume, and more. They were also able to do this early on in the month, quickly after the original update rolled out. A big difference between many people trying to "clean up their profile" and WPMU is time - getting everything down and adjusted properly meant that many people simply did not see recoveries at refresh 1.1 - but that doesn't mean it won't happen at all if the effort persists.

Additional Cleanup

Once .EDUBlogs got cleaned up, the majority of the link profile had been fixed. However, much of the junk still remained from independent bloggers who put up WPMU themes. Because of time constraints, I was really unable to move at all on the link cleanup outside EDUblogs as we attempted to get an effective strategy in place for people to remove footer links, and also avoid Memorial Day weekend for e-mailing. Despite this, we still may move forward with cleaning up the remainder "junk" links to prevent Penguin hitting again on a second iteration.

Although Penguin seems to be a link penalty, I would be remiss to only mention the large link-based changes that were made to the site in the month between updates. Farmer and the WPMU team also made the following changes during that time, any, all, or none of which may have made an impact on recovery. I want to clarify, here, that these cleanups were all done by Farmer as overall quality value-adds, and were not Penguin-specific improvement suggestions made by me, although some, many, all, or none of them may have contributed to the recovery.

Pinged blogs that were originally highlighted by Matt Cutts in a conversation with the Syndey Harold - only one removed links, but they did come from a significant volume of splogs on the Blogdetik.com domain
Submitted WPMU to the Penguin review form, twice, specifically referencing this article that was being beaten out by the links that referenced it
Used SEOmoz campaign data to implement some canonical URLs to clean up crawl errors and also kill some unnecessary links across the site
Did a bit of "SEO cleanup" that revealed WPMU.org sitemaps did not exist and/or were broken. Implemented sitemaps and submitted the feeds to Webmaster Tools, which was not happening previously
Cleaned up numerous duplicate title tag issues as reported by Webmaster tools
Continued to build natural links to the site and promote other positive signals such as referring traffic and social shares
Very notably and importantly, got this specific use case in front of Google and also the greater SEO community that highlighted it

These aren't the only changes that occurred, certainly, but were the most notable in reference to the Penguin update, and may help in your own decision making in order to better recover your own website rankings in the future.

Just as I was about to start manually e-mailing the remaining blogs to remove WPMU links, a great thing happened - recovery. On Friday, May 25th, a clear return from the 1.1 refresh of Penguin occurred, bringing ranking and traffic levels to what look like the same spot they were previously. Given that it's a holiday weekend, traffic is considerably down, so it's hard to tell for certain - but considering what we know about traditional impacts from the holidays, it looks like WPMU has made a full recovery from its original hit from the Penguin update.

This Penguin recovery is a great sign not just for WPMU, but also other Penguin impacted webmasters as well. WPMU had a lot of things going for it that allowed for immediate and quick recovery - such as getting in front of Google (which may have caused an algorithmic adjustment for this use case), being a site that DESERVES to rank with tons of other great signals already, and also the ability to pull down tons of manipulative linking root domains instantly. However, these "quick fix" solutions that allowed WPMU to quickly come back also means that the long term fixes that you're working on for your domain should also work - that is, if you implement them properly and move towards a longer term, higher quality site as you should be.

It should also be noted and taken very seriously that this post should not be considered a "blueprint" for recovery for your site. Read it and make your own educated decisions based on what you know about your link profile, your business, your vertical, and the Penguin Update in general.

Best of luck - and happy Penguin hunting!

Although Farmer and his team at WPMU did most of the heavy lifting in this recovery, I'll do my best to answer any questions you might have in the comments. Feel free to also ping me on Twitter for a quicker response. Many thanks to Melissa Kowalchuk as well for her image design work on this post.

Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!