Friday, January 18, 2013

SEO Issues - is it Penguin? Is it Panda? or is it me?

The following story is one that has been several months in the making. It's one that I have lived through one too many times as an SEO, and it is one that I am sure other SEO's have faced. I fought with the thought of writing this for fear that someone from the company might read it and get angry that the story is told. But, it's something I think that not only people out there could learn from, but speaks to so many others in this industry to show them that they are not alone.

It's long, it's a bit technical (I tried to keep it simple), and it has some personal frustrations laid out in words. My only hope is that you get value out of reading this as much as living it has made me a better person (or well, a better SEO).

It Begins


I started working on this website's SEO in May 2012 at which time I was told the site's traffic was declining due to Panda updates. In February of 2012 the traffic from SEO was the best they had ever seen, but soon after that there was a steady decline.
Traffic from February 2012 - May 2012
Before digging into any possible SEO issues, I first checked the Google Trends to ensure that the decline isn't searcher related. Often times a drop in traffic could just mean that users aren't searching for the terms the website is ranking for as they were in the past.

Top Key Terms in Google Trends
Looking at the same time frame as the traffic data, I noticed an increase in searches for the top 3 terms the website ranked for, and there appeared to be a decline around the same time from March to April that the traffic was reflecting. But there was a drop in the website's traffic in April from the 23rd to the 24th and then significantly on the 25th. The website I was working on had two SEO's already working on it: an agency and a consultant. Both had already done a numerous amount of research and some work to get the website on track. Both were stressing that the drop in traffic was due to the Panda updates by Google. I looked at SEOmoz's Google Algorithm Change History and found an update to Google's Panda on April 19th and an update to Penguin on April 24th. Given that the traffic significantly dropped on the 24th my best guess is that it was possibly Penguin related, but still needed further exploration.

Figuring Out What Was Hit by Penguin.


The site is/was broken up into sections by keyword focus. At one point, I could tell that someone really had a good head on their shoulders for SEO, but the strategy that was used was outdated. Perhaps the site was originally optimized several years before, and it just needs some cleanup now to bring it up to 2012's optimization standards. So, understanding Penguin and identifying which part of the site was driving the bulk of the organic traffic was going to be my next step in solving this mystery. Once I understood why, and where, then I could start to establish a what to do to solve the problem.

I broke the site traffic report by sections as best I could in Google Analytics. There was a bit of a struggle as all of the pages of the site resided on the main domain. Without a hierarchy in place, breaking out the sections had to be accomplished with a custom report and a head matching for landing pages. I hadn't had to do this before, so the agency that was working with the site already helped build the first report, and I began building out the other reports from there.
Click to View Larger
Section 1 over 72% of traffic

Just focusing on April and May I created a Dashboard in Google Analytics focusing on organic Traffic and identifying the sections of the site. Looking at the different sections - Section 1 was the bulk of the traffic with over 72% and Section 2 coming in second with just over 15%. Subs of Section 3 and other one-off pages make up the difference.

Both Section 1 and Section 2 dropped off after the April 24th date, so clearly they were the bulk of what was pulling the overall traffic numbers down. Since Section 1 was the majority of the traffic, I presented to the executive responsible for the site that we address any issues with that page first.

Actual screenshot of Section 1 presented
I took all of the research from the agency and consultant and we quickly reworked the pages to represent a hierarchy in the URL structure, and cleaned up any issues from the outdated optimization that was done.

Soon after Section 1 was addressed, we did the same with Section 2, and then worked on Section 3 (and sub pages, rolling them up into a solid section) and then added a few pages to grab any new opportunity.

Not Quite As  Easy as it Looks


The projects were launched in increments - first URL hierarchy fix to Section 1 and then the page redesign. Next was a full launch of URL fixes and page redesign to Section 2, and then lastly Section 3 and the new Section 4.
Section 1 - Section 2- Section 3 Launch Dates and Organic Traffic
Soon after Section 1 was launched traffic started declining rapidly. I was asked several times why traffic was getting worse, and I started digging some more. Every time I looked at the Impressions of the new URLs from Section 1 they weren't getting any traction, but the previous URLs were still.  I began looking at the history of the website, trying to find out why it was doing so well at one point, but was not doing well at that time. One of the things I noticed was that there was a lack of priority linking to these pages, but at some point there were links to some of them individually from the homepage. Google matches a hierarchy of pages to a directory structure that links are presented on a site. This site had every page on the first level, and linking to those pages from the homepage, which was telling Google that every page was the most important page. It worked at one time, but as Google has been rolling out their 2012 updates these pages were getting hit, and those links on the homepage weren't there anymore. Before the launch of Section 2, I had them put links to the main directory for each section on the homepage. The links would tell the search engines that these are important pages of the website, but not be so obnoxious with a dozen or more links on the homepage to discourage users (avoiding the appearance of spamminess).

But - even after adding the links to the homepage, the traffic to those pages was still declining. Pressure was put on me to figure out what was wrong. In addition, accusations were flying that I single-handedly ruined the SEO for the site, I spent every waking hour looking at reports, and trying to figure out what was going on. I consulted friends in the industry, and read every article I could find to figure out what Panda or Penguin updates were affecting these pages.

Then it hit me - just as the links to these sections would help them get recognized as important pages, so were the other pages that were being linked to from the homepage. In fact a set of them linked to the website's search results with queries attached to them mimicking pages, but showing search results. On those search results pages, there were over 200 links with multiple (we're talking hundreds - possibly thousands) combinations of parameters. The bots were coming to the homepage, going to the links to the search results pages, and then getting stuck in this vortex of links and combinations of parameter generating URLs - not allowing any crawl time for the pages that once were getting rankings. This also explains why the new URLs weren't showing very many impressions in the Webmaster Tools Data - those pages just weren't getting crawled.

There was a project underway that would solve the many links on the search pages, and there was also talk of using ajax to show the results. When this project would launch, the bots would go to the URL from the homepage, but would then essential not go much further. With this project a few months out, I made the case to add the search page to robots.txt to allow the bots to then recognize the Sections as important pages. After several weeks of attempting to convince the powers that be, the URL was eventually added to the robots.txt file.

Immediately after the search page was added to the robots.txt Google Webmaster tools presented me with a warning:
Warning in Webmaster Tools
In most cases, a warning from Google should never be taken lightly, but in this case it was exactly what I wanted. In fact it proved to me that my theory was correct, and that the site was hopefully headed down the right path.


Panic, Questioning, and a Third Party


As with every up in the SEO world, there must be a down. Soon after the search result page was added to the robots.txt the organic traffic to the site dropped, and continued to drop. Throughout those grueling three months there were several Google Panda and Penguin updates. I documented each and every one of them in Google Analytics, and continued to answer questions, gathering data, and dealing with being under close scrutiny that the work I was doing was complete BS.
Organic Traffic from September 2012 - November 2012
I sat in numerous meetings, some of which I walked out crying (I'm not afraid to admit it), being questioned about the road I had taken and why we weren't seeing results. There were people within the company recommending that they roll the pages back to where they were before, and even changing the URLs. I fought hard that they don't touch a thing. I sent an article posted on Search Engine Land by Barry Schwartz citing Google's patent that "tricks" search spammers.

The patent states:

When a spammer tries to positively influence a document’s rank through rank-modifying spamming, the spammer may be perplexed by the rank assigned by a rank transition function consistent with the principles of the invention, such as the ones described above. For example, the initial response to the spammer’s changes may cause the document’s rank to be negatively influenced rather than positively influenced. Unexpected results are bound to elicit a response from a spammer, particularly if their client is upset with the results. In response to negative results, the spammer may remove the changes and, thereby render the long-term impact on the document’s rank zero. Alternatively or additionally, it may take an unknown (possibly variable) amount of time to see positive (or expected) results in response to the spammer’s changes. In response to delayed results, the spammer may perform additional changes in an attempt to positively (or more positively) influence the document’s rank. In either event, these further spammer-initiated changes may assist in identifying signs of rank-modifying spamming.
 But the article and my please fell on deaf ears...

It had gotten so heated and there was fear that nothing was being done while traffic was significantly declining that the company brought in yet another SEO consultant to look at the site objectively.

Just as the consultant was starting his audit, and the traffic hit the lowest I ever thought it could possibly go, the next day traffic went up. The last week in November (roughly 3 months after we blocked the search result page) I saw an increase in traffic in Google Analytics to Section 1:
Section 1 Organic Traffic
I quickly pulled up my report to check the Section's impressions from the Webmaster Tools data, and there was a significant increase as well:
Section 1 Impressions from Webmaster Tools Data
On December 3, 2012 I logged into Webmaster Tools and saw that the warning had gone away:
It was the "halleluiah" moment that every SEO dreams of, and very few get. All the work I had done, the fighting for what I believed in, it all finally paid off.

To this day traffic continues to increase - we can now focus on some of the cleanup still left to do, and then onto projects that will attract new opportunity.
Organic Traffic from November 2012 - January 17, 2013 (day before this post is written)
Quick Note: 
I forgot to mention a post I wrote months ago while going through all of this - SEO - Panda and the Penguins. It helps to give a bit of perspective of some of the linking stuff I didn't get into in this post.