October 11, 2014
I’ve spent the past few days looking at the data. What I’ve discovered flies in the face of our accepted understanding of how Panda works.
When I first wrote about my Panda penalty, I got a lot of advice saying things like “add more text to your homepage” or “your site content is thin. Add more text to your landing pages”.
However, the data shows that my longest, best, most well-researched pieces of content have seen the steepest declines.
Below is a link to a report I pulled from Webmaster Tools. You can view the complete report here on Google Docs. Feel free to save a copy for yourself, and play with the data.
In this report, I share with you the search queries I was ranking well for, and the percentage my site has declined in visibility. In column 2 you see “impressions”. This is how many times my site appeared in the SERPs for that keyword. In column 3 is “change”. This is the percentage my visibility has changed over the course of September. This effectively gives me a SERP visibility comparison between post and pre penguin 4.1.
You’ll notice that the largest negative percent changes are for long-tail keywords. The real story is told on keywords that get 10 or more impressions during this period. We see that “trimming beard” and “full beard styles” got -90% impressions. “beard trimming”, “how to trim beard”, and “how to make a beard grow faster” all have -80% impressions. The thing is, I have unique, original, long-form pieces of content for each of these keywords.
In short, the pages that saw the steepest declines were not my short product landing pages, but rather my long, epic, well-researched, expertly-written evergreen pieces of content. This tells me that my site was not hit by Panda 4.1 because of “thin” content. It was hit for some other reason.
Here’s another report I made using my Google Analytics data. In this report, I share with you my exact page-views per page one week before Panda 4.1, and the week following panda 4.1. Below is a screenshot from the data. I have also uploaded it to Google docs for you to play with, here.
Column B shows you the exact weekly page views each landing page received after Panda 4.1. Column I shows you the same number before my site was hit.
In column D, you see the difference between each week. In column F, you see the percentage increase or decrease. The higher the percentage, the harder that page was hit by Panda.
We immediately see that the pages hit the hardest are not product pages, but rather guides and long blog posts. “How to trim a beard”, which has original images and 1,209 words, suffered an 86% decrease in page views—3,006 total. “How to grow a beard faster” suffered a 96% drop, “How to grow a beard” lost 83%. And so on.
The pages that suffered the least, contrary to our common understanding of Panda 4.1, were the pages that had the least text. My homepage, for example, previously had zero (0) words on it. I ended up changing it a few days ago, but it remained the same the week following my Panda hit. And yet it only suffered a 36% drop during that time. Still huge, but much smaller compared to some of my more epic guides. My Best Sellers page, which again has only a paragraph of text, only lost 42%. My beard wax landing page only lost 24%.
It’s Not About Content Length
Now, this is still very bad. I feel strange writing ‘only” 42%. Surely, my site was hit very hard indeed. But compared to the big losers, my shorter landing pages fared better. This turns the idea of what constitutes “thin” content on its head.
This is much more in line with how I always thought Google worked. I can’t believe Google would level an algorithmic site-wide penalty on a website with good content that has some pages that are short. That is, I can’t believe Google will ding a site that has 80% 2,000 word pages, and 20% 500 word pages. I don’t think Google works that way.
Instead, I think Google judges each individual piece of content by its own merits. Yes, domain authority has an impact, and yes, overall site reputation and topical affinity have impacts. But more than anything, I think the content of the individual piece has much greater weight, which is why we see articles from HuffPo ranking #1 for certain phrases. For example, as of this writing, The Art of Manliness now ranks #1 for “beard oil recipe” (taking the throne from me since Panda 4.1), despite not being a beard-niche site. It is an all-around “man” site.
Such articles can only win such victories if Google judges the quality of a piece on its own merits, not by the other content on the website.
The same is true for penalties. I don’t think my whole website got penalized because some of the content on it is short. I’ve been looking at content from National Geographic, The Smithsonian, and The Atlantic lately. Some of it is very good. Much of it is short, filled with ads, and doesn’t really answer my question. In fact, these kinds of websites have an overwhelming majority of content that is short, ad-ridden, and not very high quality.
To cap this sundae of data with a rather eloquent cherry, I’ll end by saying that I do not think my site was hit due to the content quality at all. I don’t think that the pages that were hit hardest were any worse in quality than the pages that were hit least. Instead, I think the site was hit because Google added a new signal to its Panda algorithm, and my site triggered that signal.
So. What is the signal? Is it too many ads above the fold? I don’t think so. I removed my ads a few days after my initial penalty. It has been over 2 weeks since then, and though my site is growing, it is nowhere close to where it was. In fact, I intend to replace my ads (within moderation, of course). Removing those ads has cost me hundreds of dollars.
All that is left—all that make sense at this point—is my non-cloaked affiliate links. From the very beginning, I brought this up as a possibility and yet I was hesitant to believe it. After all, if cloaking affiliate links actually works in fooling Google, then Google is very stupid. Maybe it will work, and maybe it won’t, but having ruled out content quality (or at least, content length), and having already moderated my ads, it’s the last thing I can try.
I’m going to cloak my links. I’ll let you know how it goes.