November 14, 2014
Before my my Panda hit, I was averaging 7,400 visits a day. After, I went down to 4,700 a day. Now, I am up to 7,000 a day and growing, nearly to where I was before my penalty. It appears that I have mostly recovered from my Panda penalty which I recorded here last September.
Click the image to see the full penalty timeline.
The “too long, didn’t read” version of this story is this: I believe Google was experimenting with ranking pages for a wider range of unqualified keywords, starting around the end of April and ending with Panda 4.1 on September 24. On that date, they stopped the experiment, and returned websites back to their original April rankings for unqualified keywords. Thus, the Panda 4.1 update did not penalize websites for lack of quality, or otherwise reconfigure the algorithm significantly, but instead started ranking sites with a greater focus on select keywords, instead of ranking them for a wider range of keywords which they did over the summer.
You say this based on what, exactly?
My Panda penalty looked like this. Some pages on my site have been historically strong, and continued to perform well after Panda, with only a minor slump. My “beard itch” page is the prime example. It has climbed steadily, with only a minor blip after Panda, then suddenly a burst of traffic at the beginning of November.
Click to enlarge.
Other pages were definitely hit by Panda, but after a month, they have recovered well. My article on “how to fix a patchy beard” was a top performer until Panda. It has recovered well.
Click to enlarge.
Others were decimated by Panda, and they have not recovered at all. These are stats from my “how to grow a thicker beard” page, which was a champion on the site for months, until it was hit hard and never recovered.
Click to enlarge.
Here is how I can explain these wildly different traffic stats. My top performer, “How to deal with beard itch“, is benefiting from Movember. Right now, millions of men all over the world are growing mustaches and beards for the first time, and understandably have questions about beard itch. My site traffic is showing this phenomenon.
My “patchy beard” page is more representative of pages on the site. It definitely got hit by Panda 4.1, but has mostly recovered. This is, I believe, due to the natural growth of my website (as I make more content and establish a longer history, Google rewards it with more traffic).
My losses can all be explained by the traffic to my “thicker beard” page, and pages that have similar traffic stats. If you see my last post on this issue, I shared some keyword stats from Moz and Webmaster Tools, that showed that I dropped from the SERP for dozens of keywords which I previously ranked on the first page for (and now am not in the top 50). These keywords were never “money words”, but were long-tail keywords like “how to grow a thicker beard at age 16” or “how long does it take to grow a 6 foot beard?”. Pages like my “thicker beard” page benefited from these long-tail and low-volume keywords. In fact, pages like my “thicker beard” page were getting a substantial amount of traffic from one or two visits from extremely obscure keywords like “how to grow a faster beard as an asian when I can’t”, but thousands of them. In fact, that page only got a few thousand visits a month from people expressly doing research on growing thicker beards.
After my Panda penalty, the visits that stopped were those long-tail, obscure keywords. The page still got lots of traffic from keywords about growing thicker beards, but stopped getting traffic from obscure keywords.
This is why I think that Panda 4.1 re-evaluated pages in the “gray area”–that is, pieces of content that were judged to broadly cover many similar topics. They kept sending traffic to pages that were explicitly about one topic (“beard itch”), but they stopped sending traffic to pages they had, previously, a hard time deciding which topic the page is about. Now that they understand that my page is about growing thicker beards and ONLY thicker beards, it now only gets traffic from people doing research on growing thicker beards.
In the three weeks prior to Panda 4.1, my beard website referred 1,279 Amazon orders. In the three weeks after Panda 4.1, it referred 933. That is an order reduction of only 346–minus 27%. Google Analytics reports that I lost 52% of my traffic in the three weeks following Panda 4.1. So, even though I lost 52% of my traffic, my orders decreased by only 27%. This tells me that the bulk of the traffic I lost was unqualified traffic, which bolsters my idea that Google was sending me a bunch of visitors using obscure, less-than-on-topic keywords.
Of course, I’m happy to have 52% more traffic, as unqualified as it may be, because any amount of new traffic will improve sales by a little bit. That 27% loss was painful to my pocketbook, and I’m very glad for the recovery. Still, it would be better if that traffic was qualified.
There is much we can learn from this, if it is true. First, it shows us the power of narrow, focused content. Focused content has a greater chance to rank well, even if published on a website with low domain authority, if it is long, well-researched, and written authoritatively.
That said, a focused piece of content will miss out on traffic from the wide range of “close-but-not-quite” queries that more broadly written content can benefit from. At this time, I am not sure how successful you can be if you purposefully write a broad piece that covers many closely-related topics. In theory, you will have the opportunity to rank for many more keywords, but you will likely not rank as well as a piece of content that focuses on one idea.
This is what I tried
If you look back on the comments users made to my previous posts on this issue, I got many suggestions, ranging from “you need to cloak your affiliate links” to “You need to rewrite all content on your website”. My traffic is back, and I did neither of these things. In fact, I did almost nothing to get my traffic back. Here is what I did:
Homepage: I completely revised my homepage. It used to be a simple grid of products. It is now a more narrative, guided page. This change did nothing to improve my homepage traffic, but I will keep it as-is because I have grown to like it better. Incidentally, here is how my homepage traffic has changed during this time:
Click to enlarge.
Ads: I revised the layout of my ads. I removed ads altogether for many weeks, thinking, at first, that I had been hit with Google’s “Top-heavy” penalty. I covered this change in Recovering from Panda 4.1 – Step 1. I ended up replacing all my ads, with a few exceptions, weeks before I recovered from my penalty, as I gradually became convinced that ads had nothing to do with my penalty.
New Content. I added a some new content–not to fight Panda, but just because I would have added new content anyway. Here is the new stuff I made since Panda 4.1:
- Guide to Movember
- Pumpkin Spice Beard Oil
- How to Make Mustache Wax
- Goatee Styles
- Beard Trimming Accident
Revised Content: I am constantly editing and tweaking content to improve conversions. I did much of that last month. See my recent article on using Amazon conversion metrics to an example of some of the tweaks I made.
Cloaked Links: I tried, unsuccessfully, to cloak my links. I purchased the Pretty Link Pro plugin, which has a nice option for cloaking external links. The authors of the plugin were very responsive to my emails and helped me set it up. However, after cloaking my external links, I noticed that every Amazon.com product landing page I clicked on was blank. Come to find out, cloaking affiliate links is against Amazon’s terms of service, and if caught, they will remove content from any page you send a visitor to using a link cloaker. Thus, cloaking links became an option I could not consider, as my users couldn’t purchase anything through my links. It may work for other affiliate networks, but not with Amazon. I removed the plugin and did not cloak anymore links.
I think the fact that my website has recovered despite not cloaking my links allows me to definitively say that:
Things I still don’t get
There are still things I don’t understand about my Panda penalty, and I have made many observations that continue to baffle me.
Historically, my website has always followed the same pattern: Friday is a huge slump in traffic. Saturday is my second-lowest day. Sunday is my biggest day of the week, followed by Monday, and the rest of the days.
After my Panda penalty, Monday and Sunday switched places. Monday is now the biggest day of the week for me, and all days following Monday get much more traffic by percentage of daily average than before my penalty. Even after my recovery, this stays true (see the first image on this page). I have a hard time explaining this, unless Google thinks that some websites are more relevant to users on certain days of the week. I can’t imagine that users would change browsing habits that happen to coincide with a Panda update.
Around the same time, my visits from Facebook decreased by 63%, and my Twitter traffic decreased by 75%. True, I wasn’t consistently posting to Facebook or Twitter, nor was I running any Facebook or Twitter ad campaigns, but I wasn’t doing that consistently before Panda either. How can Facebook, Twitter and Google all penalize my site at the exact same time by reducing traffic to it by over 50%?
In the past few weeks, my traffic from social media has increased, doubling what I used to earn before Panda 4.1, thanks to the work I have been doing on Pinterest. But my Facebook and Twitter referral traffic is still lower than before Panda 4.1.
Search engine changes
Both Yahoo and Bing are sending me over 225% more visits AFTER Penguin 4.1. I can understand getting more visits, because neither engines are tied to Google’s algorithms and I expected my site traffic to increase naturally over time anyway, but by over 200%? That seems like a crazy (but welcome) increase, to me.
While my traffic from the U.S. has recovered, I’m still suffering globally. Here is how much traffic I have lost when I compare the past 2 weeks to my pre-Panda traffic:
- UK: -32%
- Canada: -27%
- India: -72%
- Australia: -58%
I expect this to recover slowly, but it’s interesting to see my U.S. traffic increase before my international traffic.
One final word
I’m sure after reading this post, the “correlation does not equal causation” crowd is cracking their knuckles in preparation for an epic takedown. I just want to say that what is most important to me is that my websites work by making me money. The “ins” and “outs” of Google and its algorithms are interesting. I’m publishing everything I have observed in hopes that some of it will help folks like me who may not be data scientists, but who still want to make a good living writing for the Internet.
As for me, I’m going back to making epic content and building websites.