When I first started this blog, I posted some thoughts on the Mayday Update of 2010 and how the Google Caffeine architecture was maybe a part of their ability to do some interesting things. A lot has changed with Google since that time, with much of the recent interest being on the Panda and Penguin updates. I haven’t posted much of my thoughts on those updates as neither had affected me much, until now.
On September 28, all the websites I watch had a marked drop in organic search engine traffic. I would guess my sites each lost 50-75% of their traffic, in an instant. It was like a switch was turned on (or off in this case).
I’ve had websites lose traffic before, but never quite like this. I went to the usual websites to see if other people were noticing changes as well. At that time, there was already talk of an “Exact Match Domain” update which was tweeted by Google’s head of the spam team. My sites don’t qualify as exact match domains, so I was thinking maybe something in their algorithm messed up Either that, or my sites got mis-classified. I figured more information would come out, and there would be some adjustments to the results.
There was a bit of a discussion, and a few people were mentioning they had websites that were affected that were not exact match domains. People started to speculate that there were two updates going on. After a few days, Google finally confirmed that there was another update going on at the same time. The new update was an update to the Panda algorithm. Unlike previous Panda updates, this was a revision of the actual algorithm, not just a “refresh” of the data used.
As I hadn’t really paid that much attention to Panda before, I did some reviewing of what Panda was. I already suspected it had to do with user engagement, but that has never been confirmed. I wasn’t sure how they were collecting data, but there are not many different ways they can get the kind of data needed to really measure user engagement.
This is just a guess on my part, but I do think user engagement is a great way to test a website’s quality. The first was what I mentioned in my early posts on Mayday, which is click-through rates of the SERPs. That would be a bit too simple though. For a while, Google had the option of removing a website from your personalized results, with the link to remove the site triggered when you quickly went back to Google results after visiting a website.
I think it would be obvious to use data of websites selected to be removed from people’s personalized results. I was surprised I didn’t read reports of people trying to manipulate those results – although I wasn’t really looking.
I haven’t seen those links to remove sites lately, so I’m thinking maybe now Google could just check how long someone is on a website before they return to Google. A quick return to Google would suggest the person didn’t find a full answer to their search.
Getting more information than that becomes a bit trickier. Google has denied that they use Adsense data, analytics data, or Chrome browser data. In the past, people suspected they may have used the Google toolbar data but that’s never been confirmed. Generally, I trust that Google is honest with what they tell the public. However, one thing I’ve learned is that Google will be less than direct in what they mean.
Could Google be using other ways to collect user engagement data? Perhaps they are not using Chrome itself, but user stats while they are logged into their Google accounts? Perhaps the chromebooks they gave away are being used to collect data? Perhaps android operating system collects some data?
The one piece I’m struggling with sorting out is how all of my sites got stuck in this algorithm. I don’t think it is just user engagement that would affect all of my sites. Some of my sites don’t get that much traffic to even allow for user engagement data, yet ranked well for specific phrases. I’d expect them to still rank well.
So what could have happened?
My sites have enough differences that I don’t think it would just be site layout and design either. Most of the URL’s have the keywords in the URL, even though not in the domain names. I think that is possibly one reason, especially as most of my sites have the same keywords in the page title, URL, and top text of the page. This site is one exception for most pages, yet it still has seen a drop in traffic. There has to be more than just “over optimizing” with structure.
I then thought maybe there is some sort of a degrading of website quality if it is on shared hosting, but I don’t think that would actually factor in.
The only thing left to tie my sites together is that they share an analytics account and an adsense account. I hate to think that one of my low quality sites/pages would cause all of my websites to be affected. I really doubt that is the case, but I am still scratching my head trying to figure out what it was that all of my sites have in common, aside from me.