Posted by Aaron Wheeler
It's here! Google has released Panda update 2.2, just as Matt Cutts said they would at SMX Advanced here in Seattle a couple of weeks ago. This time around, Google has - among other things - improved their ability to detect scraper sites and banish them from the SERPs. Of course, the Panda updates are changes to Google's algorithm and are not merely manual reviews of sites in the index, so there is room for error (causing devastation for many legitimate webmasters and SEOs).
A lot of people ask what parts of their existing SEO practice they can modify and emphasize to recover from the blow, but alas, it's not that simple. In this week's Whiteboard Friday, Rand discusses how the Panda updates work and, more importantly, how Panda has fundamentally changed the best practices for SEO. Have you been Panda-abused? Do you have any tips for recuperating? Let us know in the comments!
Howdy, SEOmoz fans. Welcome to another edition of Whiteboard Friday. This week, we're talking about the very exciting, very interesting, very controversial Google Panda update.
Panda, also known as Farmer, was this update that Google came out with in March of this year, of 2011, that rejiggered a bunch of search results and pushed a lot of websites down in the rankings, pushed some websites up in the rankings, and people have been concerned about it ever since. It has actually had several updates and new versions of that implementation and algorithm come out. A lot of people have all these questions like, "Ah, what's going on around Panda?" There have been some great blog posts on SEOmoz talking about some of the technical aspects. But I want to discuss in this Whiteboard Friday some of the philosophical and theoretical aspects and how Google Panda really changes the way a lot of us need to approach SEO.
So let's start with a little bit of Panda history. Google employs an engineer named Navneet Panda. The guy has done some awesome work. In fact, he was part of a patent application that Bill Slawski looked into where he found a great way to scale some machine learning algorithms. Now, machine learning algorithms, as you might be aware, are very computationally expensive and they take a long time to run, particularly if you have extremely large data sets, both of inputs and of outputs. If you want, you can research machine learning. It is an interesting fun tactic that computer scientists use and programmers use to find solutions to problems. But basically before Panda, machine learning scalability at Google was at level X, and after it was at the much higher level Y. So that was quite nice. Thanks to Navneet, right now they can scale up this machine learning.
What Google can do based on that is take a bunch of sites that people like more and a bunch of sites that people like less, and when I say like, what I mean is essentially what the quality raters, Google's quality raters, tell them this site is very enjoyable. This is a good site. I'd like to see this high in the search results. Versus things where the quality raters say, "I don't like to see this." Google can say, "Hey, you know what? We can take the intelligence of this quality rating panel and scale it using this machine learning process."
Here's how it works. Basically, the idea is that the quality raters tell Googlers what they like. They answer all these questions, and you can see Amit Singhal and Matt Cutts were interviewed by Wired Magazine. They talked about some of the things that were asked of these quality raters, like, "Would you trust this site with your credit card? Would you trust the medical information that this site gives you with your children? Do you think the design of this site is good?" All sorts of questions around the site's trustworthiness, credibility, quality, how much they would like to see it in the search results. Then they compare the difference.
The sites that people like more, they put in one group. The sites that people like less, they put in another group. Then they look at tons of metrics. All these different metrics, numbers, signals, all sorts of search signals that many SEOs suspect come from user and usage data metrics, which Google has not historically used as heavily. But they think that they use those in a machine learning process to essentially separate the wheat from the chaff. Find the ones that people like more and the ones that people like less. Downgrade the ones they like less. Upgrade the ones they like more. Bingo, you have the Panda update.
So, Panda kind of means something new and different for SEO. As SEOs, for a long time you've been doing the same kind of classic things. You've been building good content, making it accessible to search engines, doing good keyword research, putting those keywords in there, and then trying to get some links to it. But you have not, as SEOs, we never really had to think as much or as broadly about, "What is the experience of this website? Is it creating a brand that people are going to love and share and reward and trust?" Now we kind of have to think about that.
It is almost like the job of SEO has been upgraded from SEO to web strategist. Virtually everything you do on the Internet with your website can impact SEO today. That is especially true following Panda. The things that they are measuring is not, oh, these sites have better links than these sites. Some of these sites, in fact, have much better links than these sites. Some of these sites have what you and I might regard, as SEOs, as better content, more unique, robust, quality content, and yet, people, quality raters in particular, like them less or the things, the signals that predict that quality raters like those sites less are present in those types of sites.
Let's talk about a few of the specific things that we can be doing as SEOs to help with this new sort of SEO, this broader web content/web strategy portion of SEO.
First off, design and user experience. I know, good SEOs have been preaching design user experience for years because it tends to generate more links, people contribute more content to it, it gets more social signal shares and tweets and all this other sort of good second order effect. Now, it has a first order effect impact, a primary impact. If you can make your design absolutely beautiful, versus something like this where content is buffeted by advertising and you have to click next, next, next a lot. The content isn't all in one page. You cannot view it in that single page format. Boy, the content blocks themselves aren't that fun to read, even if it is not advertising that's surrounding them, even if it is just internal messaging or the graphics don't look very good. The site design feels like it was way back in the 1990s. All that stuff will impact the ability of this page, this site to perform. And don't forget, Google has actually said publicly that even if you have a great site, if you have a bunch of pages that are low quality on that site, they can drag down the rankings of the rest of the site. So you should try and block those for us or take them down. Wow. Crazy, right? That's what a machine learning algorithm, like Panda, will do. It will predicatively say, "Hey, you know what? We're seeing these features here, these elements, push this guy down."
Content quality matters a lot. So a lot of time, in the SEO world, people will say, "Well, you have to have good, unique, useful content." Not enough. Sorry. It's just not enough. There are too many people making too much amazing stuff on the Internet for good and unique and grammatically correct and spelled properly and describes the topic adequately to be enough when it comes to content. If you say, "Oh, I have 50,000 pages about 50,000 different motorcycle parts and I am just going to go to Mechanical Turk or I am going to go outsource, and I want a 100 word, two paragraphs about each one of them, just describe what this part is." You think to yourself, "Hey, I have good unique content." No, you have content that is going to be penalized by Panda. That is exactly what Panda is designed to do. It is designed to say this is content that someone wrote for SEO purposes just to have good unique content on the page, not content that makes everyone who sees it want to share it and say wow. Right?
If I get to a page about a motorcycle part and I am like, "God, not only is this well written, it's kind of funny. It's humorous. It includes some anecdotes. It's got some history of this part. It has great photos. Man, I don't care at all about motorcycle parts, and yet, this is just a darn good page. What a great page. If I were interested, I'd be tweeting about this, I'd share it. I'd send it to my uncle who buys motorcycles. I would love this page." That's what you have to optimize for. It is a totally different thing than optimizing for did I use the keyword at least three times? Did I put it in the title tag? Is it included in there? Is the rest of the content relevant to the keywords? Panda changes this. Changes it quite a bit.
Finally, you are going to be optimizing around user and usage metrics. Things like, when people come to your site, generally speaking compared to other sites in your niche or ranking for your keywords, do they spend a good amount of time on your site, or do they go away immediately? Do they spend a good amount of time? Are they bouncing or are they browsing? If you have a good browse rate, people are browsing 2, 3, 4 pages on average on a content site, that's decent. That's pretty good. If they're browsing 1.5 pages on some sites, like maybe specific kinds of news sites, that might actually be pretty good. That might be better than average. But if they are browsing like 1.001 pages, like virtually no one clicks on a second page, that might be weird. That might hurt you. Your click-through rate from the search results. When people see your title and your snippet and your domain name, and they go, "Ew, I don't know if I want to get myself involved in that. They've got like three hyphens in their domain name, and it looks totally spammy. I'm not going to get involved." Then that click-through rate is probably going to suffer and so are your rankings.
They are going to be looking at things like the diversity and quantity of traffic that comes to your site. Do lots of people from all around the world or all around your local region, your country, visit your website directly? They can measure this through Chrome. They can measure it through Android. They can measure it through the Google toolbar. They have all this user and usage metrics. They know where people are going on the Internet, where they spend time, how much time they spend, and what they do on those pages. They know about what happens from the search results too. Do people click from a result and then go right back to the search results and perform another search? Clearly, they were unhappy with that. They can take all these metrics and put them into the machine learning algorithm and then have Panda essentially recalculate. This why you see essentially Google doesn't issue updates every day or every week. It is about every 30 or 40 days that a new Panda update will come out because they are rejiggering all this stuff.
One of the things that people who get hit by Panda come up to me and say, "God, how are we ever going to get out of Panda? We've made all these changes. We haven't gotten out yet." I'm like, "Well, first off, you're not going to get out of it until they rejigger the results, and then there is no way that you are going to get out of it unless you change the metrics around your site." So if you go into your Analytics and you see that people are not spending longer on your pages, they are not enjoying them more, they are not sharing them more, they are not naturally linking to them more, your branded search traffic is not up, your direct type in traffic is not up, you see that none of these metrics are going up and yet you think you have somehow fixed the problems that Panda tries to solve for, you probably haven't.
I know this is frustrating. I know it's a tough issue. In fact, I think that there are sites that have been really unfairly hit. That sucks and they shouldn't be and Google needs to work on this. But I also know that I don't think Google is going to be making many changes. I think they are very happy with the way that Panda has gone from a search quality perspective and from a user happiness perspective. Their searchers are happier, and they are not seeing as much junk in the results. Google likes the way this is going. I think we are going to see more and more of this over time. It could even get more aggressive. I would urge you to work on this stuff, to optimize around these things, and to be ready for this new form of SEO.
Thanks everyone for watching. Look forward to some great comments, questions, feedback in the post. I will see you again next week for another edition of Whiteboard Friday. Take care.