Panda Updates: Content Quality, Link Profile or User Behavior? ━ SEM Synergy Extras
An interesting debate went down in Bruce’s office last week. Usually, discussions about factors that underlie SEO strategy go on behind closed doors and away from outsiders’ ears, but not this time. On today’s episode of SEM Synergy, Bruce, Jess and I postulate what could be the target of the Panda Updates, Google’s most public actions to improve the quality of its organic search results. Excerpts from the show follow.
Jessica: I was reading a post on Search Engine Land and it said that Perez Hilton’s site actually was improved, the rankings, after this Panda 2.5. Do you know Perez Hilton? He’s like this celebrity gossip person. I took a look at his site and I’m really trying to figure out, what is it that Google thinks is quality about this site. There’s lots of ads on it. The content ranges everywhere from a post that’s like 100 words to a post that’s like 900 words, but it’s not written at quality by what our standards would be. There’s Meta information there but it’s kind of sloppy. I’m wondering if it’s really just that you have to dot your Is and cross your Ts and make sure you have all these elements there, versus it actually being quality. I mean, is it quality because people are engaging in it? Is that it?
Virginia: I bet there’s a lot of comments on those posts, right?
J: There are some comments. And actually I didn’t dig into the comments very much, but he has a huge following so I’m sure he has a lot of traffic. I don’t know. What are your thoughts?
Bruce: First, it would warrant research. But if I understand the site correctly, and I don’t read it so I don’t really study it, but what is the quality of the people who link to him.
V: It’s like the most famous gossip blog. So he would get the best quality in that field that you could think of.
B: If that’s what’s it’s about. I think that if you look at nutrition it has it’s own set of quality standards. If you look at sports it has different quality standards. If you look at art it has different quality standards. I think that it isn’t a single set of measurements. I think that we have to look at every category as it’s own category, and then determine whether or not the entire network – not just his words and not just the comments – but the network that is engaged, is it a quality network?
I think Google has expanded beyond the site. If that’s right, then certainly, a lot of Panda impact can be manipulated, if you will, by only having quality people link to you. The problem is that a lot of the sites that I saw that were penalized in this latest release and prior’s, it isn’t even so much whether the words are arguably worse than everybody else’s words, or whether it’s too cookie-cutter. That wasn’t really where I saw the big difference. I saw the difference as a site-wide analysis: is there deep linking and are people linking to you that are reputable.
V: When I was at the Google Survivor Tips session at SMX East it was all about analyzing the changes from Panda, who survived, who got thrown off the front page. And the interesting thing was that it goes beyond who’s linking, who’s reading what the words are, to how are users interacting with your page. Are they clicking through the result but only to bounce back, and in how long of a time? And it’s looking at is this user going to your site often? He called it a Google bounce, as opposed to a regular bounce, query behavior after the Google bounce? If they come back, what do they refine their search as? It’s taking consideration of your site and people’s interactions as a whole.
B: When I hear things like that, and maybe it’s just me, but I think a lot of interaction on the top sites is by type in. I think a lot of the interaction is in response to an email or some sort of an alert that comes outside of the Google world. I think that Google doesn’t have visibility to all of that information. I think a lot of the top sites, especially brands aren’t even running a Google Analytics type tool. I don’t think Google has visibility to definitively answer the question of bounce.
Now, if I go to search to find it, if I’m stumbling through results and I find a site…
V: That’s what I’m talking about. I’m talking about through search.
B: Right, and even in search, how many searches are there a day and how are they going to sort it out and how are they going to figure out total amount of time on a website? I mean it isn’t enough that somebody click-through.
V: If all of a sudden you’re back to that page, they recognize a bounce back. And even things like how Bing has a new adaptive search, they’re not looking just at previous query, they’re looking at series of queries. I think it would actually fit in really well with your review of Perez Hilton. Because people are saying [with regards to the Panda Updates] avoid things that will annoy your user like ads or that kind of stuff. But that’s a site that’s stuffed full of ads. The content is thin at best sometimes. But, if people are regular gossip readers and they’re always typing in Perez Hilton, if they’re using navigational searches to get there, and they’re leaving lots of comments, something like that, it does seem to support this idea of engagement with a site.
J: And one of the losers [of the most recent Panda Update] was PR Newswire. They have a ton of content and it’s probably written better than Perez Hilton’s but it’s just press releases. So people go there to read the press release and then they go away. Right?
B: In theory. I would think there wouldn’t be as much engagement. But we’ve been talking engagement for a very, very long time. Unless Google suddenly just turned up the dial on engagement as being a major factor recently to determine quality, not to determine value – and maybe it’s more of a value statement than a quality statement. You can end up with a site that doesn’t have much on it but everybody goes there, therefore it’s an important site.
J: Do you remember last week at Search and Social when Ian had that presentation where he did some research – Ian Lurie from Portent Int. – that the level of engagement between a social network, a brand and its users actually affected rankings, not how many people were on there.
B: I think that one thing is true: there’s a lot of factors and Google isn’t going to tell us. And we’re going to have to figure it out one step at a time. I personally think that one of the easiest things for us to do is determine if our network of interconnectivity is of value. I think we can determine whether or not or content is clearly spam-like – I mean we all know spam when we see it – if we don’t think it’s spam, as long as we’re not blind to our own content I think we can make it better. And certainly, there’s no reason why we just can’t improve the overall quality of our site.
No question we all agree on that – Google wants to provide the best results, and we want our sites to be the best for our consumers. But the question of what Google is using to determine what sites are considered of good quality or not warrants thoughtful consideration. And that’s exactly what happened in the chat room during the show today. You weren’t in the WebmasterRadio.fm chat at noon Pacific? Why on earth not? Okay, we forgive you (this time), but look at the collegial discourse and shared experience you were missing out on. (Conversation edited for content, brevity and clarity.)
bryantdunivan: someone finally brought up perez’s increase
JessicaLee: Yeah, bryantdunivan, weird, huh
bryantdunivan: it makes no sense because his stories are recycled from other sources and his benefit is almost like IBLS [inbound links] and off-page [factors] can out weigh a panda bear
Virginia: so you think panda is link related?
bryantdunivan: i think thats where panda loses out. if you are heavily linked in panda hit sites, you go down, because the links are devalued. but with him, theres no content spinning from his stuff, no real syndication, so he’s seen as reputable in that way. quality is made on the site
JessicaLee: But what does the SE have to compare it against?
bryantdunivan: human generated content. i think thats where google panda falls short. it tries to make an algo to determine writing quality, but tries to put all the sites to the same thing. humans don’t write the same, so there will be natural variation for content. thats why we are seeing (still) so many false positives, including perez. but i think somehow your link graph can maybe out weigh the panda
Virginia: ok, *that’s* where links come in
bryantdunivan: i think google is thinking (maybe?) that x people link to him, so regardless of written quality hes a reputable source to x people
MLowery: That’s a valid point. And despite his gossipy, low-quality content, he’s managed to cultivate contacts in a lot of more reputable outlets, like People magazine, etc., so that adds an element of trust, I think.
bryantdunivan: right. pandas all about “better results” so it makes sense in that way too
Do you think we’re on the right track as far as the intended target of Google’s Panda Update and the factors it’s using to determine website quality? Please, join the conversation!
7 Replies to “Panda Updates: Content Quality, Link Profile or User Behavior? ━ SEM Synergy Extras”
If Panda came down to links, why all of this false guidance from Google about making your site trustworthy enough to give your credit card to? I’d link to a site long before I’d give my credit card number to the owner.
Also, I’ve forgotten where I read this, but I thought there was this notion that a “machine” is evaluating your site by “looking” at a set of other sites that have been manually classified as representative of quality — machine learning as it’s called. It’s understandable that running the web through this machine to evaluate a site against the Panda criteria would take a long time hence the multiple weeks between updates. Most of what’s proposed above doesn’t seem like it would be that calculation intensive e.g. link counts, word count, comment count, bounce rates are all easy measures that could be assessed on the fly.
@Gino i am not sure at what extant Google can parametrize ad to text ratio as all the website are not using ads or there are plenty of websites only monetize for advertisement and doing well in search engine !!
I think ad to content ratio is one of the big reasons many people took a hit in the panda updates and how popular content is socially.
I’ve been reading a lot since my site got slapped last week. One of the common elements is that the sites that have been hurt were mostly well established, stable sites. My site, while not interactive (it doesn’t accept comments) did have a very low bounce rate and a very high average time on site per visitor. It is a commercial site, so not easy to get the type of links that Perez Hilton would attract, so yes I do think you are on to something with the theory that the penalty is based on link quality.
It is really interesting discussion, i must agree with bruce’s last point as quality of the website is always matters. I have spent over 6 years in online marketing industry and found that the panda may focuses on:
– Quality of the content (ofcouse everybody knows this)
– Thematic Internal linking (as you did in your website bruce clay)
– User engagement (i am sure all are not using Google analytics so they can not track this for all websites) but they must measure based on number of active users..ratings…comments…feedback and many more things. They may have categorized the websites like ecommerce, PR sites, article sites, blogs etc..and can parametrized the factors accordingly.
– How frequent the content is updated (this is also not new)
Apart from these, as bruce cited, quality and thematic back-links and social sharing must be the essential part of panda updates.