Welcome to SMX Advanced! Are you ready for some serious liveblogging action? I hope so. Because that’s what we’re about to unleash on the blog over the next few days. Watch out! And oh, you wanna know what sounds like a good idea but actually isn’t? Walking fifteen minutes in the misty Seattle wind to get to the expo hall this morning. If you’re looking for me today, I’ll be the sweaty, frizzy disheveled mess. Feel free to say hello. Or run away. I won’t judge you. I judge myself.
ANYWAY! We’re kicking this off with a session on the new periodic table of SEO. If you haven’t seen it, well, click that link over there and you will. Up on stage we have Danny Sullivan moderating Matthew Brown, Duane Forrester (Q+A only), Jeff MacGurn, and Rand Fishkin.
It’s the first session of the show so Danny spends some time going over the housekeeping stuff. He explains how people are supposed to tweet the sessions, how it would costs eight gazillion dollars to outfit the entire room with power strips and lots of other housekeeping stuff. You’re not here so that stuff probably doesn’t concern you. Luckily for you it’s time to get to the search stuff.
Danny talks about the new SEO periodic table that Search Engine Land recently published. He says since there’s a new one, there has to be an old one. To that end he spends some time showing mock ups of what the periodic table would have looked like over the years starting back in 1997. When I was 15. Okay then.
Up first is Jeff MacGurn.
Jeff asks – If fe= iron, then does female = iron male. Oh boy. Chemistry jokes at 9am. That’s not a good sign. Jeff ends the banter and comments that even the real periodic elements table has recently been expanded. See, even that one isn’t set in stone yet. He thinks the search engine periodic table is expanding even faster than the real one. I say that’s because we were a recession-proof career. We’re moving faster than everyone. Yep.
We’re going to talk about Correlation vs Causation. My brain hurts already. Jeff reminds us that causes underlying the correlation, if any, may be indirect and unknown and high correlations also overlap with identify relationships, where no casual process exists. Establishing a correlation between two variables is not sufficient condition to establish a casual relationship. That said, here we go.
What factors correlate with better rankings? Let’s see.
Technical SEO Factors
- Page Size: They found it wasn’t highly correlated with ranking well.
- URL Character Length: Low correlation.
- Flash Navigation: Low correlation
- Session IDs: Strong negative correlation
- Dynamic Parameters in URL: Low Correlation
- Proximity of Page to Root Directory: Low Correlation
- Page Load Time: Strong correlation to ranking well
- Keyword Emphasis: Low correlation with rankings
- Keyword in Title Tag – Strong correlation – stronger on Yahoo
- Keyword in Meta Description: Low correlation with rankings
- Keyword in H1 Tag: Strong correlation overall
- Keyword in H2 Tag: strong correlation overall
- Keyword in Image Alt Text: Low correlation
- Keyword in URL: Strong correlation overall
A link hub is a site that links to multiple authorities. An authority is a site with links from multiple hubs. With all the sites that rank for a keyword, they looked at the top inbound links and looked at the domains in common. They looked for correlation between ranking and having a link from a certain site. Did sites ranked in the top have common links?
- Internal Link count: Little correlation
- External Link Count: Marginally better than internal link count
- Keyword in Anchor: Waning, but still slight correlation.
- Hub Links: Very strong. One of the strongest factors of all the examined factors in the study.
Efficient Market Theory: One cannot consistently achieve returns in excess of average market returns on a risk-adjusted basis, given the information publicly available at the time the investment is made. The landscape changes without any of the fundamentals changing. He thinks the stock market relates a lot to search engine optimization. Things naturally change. They’re not manual changes by Google. When everything looks the same, then even the slightest thing looks huge. Having a keyword in your Title tag is no longer ranking factor for [smartphones] because EVERYONE is using it.
Don’t look for the “secret SEO factors of the day”. Look for the differentiating factors in your landscape. Ensure your are optimizing for all of the factors your competitors are optimizing for and more. Even in highly competitive landscapes there are differentiating factors:
- Links from Hubs
- Page Load Time
- Keyword in ULRL
And he’s done. Wow. Usually conferences start slow and then dig into the meatier stuff? Yea, SMX is throwing you right in. This is Advanced, kids. :)
Up next is Rand Fishkin.
Rand apologizes because he’s going to talk about correlation, too. Danny rubs his arm and says he has nothing to apologize for because it’s awesome. Aw, thanks Dad! Rand is going to talk about the results from the new edition of the Search Engine Ranking Factors report which was just published on the SEOmoz blog. Make sure you check that out when your brain is numb from SMX coverage.
Are you ready for some data?
Back in 2009, 65 percent of respondents thought link factors was the most important part of the pie. In 2011, it fell to 45 percent of the pie.
What do SEOs believe will happen with Google’s Use of Ranking Features in the Future? Exact keyword match domains will fall, anchor text in external links will fall, prominence of ads vs content will increase, usage data will increase, social signals will increase, perceived valueness will increase. This was a huge change in the balance of that pie chart.
Comparing page and domain-level link data: Domain-level data is shocking to him. Suggests page level + domain-level link signals have relatively similar weighting, just as voters predicted.
Have Exact Match domains lost their luster? Exact match domains have both fallen considerably in the past 10 months.
Is Google Evil? Google has said that linking externally is good, slow pages are bad, and using Google services won’t give any special benefits. His data supports those statements. Drat. Sorry, Google conspirators. These are not the droids you’re looking for.
For Twitters, voters felt authority matters more, while for Facebook, its raw quantity (could be because GG doesn’t have as much access to FB graph data). The raw number of Facebook shares is the single highest correlated metric with high Google rankings. HUGE! There is clearly something going on there. If you thought that’s because just not that much stuff is shared on your Facebook, you’re wrong. FB Share data was present for 61 percent of pages in the top 30 results. Duuuude!
Are pages ranking well because of links and social metrics are simply good predictors for linking activity? Nope.
Don’t misuse or misattribute correlation data. Think of correlation data as a way of seeing features of sites that rank well, rather than a way of seeing what metrics search engines are actually measuring and counting. [In others, don't lose your panties, folks.]
Please do Rand and this survey justice by reading the actual support. Liveblogging Rand and data first thing in the morning is not the easiest of maneuvers. :)
Next up is Matthew Brown.
He throws up a picture of a GTR (that’s like, a car or something) and shows how each of its parts are like a part of your Web site. That’s really all I can tell you, other than that Matthew is super funny.
Some factors like social factors, trust factors, etc, may affect rankings but they really change up the search results. The search results you see can be vastly different when brand, social and local all get together. Brand lives in the middle of trust and authority. When Google talks about brand, they don’t say brand. They use words like “trust” and “authority”. When Bing is talking about it, they’re talking about an online version of what’s happening in the real world. You may not be Google or Coke, but to Google, SEObook is a brand. Search Engine Land is a brand. We’re not just talking about Fortune 500 companies.
You type in [microwave oven] and you see nothing but retailers and brands, even if you didn’t type in [buy microwave oven].
How do engines find brand signals? From clickstream data. People type your brand name into their toolbar or address bar, associate your brand with related keywords, build non-search brand traffic to get those citations.
Great, but how do I build my brand and signals on MY site? ThinkGeek does a good job of this. They private-label their things as Think Geek even if they don’t make it in their laws and they use object-based social signals. You like the product, not just ThinkGeek. Identify your site’s social authority hubs. Which sites work for your types of content? Look where your audience is, this will tell you which social buttons to put on your site. Do you really want to lose your LinkedIn button for a +1 button?
Merge local modifiers with keyword sets using available tools like MergeWords.
First Move Status – takes chances and makes guesses. The ones that pay off will be worth it. Don’t just wait around for SEOmoz’s data. You’re behind the curve.
Duane: Don’t stray too far away from the core concept that content is King.
And we are done. Wow. That may have been the best “kick off” session I’ve experienced. We’ll be back in a bit with more. Stay tuned! :)