The first SEO session of the day and we’re on to the question that will never die so long as search engines exist, ranking factors. A great lineup for this one, including blackhat legend deMib (who I’m assuming will give us the real deal, non-Google-approved shizzle on whats helping him rank); Rand Fishkin who will no doubt be referring to the SEO Ranking Factors survey (although I hope it goes beyond the frankly ‘bored now’ question mentioned in the conference blurb, “how much does that H1 tag really matter”); Rob Kerry who I know will be interesting and talk from his extensive experience of working on some of the most competitive niches out there; Will Critchlow who’s moderating the Q&A but I’m sure will pipe in with some interesting observations.

Rand Fishkin – SEOmoz

Rand starts by saying he’ll touch on the ranking factors survey but has got a lot more to share – good stuff! He mentions Bill Slawski’s post about the ‘reasonable surfer’ patent granted recently, and stresses the importance of reading that.  Whereas SEOs have been using the original PageRank Random Surfer model as the basis for our work, which says that every link on a page has equal weight in terms of PR passed.  However the Reasonable Surfer model takes into account how likely a link is going to be clicked in attributing weight.  According to Rand his ranking tests, links higher up on the page do pass more weight than those lower down.  More generally it seems to be the case that links that are more prominent also correlate to better rankings.  I think it’s easy to get carried away with this – I still believe on the whole a link is a link and just because a link is tucked away within a paragraph at the bottom of the page it doesn’t mean it’s less important than the massive flashing ‘click here’ banner at the top.  But it certainly is interesting that Google *might* be considering this as a factor.

On to Twitter data influencing SERPs, a really important issue IMO.  According to his observations, links from twitter certainly do affect ‘Query Deserves Freshness’ (QDF) results, given that both Google and Bing have a data deal with twitter now – in spite of the ‘nofollows’.  Rand mentioned this at ProSEO last year but I think the size of the twitter link graph means that it will continue to be important to investigate/discuss.

Rand then moves to how the social graph and particularly Facebook ‘like’ data might affect SERPS.  At present, as we all know, it’s not taken into consideration in ranking models. But he speculates that this could happen, potentially from Bing who have an investment in Facebook.

On to some correlation data about the importance of putting a keyword at the front of your title tags.  The stats in his correlation analysis clearly show that it has a big impact.  I think most SEOs would agree that this is the case from experience.  ”A while ago some dude said it doesnt matter, just put your brand at the beginning – I think his name was ‘Rand Fishkin’” However I still think there are cases for it in terms of clickthrough rates for big brands.

Ahem. We come to H1 tags. Yep, they’re not as important as people think. That is all.

He then touches on the dreaded Keyword Density.  The data shows that while putting keywords on your page helps rankings (duh), if you look at keyword density % as a fixed causation factor you are an idiot. (or words to that effect)

Rand then shows up some data related to “Latent Dirichlet Allocation” which is about analysing which phrases are related and relevant to others.  Um… to me it seems a bit like “Pantene shampoo science” but that’s me being flippant and dismissive… one to investigate in future I think

[Update: Rand's slide deck now added! Note how I got you to read my notes first - hehehe...]
SEO Ranking Factors 2010 SMX London

Rob Kerry – Ayima

Rob starts by talking about the Death of the External 301.  It seems in January Google clamped down on smart affiliates using cross-site 301s on affiliate links and stopped them passing value.  He uses the example of Top10 Broadband which suffered but recovered.  The key message is, if you’re moving sites, don’t rely on 301s to save all your link juice – build up the link profile to the new domain with fresh linkbuilding,

Rob then talks about how Google also seemed to improve their link filtering back in January – a lot of sites’ links were devalued where there was not as much of a natural link profile – overoptimisation of links and particularly over-focusing on homepage.  You need to make sure the signal:noise ratio in your link profile is good, and make sure there is a good proportion of brand/natural links across the site.  ”Deep pages need to justify their existence by having external links to them” – fully agree with this comment!  So often people only linkbuild to their homepage and they dont realise how much they are losing out – deep links rule the SERPs, it’s a fact.

Rob then talks about the “MayDay update” – take this with a pinch of salt! People like to come up with new names for so-called updates – in his opinion this could just be an example of what he has talked about above, ie lack of deep links affecting long tail rankings.  However he does say you can’t rely on domain authority alone any more. You need to have specific content and pages related to specific terms with links coming into them.  This may be the issue for people.

Mikkel deMib Svensson – deMib.com

Great opening slide – “Get Rid of the Crap on Your Site” – LOL.  There are two common types of crap on a site: ‘code junkyards’ from lazy or incompetent developers who leave opportunities for hackers; and malware.  He talks about how ignoring cleaning /securing your code can completely slam your rankings.

So how do you secure your website?  Demib goes through a few steps you can take but sadly this particular liveblogger completely missed them – I will try to update later! (sorry)

Site speed has always been a factor – the faster your server responds, the faster the engines crawl your site.  Completely agree with this – check out my previous post about site speed and SEO…

Does the ‘quality’ of code (ie validation) matter for rankings? Demib says, No, emphatically.  Google can’t ignore a site just because it doesn’t validate – they don’t even validate themselves.

He then moves to the importance of clean code – don’t give Google unnecessary crap to wade/crawl through before getting to your content:

• .net viewstates are a total pile of crap! Most people don’t realise how much this can harm you (I totally agree).  Demib has printed out a typical viewstate from a site’s code and stuck the pages together – it’s about 10ft tall in total! But there are ways of keeping this to a minimum – see this post for example.

(while this is going on there’s some discussion about .net viewstates and Mark Cook comes out with a lightning quick solution of how they hide viewstates from googlebot (a bit of ‘whitehat cloaking’) on the Further site – great stuff and I love how twitter responded to SMX in realtime there!).

• Keep all JS / CSS external – we all know this but do we do it? good kick up the arse here
• Minifiy your code – it can make a big difference
• Get rid of HTML comments – they just bloat your page
• Remove those ridiculous, unneccessary meta tags!

Q&A Session

• Question about the ‘MayDay’ update as well as Google’s new interface.  I missed Rand’s answer, but Rob Kerry makes the point that it’s not necessarily a negative – a change in the algorithm is bad for some but an opportunity for others.  It comes back to the point that you cannot rely on domain authority alone any more – need niche content, and perhaps this helps the specialist niche sites.  (Interesting – are we seeing a corollary to the brand update taking place?).  Demib warns of the dangers of  always chasing the latest update or tweak because when it changes again you could lose out. Stick to the fundamentals.  Regarding Google’s new layout Demib reckons this might be a win for Adwords – essentially because people are unfamiliar with the new organic results layout they tend to click on sponsored results more.  Really interesting theory and I wouldn’t rule it out.
• A question about UK vs US SERPS. Rob says that in google.com it’s homepages that tend to rank due to the sheer authority of US based sites, tends to be more deeper pages in the UK.
• Regarding QDF and realtime search.  Demib makes the point that a lot of smaller language / country results have much smaller datasets and you don’t see this happening as much as you do in .com.  Rob gives an example of the “credit cards” SERP in the UK where you get a news results onebox with US news stories!
• Question about 301s and canonicals for Rob Kerry.  Are canonicals now the replacement for using 301s? Well the most important thing is to try and repoint old links rather than using 301s.  However failing that canonical does seem to be the solution of choice at the moment.  It’s no coincidence that Google launched cross-domain canonical tags in Dec 2009, and then devalued external 301s in January.  For Rob this is an example of Google taking away one solution and replacing it with another, Google-approved one.  He has noticed it working and in fact a lot of sites are getting link juice via canonicals that don’t really deserve it.  Demib doesn’t really like canonicals – he’s seen Google treat correctly used canonicals wrongly.  However Rand has seen canonicals
• Question regarding future trends…responses below
• Rand: its insane that we tweet so much (good) content, but don’t use that content for SEO value.  We should be capturing this as webmasters.
• Rob: If you value your short tail traffic, target links to the root of your domain (and vary the anchors a lot / include lots of Brand links) and don’t change your domains.  If you value long tail, focus on deeper pages and deep linking.
• Demib talks about how so many sites are missing a trick by having stale pages / content and never updating it.  Make use of UGC / pull in (or even scrape!) content from other sites, use dynamic elements – whatever you can to get your pages refreshed.
• Rob Kerry: at a past conference in the “Give It Up” session I mentioned the idea of bringing in affiliate links to your own domain and then using rel=canonical to your money pages – ie pretend it’s like a session id.  A few months later at SES a big site said they had done this and had excellent results without having done much link dev.
• Good Q about rel=canonical: How important is it that two pages are identical in content? How far can you go with variations in content?  I believe the answer Rand gave was that it doesnt seem to matter too much – Google really strongly values this (pls correct me if I’m wrong on this)

The end of an excellent session, loads of useful info.  I think Rob Kerry’s thoughts were the highlight (as I predicted), I definitely need to come back to them!

liveblogging · ranking factors · smx

<< Leveraging Digital Assets For SEO: SMX London 2010 Recap

Keynote: Barak Berkowitz, MD of Wolfram Alpha – SMX London 2010 >>