The Knowledge Graph is the name given to the layer of data that sits under Google’s regular algorithms to provide answers to questions. It’s all about linked data, which in turn is all about semantic search — the idea that links and associations can give us better answers to our queries that straight out web search can.

Beyond Keywords

Google’s Matt Cutts tweeted hard about the move, saying that it represented “moving from keywords toward knowledge of real-world entities and their relationships.”

This is the nuts n’bolts of what we’re talking about here. Historically, search has always been about keyword content. No matter how increasingly sophisticated ranking algorithms have become, they only ever play out on a subset of the web that happens to contain the keywords you’ve specified.

So if you’re searching Shane Warne (and for heaven’s sake, please don’t), Google first looks for pages that contain the word Shane and the word Warne, and then it applies its algorithmagic (or whatever it’s called) to web documents that contain both words.

A semantic approach to the search would understand that Shane Warne had a brilliant test career, characterised by a handful of select defining moments, that link him to other entities like Ricky Ponting and Kevin Pietersen. The approach might then link all this together, and serve a set of search results that includes Warne’s test highlights, biographical information, links to multimedia content and, inevitably, something about Liz Hurley.

In other words, it takes a lot of power away from SEO, at least at the keyword level. If a search engine is able to answer a question, it doesn’t need to listen to web pages that are awash with signals suggesting that they can answer the question. Webmasters lose the ability to influence based solely on keyword content, and will need to move firmly in the direction of semantic markup (and I humbly suggest my post from last year on how businesses can get started with schema.org, which is till relevant.)

So despite having too hectic a schedule, I can still bash out a 3am-style rant on this. It’s major news, although its full features don’t yet seem to be available in Australia for us to have a proper play with yet.

It sits on the backbone of their very strong index, which has benefited from years of daily index updates. It’s always been a very data-rich and information-heavy resource, which has been just as much a problem as a bonus: working its data into an informative, actionable set of insights has been the challenge, particularly when presenting solutions and plans to clients.

Step forth Flow Metrics, which looks like this:

The two metrics — Trust Flow and Citation Flow — are based around the amount of equity that’s transferred to the site you’re analysing.

The above is for BBC.co.uk, a fairly massive site. Compare it to the below chart for XLPharmacy.com:

The greater density at the lower ends of both Trust and Citation flows shows a darker colour in the bottom left of the chart, and with a greater spread around the higher ends of both scales, what we see with XLPharmacy.com is a set of backlinks that are generally less authoritative and less “important” than those that the BBC has.

Not Just A Pretty Face?

This is all well and good. But how actionable is this data, really?

We can see, visually, that BBC has more and better quality links than XLPharmacy. But wouldn’t we know that already, from conventional charts? Would we not have the same information, and the same conclusion, from looking at the raw data itself?

There’s probably some very cool numbers behind this, but unless we can have access to the raw data, it’s just a bundle of fancy schmancy graphical representations of stuff that will be difficult for people to understand.

There’s already a big of glazing-over of clients’ eyes when we mention stuff like PageRank, let alone ACRank or Mozrank. What clients want is to rank better — they’re not really fussed about how it’s done.

Whichever way we look at it, it’s good for the industry. MajesticSEO are refining their offering in a way that other players in the space aren’t doing, and if that means that we’re seeing better and more frequent updates to link analysis tools, then that has to be a good thing.

The ad features a surely-fictional teenage documentary film maker who apparently is only able to find an audience for her footage of sharks using Bing, a tactic that would restrict her to less than 4% of Australian web users. Now that’s savvy marketing!

I guess Bing is fully entitled to do whatever it can to chase market share (which is much, much lower here than the 30% or so they regularly pull in the USA.) This ad is fine, even if it’s based around a fairly patronising idea of Australia as a Steve Irwin / Blue Juice crossover (imagine the ad panel meeting: “so they have scary animals, and they all surf… how about sharks?”)

What gets me however, is that Bing isn’t doing “whatever it can to chase market share.” That’s the whole problem. We’re still waiting on an Australian release of Bing Business Portal, which looked awesome when it was released in the states a full twelve months ago.

Neglecting Australian businesses is, of course, bad for Australian businesses. But it’s also bad for the Australian users that are accustomed to relying on search engines to find out info about local businesses.

I’ve ranted on here about how this leads to flimsy results for Australian web searchers, and it’s been noted several times here and everywhere about how their results stack up poorly against Google’s.

Bing, seriously. Running an ad on TV is one thing, but unless you reach out to Australian businesses and give them the support they need to grow on your platform, there will always be a disconnect between what Aussies can find on Bing, and what they can find on Google.

To put it another way: we’re not fourteen year old shark documentary film makers. Give us proper, grown-up solutions to help us market and conduct our businesses online. We might then start taking you a little more seriously.

Michael Martinez over at http://www.seo-theory.com maintains a consistent and quality blog focusing on the more nuts ‘n bolts side of what makes SEO tick. His stuff always interests and entertains, but today there’s a post that totally blows those high standards out of the water.

It’s head and shoulders above anything else I’ve read on its subject matter, which us why I’m inking to it from my blog (which nobody supposedly does any more) as well as sharing it on Twitter, Google+, and Reddit (which we all supposedly do instead).

The link, then? http://www.seo-theory.com/2012/05/08/my-bird-is-a-lark-in-the-park/

It’s a brilliant study on the recent Penguin update, that cuts through the hype and presents a totally clear-cut view of what we know, what we don’t know, what we’ll never know, and what we should really have known all along.

Aside from the subject matter, it’s the method that really stands out for me. Good SEO is all about accurate observation followed by reasonable and measured conclusion, and as the industry grows and becomes, for want of a better word,  “overground”, there’s a tendency for quick-posted articles to explode around the web and for their conclusions to be accepted as truth way too easily.

We need more like this. I strongly urge you to check the article for yourself;I don’t think I’m being too over the top here… So much crappy content gets washed around this industry, that when something this good appears, we owe it to ourselves to make it rise to the surface.

They’re all sites that are tied to legitimate, bricks-n-mortar businesses operating in Australia. They all engaged in strictly white-hat SEO (aside from one who engaged an SEO company that made extensive use of Blog Networks), and their current and historical backlink profiles didn’t show any excessively manipulative anchor text engineering, boilerplate links.

The only thing they had in common? The SEO-Friendly Footer.

Yep, that big mash of text with keywords and locations all zipping users off to relevant pages on a website. Except, most of the time, they aren’t clicked by users. Google views them as chunks of keyword-rich text below-the-fold, that has no practical function for human visitors to a site.

I’m not one for jumping to conclusions, but even early evidence is suggesting that removal of the footer can see rankings recover, at least partially.

Here’s some more direction around this idea:

• Have you had the SEO footer in place for a number of years?
• Has nothing else changed on your site?
• Have your search engine rankings dropped?
• Does your footer take up more than 30% of the vertical screen space on your homepage?
• Does the word count with the footer dwarf the word count without the footer?

If you’re answering yes to those, then it’s time to take action. Don’t get attached to something that people aren’t using, and that’s only bringing you negative movement in the rankings.

If your site’s taken a bit of a stumble recently, and you had one of these footers on your site, I’d love to hear if you see positive change from removing it. Let me know in the comments, or shoot me an email: hello@damionbrown.com

In contrast, then, to what we saw last week, we’ve got a Google release about a forthcoming update, and their head of webspam confirming that it’s out there in the world.

The main points of this update are:

• It’s not an over-optimisation penalty, but an anti-webspam penalty. Google have dropped the idea of over-optimisation as something to be penalised
• Keyword stuffing and unnatural inbound link patterns are specific tactics that the update is targetting
• Spun content appears to be targeted
• Some believe that the placement of links in an article may also be targeted
• Suggestions that patterns of domains linking and being linked to might be targeted

In short, nobody really knows anything. But rankings are being pounded all over the stop. There’s a lot of talk on this Traffic Planet discussion, with multiple reports of sites’ rankings dropping like stones. Some of the screengrabs of tanked rankings are sobering.

One comment in that thread relates to the SERP for the word “viagra”, where the leading result viagra.com is nowhere to be seen. It looks as though Google’s update saw loads of links to viagra.com with the anchor text “viagra” and penalised the site. Ouch.

Moreover, here’s the screengrab of a search on [viagra] showing a very fuzzy assortment of results:

Like I said, ouch. Wikipedia ranking in number one; but a hacked site warning in positions two and five. Really? What would you rather Google was demoting in the rankings — SEO or malware?

In amidst the results, there’s an entry for the Melbourne digital agency Paste. Somehow, these guys are ranking on the first page for an incredibly competitive keyword; my first thought was that they might have had a rival sponge a load of links to their site with “viagra” in the anchor text, but neither MajesticSEO nor Ahrefs could find any.

Still, Google thinks that there’s some relevance in there somewhere, as this site: search shows:

Yikes. And still more Yikes is the bottom result there, for the Australian Multicultural Foundation at http://amf.net.au. Their site has nothing to do with viagra, but has some spammy anchor text pointing to it:

Argh. While I accept that this is all in the middle of the algorithm rolling out, and we can therefore expect to see some weird results, it’s hard to understand exactly what is going on here, particularly in the context of improving search results and rewarding real, honest webmasters.

On the one hand, we’re being told not to manipulate with links or other dodgy tactics; and on the other hand, we’re seeing that it’s possible to outrank the manufacturer of viagra just with some crappy anchor text links.

I’ll revisit this in the coming days, as it’s definitely one for the industry to watch. I’m running rank checks all day and I’m bracing for a major upset. If you have any to report, or any other observations on weird search result behaviour, do let me know!

While I didn’t see any negative movement on sites I look after, others in Australia did, and combined with the chatter on various forums around the interwebs, it looked for a while as though weird things were happening.

Earlier this morning, Matt Cutts posted the following on Google+

I saw a recent post where several sites were asking about their search rankings. The short explanation is that it turns out that our classifier for parked domains was reading from a couple files which mistakenly were empty. As a result, we classified some sites as parked when they weren’t.

I apologize for this; it looks like the issue is fixed now, and we’ll look into how to prevent this from happening again.

So there you have it: it was a mistake, a misconfiguration on Google’s side. Or so they say.

Of course, if it really was the over-optimisation penalty and it was running rampant on many, many sites causing damage to legitimate sites’ rankings, you’d expect Google to blame it on a “tiny inadvertent error”. Tinfoil hats at the ready, we might also say that this sort of error isn’t the sort of thing that Google’s layers of PHDs would make, and not the sort of thing that would get into the wild in the context of Google’s progressive and extensive testing of new features before launch.

And in any case, I don’t recall seeing parked domains anywhere in search results in the last few months. Why the need to act on them now?

This is one of those things that we’re going to have to take at face value I guess; it’s good that Cutts has responded, and it’s good that folk that lost their ranking yesterday might be getting it back today.

What leaves me with an uncomfortable taste in my mouth is just how one change by one company — whether erroneous or misjudged — could spiral so many people’s livelihoods down the lavatory.

Barry Schwartz over at SE Round Table has a good summary of what’s going on at the Google Webmaster Forums, and also Webmaster World forums.

I’m not seeing anything on my sites yet (in Australia we often get a day or two’s delay before what happens in the US rolls out to local datacenters), but from what people are saying on the usual places there’s definitely something up.

It might be connected with the recently-mooted “Over-Optimisation Penalty” but there’s a lot of initial evidence suggesting it’s connected to specific hosting providers, according to folk on Warrior Forum and Traffic Planet.

Thing is, whenever something like this happens the forums will burn white hot with speculation, and already there’s a heap of dross to wade through before actually finding reliable and methodical info from contributors. Possibly the closest thing to an empirical study at the moment is this thread on Warrior Forum but even then, one wonders how accurate the insight really is.

At this stage then, it’s too soon to say what changes are, and what their effect might be. However I’m concerned that white-hat and fully compliant sites might be getting hammered here, if indeed it’s the over-optimisation penalty or a penalty connected with cheap hosting.

One of the great things about the web has always been the absolute democratisation of everything — if you want to be am online publisher, then with ten bucks to a shared hosting company and a bit of WordPress, you’re a publisher. Part and parcel of the lifeblood of the internet is the ability for a site to exist like that; a quickly setup, cheaply hosted site has just as much of a right to be there as any other site.

If Google really is taking a sideswipe at these sorts of websites, it could be a hugely sad day for the internet.

Although new to Australian shores, Econsultancy have been operating in the UK since 1999, and have branches in Dubai, New York and Singapore with Sydney the most recent addition.

The courses I’ll be conducting for Econsultancy are Digital Content Strategy, SEO Marketing and Web Measurement & Analytics which are offered in Sydney, Melbourne and Perth. I’ll also be running their two-day Fast Track Digital Marketing course in Melbourne.

I’m excited to be delivering these courses under such a prestigious banner; Econsultancy’s research publications have been a defacto authority resource for both my own training courses, as well as materials I’ve delivered on behalf of agencies. For years they’ve been a consistent voice across an often muddled industry, offering high-value insight to their members and course attendees.

Being invited to work with such an industry authority is extremely humbling. It gives me the chance to lift my game to the next level, working solidly on the stuff that resonates with me best: sharing knowledge, concepts, and occasional jokey asides, delivering content I believe in, and helping digital people reach digital goals.

This also means some changes elsewhere. I’m moving away from agency work, having had an awesome run with Komosion over the last two and a half years. I’m moving into funky new offices (in Brunswick East, and at which you’d be most welcome for a catchup) which I’ll be sharing with the Content strategists Weaveweb.

You can find out more about Econsultancy’s Australian launch here, and view the full programme of public training courses here.

The upshot, today, is that it’s now no longer okay to bid on your competitors’ business names. Previously, courts had openly said that doing so did not amount to any sort of infringement, but now they’ve changed their tune:

The user asks a question of Google and obtains Google’s response. Several features of the overall process indicate that Google engages in misleading conduct.

(My emphasis added for dramatic effect.) Read the ACCC’s full release here.

It’s all very early to say how this will pan out. Based on today’s ruling, advertisers should remove any keywords in their campaigns that are references to competitors (and this may or may not include misspellings.)

Several advertisers, whose competitors are bidding on their brand names, will also no doubt be anxious to get some sort of recourse on their brand equity by demanding that Google block competitor ads against their name.

Interestingly, the Court appears to place all the blame on Google, and not on the individual advertisers:

Google did not merely repeat or pass on a statement by the advertiser: what is displayed in response to the user’s search query is not the equivalent of Google saying here is a statement by an advertiser which is passed on for what it is worth.

It’s implicit in this part of the ruling that Google somehow creates a distortion between brand searched for and brand presented, and that this distortion is misleading.

While this is good news for advertisers represented by the ACCC, I can’t see any change happening on this in the immediate term. Google will appeal, the ACCC will appeal back, and not a great deal will change overnight.

What’s most significant is the idea that there’s a bait-and-switch going on that’s deliberately misleading, and that Google are the architects of this. Let the endless cycle of appeals and counter-appeals commence…