Posted by simonpenson
This post was originally in YouMoz, and was promoted to the main blog because it provides great value and interest to our community. The author’s views are entirely his or her own and may not reflect the views of SEOmoz, Inc.
Rand’s recent WBF about co occurrence was a real wake up call for those still transfixed with link building practices of old. While anchor text based links may still have some effect there is little arguing the fact that the factor’s importance is dwindling. In its place are things like social signals, link age and most importantly a growing reliance on relevancy and how that is deciphered.
For those that haven’t read it yet, I wrote a piece a few weeks ago about what I felt recent Google penalties are really trying to solve. My view is that Google is really trying to clear up the link graph and with it valueless links so that it can clearly understand relevance and associations again.
It’s something that web ‘creator’ Tim Berners Lee first wrote about back in 2006 in this magazine article and Google has been talking about ever since, ploughing lots of cash into acquisitions to get it there.
So, why invest so much time, effort a resource into such a game-changing project? Quite simply because its existing keyword based info retrieval model is in danger of being bettered by semantic search engines.
To understand why this is the case we must first delve into semantics and why it changes the way search engines work.
Put simply Google wants to better map the associations between content so it can deliver what it believes will be a more personalized and effective result.
The nirvana for the engineers working on the project is to map the associations between types of content with an understanding of the intent of the user when typing in a query.
So, let’s say I type in ‘what’s the weather today?’ At present Google might know where I am but would find it difficult to associate other content to that query. The reason I’m searching for it may well be because I want to know whether I can BBQ, or complete that landscaping project I’ve been researching online.
Google can improve its results by ‘knowing’ why I’m searching for the weather so it can also throw up food offers or home improvement guides.
It can only do this if its data set is clean, and right now there are far too many spammy links muddying the waters; hence why Penguin came in to begin addressing that issue.
Why is Relevance important?
It is quite clear to see why relevance is therefore important and why search engines may reward those helping them out or working to this new system.
How does Google Measure this?
Clearly we are still a long way from a purely semantic engine and Google may never get to that point. The important thing is that they are certainly motivated to make much more of relevance to diversify search results.
As a search marketer your first thought will undoubtedly be ‘how can I ensure I work in a way to take advantage of this change?’ The answer to that question begins with an understanding of some of the patents Google currently holds that may help it do this.
(Hat tip to Bill Slawski and Dan Thies on some of the below)
Google had a busy time of it back in 2003, bringing in Taher H. Haveliwala, the genius PHD student behind a new way of applying topical relevance to the company’s faltering PageRank model.
His research around applying greater relevance to links from topical pages as graded by their newly acquired Applied Semantics’ CIRCA Technology meant that they could start developing ways of measuring relevance.
Reasonable Surfer Model
This theory is then taken another step further by applying differing weights to different links on that same page based on their ‘likeliness to be clicked on’. The more likely it is they will be ‘used’ the more authority handed to them. Everything from font size to position and even colour are taken into account in this computation.
Phrase Based Indexing
To further complicate the picture Google then also looks at co-occurrence of words and phrases on a page to work out their ‘meaning’. If you take the phrase ‘hair of the dog’ for instance, Google needs a way to understand its meaning.
To do that it will look at other pages that mention that same phrase to see what else they mention. If they also mention things like ‘drink’ and ‘the morning after the night before’, for instance it will understand that it, and the page it is linking to is talking about a drink to offset the impact of a heavy night out it will assign more authority to that link as it is extremely relevant.
Had it talked about dog hair’ it would be less relevant and therefore a less valuable link.
This is a key development as it is most likely responsible for a lot of the penalties we are currently seeing as a result of spammy link building practices. To stop a page ranking Google can simply remove the connection between the page and any particular term in its index.
It also throws up some interesting opportunity and new ways of working for those looking at how to optimize sites, and we’ll come onto that a little later.
While this wasn’t a direct algorithm patent or algo change Google’s purchase of Metaweb, an open source entity database of people, places, things, powered the development of ‘Knowledge Graph’ and fast-tracked its move to add more diversity and ‘user intent understanding’ to search results.
It’s addition allows Google to better understand associations between pages based on their real life connections, not just how they are linked to.
How can you develop a Semantic Strategy?
Knowing all of the above is useless with some actionable ‘next steps’ in terms of how it affects your own search marketing efforts.
So let’s look at some of the ways in which this kind of knowledge has helped me structure our own on and off page process at Zazzle.
The first thing you must work on when considering your off page plan of attack to proactively improve your own relevance profile is to understand what is considered ‘relevant’ to you, and how, in a semantic world. Below is an example of related words to ‘content marketing’ and how they are connected:
The good news is that there is no need to use guesswork here. Tools exist to take the hard work out of the process and a few of the best are listed below:
http://ctrl-search.com/blog/ – this is a great tool to enrich your on page content. Effectively semantically optimizing your own site. By pasting in snippets of your post the engine finds semantically associated images and other content for you to link out to and add.
http://lsikeywords.com/ – a few great blog posts have been created recently around the subject of LSI, or Latent Semantic Indexing, including this one linked to on our own blog.
We wrote about it, as it’s a key part of our own outreach process now. For every piece of work we do we will use a tool like this to ensure we stay relevant.
LSI Keywords is one of a handful of tools that will present a list of semantically relevant keywords and phrases for you to widen your outreach approach.
http://bottlenose.com/ – is a tool I have mentioned before here and its great for a multitude of things, especially big data led content cu ration. One of its ‘tools’ however is great for understanding degrees of relevance separation. Once you type in a keyword you have the option to scroll through a number of different tools but the one that we want to use for this is the Sonar+. It visually maps real time semantic relationships between concepts based on Twittersphere sharing and other big data.
Google Semantic Operator – not a tool per se but a really useful operator to help define semantic keyword relationships. By adding the ~ Tilde symbol when searching Google for your key phrase (e.g.: ~travel) you will see other words that Google has mapped against that word, such as Hotels, Flights, Holiday, Tours.
http://ubersuggest.org/ although it is not officially a semantic tool ubersuggest is built on Google predictive search engine and so by default it delivers semantically relevant searches, which makes it great for building outreach keyword lists.
All of the above tools give the user the ability to create a keyword-based map of where to outreach to if links are your project aim.
Building the Outreach Plan
Once you have a view on where you want to outreach too then the next step is to construct a plan to do that.
The next stage is to create a time-based project plan to detail each and every step of the process. This is extremely important when carrying out outreach, as it can be very easy to get distracted and pulled sideways and out of your defined semantically relevant zone.
We use a simple excel table to plan this and below you can see an example based on a two week outreach campaign for a fitness brand.
As you can see we have planned time days on specific areas to ensure we cover off as much semantically relevant opportunity as possible. Into this plan we would then add outreach contacts and note what communication we have had with each.
How to outreach well has been covered in detail by posts like this and this and this post is already far too long to delve into this right now but one tip that must be followed is to be as exhaustive as possible in exploring each avenue. Think Face to Face, Phone, Twitter and finally email in terms of contact medium hierarchy as the further you get down the list the lower the conversion to placement will be.
Where things get VERY interesting in a semantically driven project, certainly in terms of off page activity, is when you begin to consider what the real value of that work is; the metrics you’ll be monitoring as your KPIs for the campaign.
Posting without links
Posting content without the need to obtain a link may seem like an insane proposition, especially if you are measuring success by ranking and search engine visibility metrics, but that may not be the case.
Real marketing is not about links. It’s about connecting your brand or business to people with similar interests and beliefs. Links are simply a mechanism that drives Google visibility to get you in front of more of those people more often.
Google understands that while its entire business is built on links it really needs to get away from that model and motivate us to act like above-the-line marketers. And that is where Lexical co-occurrence comes in.
In simple terms however it is a way of ranking websites and pages not on inbound links but by how many times they are MENTIONED in close proximity to key phrases.
That’s game changing.
If Google can work out what you are relevant for not by looking at dumb anchor text but what people write about you and what other phrases you regularly appear close to it changes the way you outreach and market your content.
Imagine being able to outreach awesome content without having to look for links. Simply make people aware of what you are doing and get them to talk about you. It’s how it should be and it would have a profound effect on the type of content you might produce and brand-marketing activity you might pursue. Expect PR stunts galore!
On Page Semantic Optimisation
Another key element of semantic ‘link building’ is to build out your relevance to widen the scope of what you are ‘about’. If Google is looking to diversify results then the more words and phrases you can associate yourself with the better.
This means expanding your repertoire. Writing more about those peripheral semantic phrases that are still on brand but may help you rank for a greater number of related searches.
In many ways this is not dissimilar to how any good content strategy should be constructed anyway but below are a few simple reminders and additional points to consider when designing content for a semantic engine:
- Proximity > keyword mentions, as we know are useful for helping keyword-based retrieval engines like Google work out what you are about. To improve this further augments the phrase with synonyms, as these are strong semantic signals that you are relevant for a cluster of phrases. Try and then align URL, H1, bold and italic text etc. to ensure continuity, as you normally would to strengthen the page. How close keywords are to key modifiers, and to other links, and the higher up in the code they are the better.
- Relevant Shingles and Keyword Co-Occurrence > Use the aforementioned tools and ensure you are adding in those co-occurred phrases within the copy of the page.
- Synonym keyword linking > Ensure you link from semantically relevant keywords back to key landing pages for bigger terms to create a strong semantic theme. So link internally using ‘holiday’, ‘hotels’ and other terms back to a key ‘travel’ landing page.
- Linking out > again this is not new but by choosing ‘Authority’ or ‘Expert’ documents within the niche, as described by the Hilltop Algorithm. This means looking for high authority sites already ranking for the term you want to be relevant for.
We have covered a lot of ground in this lengthy piece. My hope is that it gives a solid overview of where Google, and other key search engines, are heading. More importantly is gives some actionable tips and suggestions for you to begin implementing now to ensure your site benefits from these forthcoming changes.
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!