Links are dead – long live links!

Links for SEO are already dead. A lot of people haven’t realised that yet.

More specifically, I’m talking about links for SEO in Google - the use of link building to increase rankings on Google.

I don’t make that statement lightly – I run a company whose core business model has been link building services.

So why are links dead?

Five years ago links could be taken as a non-biased third-party recommendation of other websites. Google built the foundations of its search business on that principle.

Nowdays, links are no longer trustworthy in themselves – certainly according to Google. There’s a thriving industry in providing links which are essentially for SEO purposes – to help websites rank for their targeted keywords.

The links themselves are not necessarily bad, but Google has always made it perfectly clear that it doesn’t like third-parties having undue influence on Google’s search results.

Google wants to organise the world’s information – not the link builders.

The death of the link is already in process, and actually has been for years.

Florida

In November 2003 I made first public call on an update that would live in infamy. A few days after I noticed it, the update was given the name “Florida”.

What Florida did was apply some major algorithm changes – which almost certainly included a re-evaluation of links (not least by IP ranges), a focus on authority issues, and some semantic processing (stemming, etc) thrown in.

Google gave plenty of warning about that update coming – it published papers years earlier, such as Hilltop and LocalRank, which clearly showed that Google was interested in determining an authoritative hierarchy to the web, along with devaluation of links within close IP ranges.

What we learned then was simple – forewarned is forearmed. Nowadays, Google patent applications can be held as potential glimpses of the future.

From some of the more recent Google patent applications, there is a clear message – links alone cannot be trusted.

Sandboxing

Sandboxing has become a euphemism for “crap site, can’t rank”. Past the arguments, there is a very real process at work.

The Google Sandbox became manifest as way to delay the impact of text link advertising.

It meant you could no longer drop a few hundred thousand links and expect to rank for your anchor text quickly – you now had to wait those links through a probationary period.

Sandboxing began as a simple three-month delay on your anchor text hitting.

Then Google saw it worked, and extended it – now the Google Sandbox has become a complex entity in it’s own right, with perhaps as many as a dozen key parameters involved, not least the age of the domain being linked to.

The Google Sandbox is currently being used to try and limit manipulation of the Google index through link building techniques.

Clickstream

If Google is no longer able to use links as an indicator of value given to websites, what can Google use?

The answer is Clickstream.

In short, user behaviour that Google can log, measure, process, and data mine for human behaviour patterns and associations.

Couple that with Google’s range of other services, from Gmail to Google Analytics, and Google becomes empowered with a wealth of user behaviour data.

Google can also see which sites deliver a poorer user experience - for example, a website whose description does not invite clickthroughs, where users do not remains long on the site, and is rarely bookmarked, is flagging itself as a low quality website for human users.

Clickstream indicates quality

The insidious part is that Clickstream is very difficult to manipulate – the most effective way to manipulate clickstream data is to sweat it out and build the best quality websites.

In short, you have to focus on building for human users – and they have to like it enough to want to visit it, bookmark it, and re-visit it, in numbers that Google determines is significant.

Truly, the fabled democracy of the web comes not from the structure of the internet, but the usage of the internet – votes counted not in what websites tell one another, but how humans use those websites.

With that in mind, the SEO power tools of the future aren’t built around the link – but instead built around human interaction.

Building for Clickstream

How do you build for clickstream? Here are a few simple recommendations:

Clickstream vs Links

Google is evolving and Clickstream is a clear future that Google is evolving towards.

Are Google testing any Clickstream data now for ranking purposes? Almost certainly. Are Google likely to ignore links entirely for SEO? Not really.

You can be sure that with the amount of user behaviour that Google have already collected through their myriad of applications, that they are already experimenting with injecting small influences into the ranking process.

It’s unlikely to be a major compenent yet as compared to links - but there is an ongoing shift in the balance of power between them.

Issues such as the Google Sandbox are simply to help work as gatekeepers on containing link influences - as Clickstream becomes more and more mainstream, and increases its influence on Google search results, so will the reliance on raw links and the need for sandboxing.

After all, human user data is much harder to influence - leaving Google alone to process the resulting pure information.

Are links dead then?

Links for SEO in Google is a dying game. But links are not dead.

I’ve mentioned before that the best use of links for general webmasters is not for link bombing purposes, but as a means of communcation.

Going back to founding principles - links are recommendations. What Google is slowly but surely doing is shifting the emphasis, so that link recommendations come not from the search engine crawler experience, but the human user experience.

Links are and still are a valid form of advertising - Clickstream data will ensure that good links that encourage human user clickthroughs will still count - simply less directly.

Almost perversely, this means that human user clicks through PPC channels could become an important element of SEO work. This would blur the boundaries between SEO as optimising websites, as opposed to search marketing based on paid-keyword management.

Even running a successful AdWords campaign could eventually become an indirect influence on natural rankings.

Conclusion

Links for SEO in themselves are not dead.

For a start, Yahoo! search and MSN search have a less punitative approach to links, applying other methods to quality control of their search results - not least a focus on human editorial judegment at Yahoo! and focus on content at MSN.

However, the industry that formed to supply links to directly influence Google rankings, is dying. It’s a process of death that will take a couple of years yet to be fully realised, but the process is already well underway.

As that reaches conclusion, so will links become more directly useful as a pure advertising medium, rather than technical SEO benefits.

There’s nothing inherently wrong with links - if they are not misleading, then however they are processed, the human user experience is satisfied.

However, Google will ensure that they have greater and greater control of their own index against third-party manipulation. To do that, they will need to move more and more towards a Clickstream model.

I trained up in the “Links are King!” school of SEO. Now I’m defecting to “Clickstream is King, and links are its servant”.

That’s why in 2006, if I want my own Search Engine Optimisation company to survive, I have to evolve with the internet. I will become an advertising company.

Links are dead - long live links!

(For personal comment on this issue this week - see Waving a Red Rag at Google on my personal blog here at Platinax.)

* DMOZ descriptions have unfortunately also been used by Google - you can often see lame one-line “descriptions” which can only impact your clickthrough rate. If you’re affected, you can always try to ask DMOZ to update your listing - but don’t hold your breath.