Google's is investing big in AI and announced last October that RankBrain, it's 'deep [machine] learning' AI will be overseeing the core search algorithm, acting as a conductor to the 100s of ranking signals and core algorithms that determine the best results for users.

But RankBrain, much like Google Translate, gmail spam filters and even driverless car software are all examples of Artificial Narrow Intelligence (ANI), meaning they are designed to perform and improve upon a narrow field of activities (if you think analyzing 2.3M searches a second is anything but narrow!).

RankBrain builds upon Google's 'connectionist' belief that all knowledge and understanding can be derived from the semantic connections between entities, similar to how neuron networks work within our own brains. 

Building on previous Google infrastructure updates, such as Hummingbird, RankBrain is learning to better understand natural language and the 15% of unique daily searches it has never encountered before -- and it's learning fast! 

The brilliance of 'deep learning' and the law of accelerating returns means that RankBrain has already self-improved to the point where it can even outperform Google's top engineers in delivering better search results and remember it's only still a AI infant!

But what does this mean for search marketers, general users and the future of Search Engine Optimisation?

One Size Doesn't Fit All

Well technically search hasn't for a while now but RankBrain acts as the supreme algorithmic overload and evaluates each search query (along with location, intent etc.) individually instead of with the broader algorithmic brushes of the past. 

Because it's been fed so much data the not-so-fledgling AI can better understand queries independently and decide the weighting of lower algorithms and signals needed to deliver the best result.

For example:

  • 'Where to buy the best X': Will give greater weight to local and shopping results with positive review signals.
  • 'What does X mean' or 'How to do Y' will deliver answers results, videos and rely on positive links, content quality and social signals to deliver what the user wants.

And this continual blending of the algorithmic mix under the benevolent dictatorship of AI will up the complexity beyond anything we and even the brainiacs at Google can comprehend or articulate.

No More Loopholes

As AI pushes search to evolve at break-neck speeds the standard analysis models and deployment of SEO to meet certain criteria across a wide range of sectors and topics will no longer drive the universal benefits delivered by implementing best practices that tick the same algorithmic boxes.

Hence the days of universal quick wins are over and SEOs will need to abandon the idea of 'chasing the algorithm' altogether because let's face it you're not as smart as a search engine and haven't been for quite some time.


Focus On Your Niche

Another trait of the new RankBrain architecture is that it learns by comparison, identifying vertices, rewarding what it sees as typical in a given category  and what are good and bad characteristics of the sites within (then learning and repeating).

RankBrain will evaluate what are the best sites (generally big brands) and expect other 'valuable' sites to provide a similar experience, architecture and link profile to be branded as good and rankworthy (with criteria obviously varying from sector and search term).

So this provides a real opportunity for digital marketers to take advantage of a more complicated landscape and do better than the competitor next door.

Starting with more detailed analysis of your niche, future proof digital strategists will evaluate who is being rewarded by RankBrain and what they need to do to receive the same rewards.

By Judging:

  • What is the prevailing site structure, UX, product categorization etc of top sites. 
  • Not just keyword volumes but what is being delivered in the blended search results (answers, videos, images, maps etc.) and were are there long tail opportunities.
  • What content are top sites producing, what is the format, what are the related topics which satisfy both the quality and relevancy RankBrain demands (and where is the topical gaps and white space to compete?). 
  • What is the link profile of the top performing sites in your niche? RankBrain will not just be judging sites based on good or bad links in a traditional sense but what is the makeup of truly connected brand and reverse engineer what it would take to be featured by these same 'category specific' influences.