Google Search will be all about relevancy and context (not keywords and links)
Google hasn’t really changed that much since 1997:
Historically they have also been good at analysing keywords and links and crawling every page on on the internet using spiders.
In the old days you just needed more keywords on the page. This couldn’t last though. Everyone could do it and rank highly. Google then figured out a better way – PageRank. PageRank was (and to some extent still is) their biggest USP. This iterative and very clever link analysis algorithm allowed them to use vector space models to store link data about all websites on the web more efficiently. That worked… for a while at least. Then this happened:
As it was the case with keyword inclusion PageRank was also prone for misuse. True, it was good at analysing link and link sources but it lacked one critical ingredient – relevancy. This was quickly rectified and the modern version of PageRank looks at the relevancy of the referring site as one of the most important elements.
Then Social exploded.
So what Google did? They’ve rewritten the back end to be able to index new content a lot quicker and called it appropriately Caffeine.
The next step in evolution of Google was personalisation… by IP, location, history, language, device and obviously Google+ activity.
So far so good. They fixed PageRank and ensured pages get indexed a lot quicker (in response to Social). The time was up for dodgy SEO spammers.
The introduction of every iteration of Panda and Penguin algorithm updates is met with great anticipation by many SEOs and brands whilst others are still oblivious to the fact that Google is officially the biggest policeman on the internet.. for commercial websites. The list of big losers is long and painful but here are some of my favourites:
So this is where we are now. Google anti-spam updates render old school link building tactics useless and suddenly everyone is a content marketeer.
With the exception of just few countries on the globe, Google has a lion share of searches happening every day.
Judging by the average click-through rates from Organic Search Google does a pretty good job at presenting the most relevant results at the top:
Yet still they face some serious challenges, including:
- Revenue diversification – most revenue still coming from advertising
- Declining desktop searches and ever declining CPCs (cost per click)
- High expectations from investors
- Facebook Search in the next five years
- SEO – the arch enemy of Google’s search revenue
See my post on TMW website for details: The key challenges facing Google in the coming years
So… what’s next?
To understand the rest of this post we have to get under the skin of what some of Google’s most outspoken representatives talk about.
My vision when we started Google 15 years ago was that eventually you wouldn’t have a search query at all. You’d just have information come to you as you need it.
search is for users, not websites
What this would suggest is that Google will be not be the search engine as we know and understand. During one of conferences Amit Singhal, Google’s Senior Vice President (nice title btw), has announced that Google experience will be focused on three key aspects:
Essentially Google wants to be an Answer engine which you can have a conversation with and which anticipates your search and gives you the info before you actually type in the search query. Clever stuff!
To achieve that they need data… a lot of data. Hold on, but they already have a lot of data and more is on the way.
They have Google Fiber serving super fast broadband to people (and know what sites you visit and when).
They will have balloons beaming wireless internet to underprivileged regions of the world (also gathering behaviour data). They are investing $60,000,000+ in Titan Aerospace drones… also serving internet from (and guess what, gathering search behaviour data).
But why stop there? Sky is literally not the limit. Why not invest $1,000,000,000+ in launching 180 communication satellites. Please don’t ask why.
Closer to home again, you want a website? Google Domains will register it for you…. more data!
Bored of driving? Google car to the rescue. In far far future it will not just get you home safely but will also answer any burning questions on the way and will remind you about that wine you forgot to buy.
When you eventually get home and try to chill out on your sofa Google Chromecast will register and record your TV preferences. Oh, and by the way, they will obviously know when you usually get home… thanks to the Nest in your living room. Just in case, the always-on WiFi Dropcam will do the rest. Getting seriously spooky now.
Then there is Google Shopping Express delivery service… just in case you decided to ignore buying that wine and realised you really really need it.
Let’s not forget about the Glass recording your location and activity LIVE!
If you add the existing services like Chrome, Gmail, Android, phone handsets, TVs, Chromebooks, Google Drive, Google Surveys and many many more you can start to appreciate the vast extent of the data they do and will know about you.
That’s all great but how does that relate to Google Search? The simple answer is relevancy and context. By understanding what you want right now and what you will need in one hour, month, year, they can serve much better ads, can charge for it more and can get around the problem of declining desktop searches
The (near) Future of Google Search
One of the fundamental issues with the way Google relies on keywords and links is the fact that they don’t actually understand website content beyond analysing strings of characters. The simple truth is that any machine, including Google, sees a web page like that:
Some clever people a while ago thought: maybe if we wrap it in some meaningful tags the machine can understand it better?
Lo and behold.. XML (e.g. HTML) was born. Sadly, as it turns out, the machine can still see this:
Just think about it. Try explaining this to the machine:
Seriously, try to explain it to an alien or newborn. What is this mysterious liquid contained in that weird container, what is it standing on, what is the texture, taste, smell, and so on.
Credits for the analogy must go to my Semantic Web professor Michael Zakharyaschev.
What you need is an ontology, a standard system with real world entities and relationships between them.
In fact Google has been using many open source standards for some time now. The structured data often manifests itself in search results.
How can you influence that and make the most of it? You need to be featured in Wikipedia and Freebase, and implement relevant structured data following the guidelines from Google Rich Snippets page or Schema.org.
Let’s also not forget about Google Hummingbird. Hummingbird is not an update like Panda or Penguin. Instead, it was a complete rewrite… which no one noticed for three months!!! Why? Because nothing major has changed in search engine results. Hold on… so what have they done?
As it turns out Google have rewritten the algorithm to process longer, conversational, natural language search queries, e.g. ‘give me directions to italian restaurant near me with good ratings that serves carpaccio’. The next reason is to marry the new algo with personalisation and their synonyms / entities from the semantic search. Lastly, Google Hummingbird will focus on the context and meaning behind the words.
But seriously why? The internet of things!
You can already taste the future. Just download Google Now app and explore all the cool things you can do with it. It has some serious implications for the way we’ll be searching in the future. On one hand this is brilliant. Google will be able to do that anticipatory search really well (show me what I want before I ask you/search for it). It will also allow users to search and interact with Google hands free – great for many moments, from making dinner to safely getting you to the destination. On the other hand though, it will potentially limit users’ choice of results and, dare I say it, will render Organic and Paid Search useless… food for thought.
Obviously Google Now is not the only answer app / personal assistant:
Sounds great but Google Now is still a long way from perfecting this. One of my favourite tests for whether Google has been successful in Semantic Search is the search for ‘animals that use echolocation but not a bat or dolphin‘. Right now you’ll see these results:
Err… that’s not what I wanted Google! For those of you really interested there are few more, e.g. owls. I hope you see what I mean. Another example showing how Google has failed “me”:
Come on Google.. I’m in London! OR this one:
Come on Google… I’m at BrightonSEO!
Summarising it all sounds pretty spooky. Is Google building one of those?
While it may seem they are building something like that, I honestly believe the truth and the fact is that Google is and will be more about data and not search. The application of this data is what no one seems to be able to get their hand around.
What does it mean for SEO? Well it means you have to be ready for it. That includes coding the sites in a structured way and ensuring brands are established as semantic search entities. What about PPC? I’m sure that as with everything else, somewhere deep inside, Google already knows how to monetise this. In near future could this be about advertisers paying for voice search responses?
You can find the full deck where I talk about the Future of Search on Slideshare and below: