If there is one trend which has been very visible in the last few years it is personalisation on the web. The more you did on the web, the more personalised information you were getting back. It was in 2009 that Google opened up personalisation for everyone, meaning that your search results were from that point on influenced by your behaviour.
At first this was just search behaviour, but slowly that has evolved as well into social behaviour. Still if you do searches on Google, or whatever other search engine you prefer, there still seems to be a lot of flaws in the accuracy of the personalisation. This is partly because for search engines it is hard to pinpoint what it is you exactly want. It takes data to figure that out. Lots of it.
And it becomes even more difficult to personalise when your opinions change. We are all human after all and we can change our mind. Hell, we do that all the time. Russian search giant Yandex last week took another step towards perfecting personalised search, by looking at your every move, even your latest move. It was something Ilya Segalovich talked about a few weeks back at the International Search Summit.
Yandex explains what they are doing in a video:
A big difference in technology between Yandex and others is that Yandex personalisation technology doesn’t just take the last weeks into account, but it actually looks at seconds, so within the session you are in.
To do this Yandex created a real-time data processing system, which processes more than 10 terabytes of data a day, at up to 500 megabytes per second, continuously correcting its knowledge of users’ needs.
The general person has a lot of different interests, some of them being more stable than others. When you like football for example chances are that is not changing too rapidly, which means it is a ‘stable’ interest. But a sudden interest for a specific movie for example might show you have a stable interest for movies if you’ve looked at movies before, but it might also show the less-stable interest in the specific topic of that movie. It could be that you are looking for when you can actually see that movie, so adding “showing in theatres’ or words like ‘when showing’ will give search engines insight into the interest you have at that specific point of time.
Yandex now acts on the less-stable interests, for example by giving you suggestions or making it much more relevant (like the 2013 version of a movie instead of the 1925 original novel):
Off course we don’t know exactly how Google is personalising search, but as far as we can see Google does not take real time into account. The big difference between Yandex and Google at this point is that Yandex is looking at the short term much more than Google is. Next to that Yandex is also turning on the volume of the number of searches affected by personalised results. Where Google (at least that’s what Google claims) ‘only’ takes a percentage of their searches and personalises it, Yandex has been personalising search for 30% of the users not for ‘just’ a percentage, but for about 60%, and they intend to grow that to no less than 85%, for all of the users, a huge difference.
This will make every search unique. It will make that the same search will have different results for different people.
So why should we care about what Yandex is doing here? Apart from the fact that Yandex is a fast growing search engine who is not just focussing on Russia anymore and you there fore should pay attention Yandex is also setting the stage here.
The increase in personalisation of search results makes sense. It ail give searchers much more relevant content to look at. This will without a doubt be followed by others as well.
The relevant content means people will get to see what they need right there and then. So you want to see a movie now, for that you need the relevant results based on short term, not long term. Or you want to know now where you can buy something. In this mobile era having more relevant results which you don’t have to ‘ask for’ is exactly what the searcher wants.
As a marketer this means you have to adapt to that. You have to make sure your information is as up-to-date as possible and you have to realise an important element: you are looking at individuals here, not groups of people with the same behaviour. So you really have to get to know your audience.
Finally it will shine a very different light on rank tracking. If only a few percentages of searches are personalised it makes sense to look at rankings. But with this many results being personalised you might be looking at the wrong numbers to draw conclusions from.