The evolution of personalised search: Brilliant or just creepy?

Last week, in what can be characterised as a late move, Yandex announced that it has started personalising its results.

What particularly struck me following the announcement was the fact that Yandex’s proprietary technology would potentially allow the company to completely personalise the results it displays to users. But it does not. Why? Because during the testing of this approach throughout the last year Yandex discovered that by fully personalising its results it would actually be doing its users a great disservice. In short, they simply deemed these results less relevant.

So. what really got me thinking was this: If hyper-personalised results are not necessarily what users prefer when it comes to search, how come Google and Bing, in particular, are putting such a massive emphasis on it?

Innovation happening faster than we are able to adapt

If you are one of the cynics, you would probably argue that it is a purely selfish move for economic reasons, through which Google, as well as others, provide marketers with extensive targeting capabilities in return for higher CPMs, CPCs etc. Obviously, bumping up advertising revenues is an essential part of it… but is it really the only one? Could it be that it is also driven by a strong desire to create amazing, game-changing products that also happen to be monetisable? I would not know, I am just asking.

Ironically, however, it seems that the breath-taking speed of innovation occurring in technology today means that technology itself is evolving faster than we are able to adapt… and because there is something very pleasant and comforting about what we already know that is indeed scary to many of us. It is not necessarily that people do not appreciate change, however too many changes happening too quickly will often result in some form of negative or worried response. In some odd way you might say that technology is standing in its own way.

So perhaps the real questions to ask are:

Are the ways in which some search engines use our data to provide hyper-personalised results, in fact, creepy or less relevant? Or have we just yet to adapt to this new reality?

Whether good or bad, it seems clear that there is no way escaping it; personalised results are here to stay in some way or another. So in order to better understand and answer the questions above, taking a closer look at the past and present of personalised search and the attitudes towards it is probably a good way to start.

So, here we go (hope you are still with me :))…

“Big data” paves the way

In marketing circles personalisation has been one of the bright stars in the buzzword bingo for what is now decades. And in all fairness, it is not without reason. Personalisation is a tested-and-true tactic and has arguably become more important than ever before due to the increasingly cluttered information landscape we as consumers navigate in.

Surely, in an age of Big Data, marketers and publishers alike have never had better conditions of personalising their communication. The amount of personal data we are sharing online these days, often unknowingly, is feeding data giants, search engines in particular, with an abundance of information.

Bing and Google both say this data enables them to deliver a much better user experience by tailoring results to fit the specific preferences of the individual searcher. The critics on the other hand fear that Google will ultimately know more about you than your partner. They also argue that the increasing degree of personalisation in search results is detrimental to relevance – more on both of these a bit later, but first let’s take a look at how personalisation in search plays out in practice.

The three major areas of personalisation in search

Search behaviour, language, and geographic location of a searcher have historically been the main determinants of personalisation. These still carry great weighing and have gradually become more sophisticated signals in search engine algorithms. However, Google’s introduction of Search Plus Your World in early 2012, which arguably marked the most radical iteration of its search results ever, has shook up the entire search engine industry and added an extra layer to the personalisation equation so it now looks something like this:

Search hehaviour/history

Historic searcher data is what we typically associate with personalised results – be that:

• query history (what has a user searched for in the past)
• SERP interaction (what has a user clicked on in the past)
• on-page browsing behaviour
• bounce rates
• bookmarking

In the recent case of Yandex, its algorithm takes into consideration several factors when serving up personalised results, though it seems to rely mainly on the query history of the user. The more a user searches, the more tailored results will naturally become. In the example below, it can be seen how the same search query will trigger different suggestions and reorganising of results based on a user’s previous searches. For queries where Yandex does not have sufficient information on a user’s preferences, personalisation is switched off.

Personalisation based on search behaviour is an iterative process in the sense that, as your search patterns change, so will the results that are displayed to you. Accordingly, your most recent searcher history will bear greater weighing than search behaviour detected, say, a year ago.

Geography

Search engines have long used geographic signals to determine the relevance of their results.  However, the sophistication and precision with which results are nowadays served up have dramatically improved since these signals were first deployed by Google more than a decade ago.

In the wake of the proliferation of mobile internet consumption and the greater local precision on these devices, the days where geographic signals were restricted to the country-level are long gone. Heck, the far superior geo-location technology on mobile devices means we are even moving rapidly beyond personalisation at the city-level when searching.

In the example below, I have conducted two generic searches for Starbucks – one from my office and the other at home. Whilst it is not a completely different set of results, one listing has been swapped. Perhaps it retains the two stores that are closest to my home, even when searching from the office, since I primarily search from home on my iPad?

Even for PC searches, Google continues to place a greater emphasis on local results; probably best exemplified by the company’s Venice update this year. In essence, even for generic, non-localised search terms conducted on a desktop, local pages will now much more easily find their way into the top slots if in close proximity (or deemed in close proximity) of the searcher.

Note below the striking prominence given to “Seattle listings” when searching for a highly competitive, non-localised term such as SEO, as noted by Mike Ramsey in his great piece on how to understand and rock the Venice update.

• Tip for global businesses: For international corporations with footprints in multiple markets this further underlines the importance of making each presence truly local for maximum visibility. Consequently, restricting yourself to targeting at the country-level may no longer be enough, even if you have a stronger backlink profile than local competitors.

The added social layer

The integration of social into search is no longer any news flash, nor does anyone seem to doubt the decisiveness of the social dimension in search going forward.

Surely, in many ways it makes sense to make search reflect how we obtain and deem information useful in most other aspects in life; by turning to the people we know and trust for advice. Having said that, the extent to which these signals are already given prime shelf space in search results is striking – be that Google/G+, Bing/Facebook, or Baidu’s relatively recent integration of Sina Weibo tweets into its results.

In the specific case of Google, its determination to make its social project work is particularly noteworthy. Following Google’s launch of SPYW earlier this year, an abundance of articles have analysed the impact G+ connections have on search results. And the conclusions are fairly unanimous; social connections & activity on Google+ can be the number one factor trumping all other SEO efforts like links, title tags etc.

Google Authorship also reflects a move from what you might call the traditional brand (e.g. website) to the personal brand, i.e. the author. In essence, this means that the person linking to you can bear greater weighing than where it is from. And the Google+ profile of an author is for Google a strong  indicator as to how influential that person is. Expect this to gain impetus in 2013.

It therefore goes without saying that ALL companies are now neglecting G+ at their own peril. Since the content you add on Google+ can reach not only the people who have you in their circles but also their followings, boosting the amount of circles you are in can reap great rewards for SEO.

• Tip for global businesses: For global businesses in particular, this means identifying and mapping out local influencers and authorities within your vertical in each locale. At the International Search Summit in London some weeks back, Bas van den Beld did a presentation on Social Search and used the analogy of “local hairdressers” to describe these influencers; the ones who are likely to talk about your business and share it with the local community. By identifying and influencing those influencers, you will not only benefit in the social networks but also directly in the SERPs.

How it all comes together

So personalisation anno 2012 is a highly complex technical process that factors in a plethora of different signals. In practice, I imagine a step-by-step process could go something like this:

But are personalised search results really what users want?

Well, yes and no it seems. Search, as the word implies, is exploratory in nature. On the one hand the web is a BIG place, which need filtering and tailoring to fit personal preferences, but hyper-personalised results will on the other significantly limit the scope based on previous behaviour and/or social connections and actions.

Indeed, it appears to be a very delicate line to tread. As Grigory Bakunov, Deputy CTO of Yandex, said last week at the press conference in Moscow when announcing Yandex’s introduction of personalised search:

Our testing and research shows that our users appreciate an appropriate level of personalization and increase their use of Yandex as a results.

I have highlighted “appropriate“ because that somehow seems to illustrate the catch here. Yandex says it has seen some very encouraging figures when personalising results, such as 37% higher CTR on a personalised top result; however, when providing users with completely personalised results these were deemed less relevant.

As Andy Atkins-Krüger, CEO of Webcertain, puts it in his Search Engine Land article, “users do not want to get locked down in their own interests”.

So while search engines say it provides better results, findings from a survey by Pew released earlier this year point to the contrary; 65% of respondents perceived personalised search as a bad thing due to the limited information you get online, compared to just 29% stating that it provided more relevant results. And Google’s heavy push of its own social network in its search results surely has not helped in improving this perception.

Personalisation raises serious privacy concerns

As touched briefly upon in the opening section of this piece, serious concerns over privacy issues have emerged following the increasing degree of search personalisation. Google’s move, in particular, to unify its privacy policy earlier this year, through which more than 60 separate policies have been boiled down to one, has had privacy experts and lawmakers up in arms in concern that Google will ultimately know more about you than your partner.

In essence, the big G is now able to collect and merge user data across almost all of its services (something which was not possible with the previous structure), which in turn allows Google to gain a much deeper understanding of who you are and what you want to know at a given moment in time. More on this is in the final section.

It also appears that it is not just lawmakers and privacy crusaders who worry about the means of personalising search. The aforementioned survey by Pew took a look as well at the extent to which people found personalised search an invasion of privacy, with findings showing that a whopping 73% were “Not Ok” with the increasing personalisation and storage of their user data.

In Europe, this concern has been followed up by formal investigations carried out by French watchdog CNIL, who subsequently has found Google’s policy to be breaching EU law on several counts and has given the American search behemoth until spring next year to formally respond to its recommendations. Given this, it is far from unlikely that we will get to a point in 2013 where Google will have to adapt the way it handles user data and ultimately how it operates some of its web services in certain countries.

• Tip for global businesses:
  If you are an international player, beware that this affects you too (or at least it should). For example, if you are doing business online in Europe, I probably will not have to remind you about the “Cookie” Law! As regulations and general privacy concerns largely differ between countries, so should your market approach. In the worst case scenario, you might be in breach with national laws yourself.

The future: Google knows what you want BEFORE you know it yourself

Well, in that case the future is already here.  Sort of. Search is certainly undergoing a radical transformation these days, but since the building blocks for the next generation are currently being laid, the true revolution at this point in day is mainly happening behind closed doors. Having said that, we are seeing product launches that provide us with a good indication of where we are headed, and although these products can already seem amazing – magical at times even – they are merely beta tests of what is in store for us in the near future.

Google Now is one such product – an ambitious cross-over company project that gives the first indications of how Google’s many products may come together after unifying its privacy policy. The Google Now app, which essentially comprises of voice search and automated cards that “pop up” with relevant contextual information (without you explicitly searching for it), is currently only available on Android devices running “Jelly Bean” and up.

Indeed, it radically alters the paradigm of search as being a blank box in which you actively go to seek information. Because Google is now able to pull information about you from almost its entire range of web services – be that calendars, Gmail, maps etc. – it is developing a much deeper understanding of not only who you are but also a contextual understanding of what you need in real time.

For example, it might “tell you today’s weather before you start your day, how much traffic to expect before you leave for work (it already knows the route of your daily commute), when the next train will arrive as you are standing on the platform, or your favourite team’s score whilst they are playing… All of this happens automatically. Cards appear throughout the day at the moment you need them,” Google explains on the landing page for its relatively new cross-over service.

• You can check out the short Google Now video here.

In its current shape, it functions more like a personal assistant you just hired. He/she might bring you tea when you asked for coffee, but with time, effort and training this assistant has the potential to become better than you could ever have imagined. Down the road, this personal assistant will always be one step ahead of you, handing in documents you need before you even thought about it yourself, reminding you about appointments that slipped your mind, and providing you with exact directions to get there. In short, a proactive rather than reactive personal assistant.

Increasingly, you will be opting-in by default. So if you are an avid user of Google products and do not actively go change your settings, Google will know when your flight departs, in which gym you normally do your afternoon workout (and when you do it), if you are in near proximity of any of your friends or family etc. and provide automated suggestions accordingly.

So basically, although these are baby steps to what search has the potential to become, Google is in some cases scrapping the second step – the one where the user types in a query – in the search personalisation process that I outlined earlier in the piece.

This is radically different from the current “bar bet settling” function of smartphones as a blunt search instrument, as wisely pointed out by Dieter Bohn in this in-depth article on Google Now & the future of search.

And one thing is certain; it truly divides the waters. Some people seem to love it, while others are genuinely freaked out by it (as illustrated in some of the comments that I have scraped from articles and videos about Google Now):

So now that we have a better understanding of the evolution of personalised search and where it is headed, let’s go back to the opening question: Creepy or brilliant? To be honest, although I do not even have a spouse to remind me that I need to go exercise, as opposed to the person commenting above, I am still not quite sure which of the two to go with.

Now over to you! Are you a supporter or totally creeped out by this? Are you rushing to your Google settings to opt-out as you read this? Or is this just the inevitable evolution of search, technology and society in general? Have your say in the comments below. It is much appreciated!

The following two tabs change content below.

• Bio
• Latest Posts

Immanuel Simonsen

Research Manager at Webcertain

As the Research Manager of multilingual web marketing agency Webcertain, Immanuel heads up the company’s global market research activities and large content marketing projects. He is the author of several reports and guides, including ‘The Essential Guide to Rel-Alternate-Hreflang’ and ‘The Webcertain Global Search and Social Report 2013’. Apart from being a tutor at the International Marketing School -- teaching online marketing professionals on business opportunities around the globe -- Immanuel is a regular speaker at the International Search Summit, a leading event series dedicated to multilingual search and social media marketing.

Latest posts by Immanuel Simonsen (see all)

• Is Baidu losing its crown in China? - July 31, 2015
• Global logistics brand DHL eyes Chinese e-commerce growth - July 27, 2015
• VKontakte To Launch Rival To Instagram - July 21, 2015