Enterprise Information Management

Is Personalization overrated?

Personalization has been one of the core features of portal products. But it is rarely used to its full-extent. It is just not considered worth the money. But as the attention market heats up, personalization will become the differentiating factor. The better the personalization, the sooner the end user can find the information they need.

How personalized is personalized enough?

Most people think that personalization based on demographic information like location, language and age or organization specific information like membership, groups and department is enough. But these factors rarely help you understand the user intimately.  One technique which has gathered momentum is analytics based personalization. User behavior is tracked and used to predict what they might like. This is more effective and it lets you get closer to the subconscious mind of the user, even if they do not explicitly call it out.

But is this personalized enough. Not nearly. Web 2.0 changed everything again. Collaborative filtering is another technique which has emerged which helps leverage crowd intelligence. Analytics are gathered from multiple users and used to make recommendations on what you will like. This goes closer to the subconscious of the user as it analysis not only what you have done in the past, but also what other similar users did in the past.

End of Information Portals?

So I get thinking, If someone can really personalize the information and deliver it to me, do I really need a Rich, attractive Portal with heap loads of content. Maybe not. What I need is an Attention Management system which knows and learns what I like, which goes and fetches the information I am interested in , which lets me configure that information and view it the way I want to.

Putting it all together – An Attention Management System (AMS)

So how will personalization change from what exists today in the form of personalization engines and portals? I believe the entire machinery of information generation and delivery will be tuned to deliver personalized content. An Attention management System might have the following components :

a)      Personalization Information source – This will be the component which provides the information required for executing the personalization rules. This could further be divided into

1.       Implicit information source -  This includes well know information about the user like location, department, groups etc. usually stored in directories.

2.      Explicit information source – This includes information provided by the user in terms of their areas of interest, likes and dislikes, opt-in/opt-outs etc. which is stored in systems like portals or social networking sites.

3.     Behavioral information  - Information about the browsing behavior of the user e.g. sites visited, time spent online.

4.     Collaborative information – Collective information about browsing habits of numerous users , and their interests. This could be gathered

There are concepts like Attention Trusts which might emerge as the source of majority of this information.

b)      Content source:  This will be the source of content that needs to be personalized. It could be content management systems, feeds or applications. But the change will be that it will no longer be enough to just create the entire page and tag it. Content will have to be created to be consumable in a semantic way. This is because instead of it being directly consumed by end users, it will be read and filtered by Personalization engines. So the simpler you make it for the personalization engines to find, the better.

c)       Content categorization engine : Personalization engine may not be able to pick up raw content, as they will not be able to read and categorize content. They can only provide a recommendation on the kind of information to pickup. Categorization will need to be done by another specialized component, and what better solution than to use a search engine. Search engines can read both the metadata as well as contents and categorize the content into different buckets e.g. news, United states, athletics, home improvement etc. When the recommendation engine throws out a recommendation, it can be matched with these categories to pick up the content. Ofcourse, Web 2.0 also plays a role here. Collaborative tagging can help categorize the content better through collective intelligence.

d)      Personalization decision system: This will be very similar to the personalization engines of today. This is where the rules for personalization will be written and executed. This engine will utilize the rules and the information from the personalization sources to provide a recommendation. These engines might evolve as standards like JSR94 and collaborative filtering algorithms mature. This will be the heart of the system, and the vendor who gets this right will win the game.

e)      Content Delivery system: Structured websites (read portals) might be a thing of the past. Instead, a Attention Management system might give you a completely configurable front-end (read Mashups). You will be able to assemble your portal by puling together attention blocks and configuring them to show information of a particular kind e.g. news, quotes, Tweets, shopping,  books, music, etc. The Attention management system will use its personalization engine to deliver the best content in that block. The content Delivery system could also become the source of behavioral information as it can track what you are clicking on.

The Attention Management Cycle: Spinning the Web

Well, what I mean by the Attention management cycle is that soon the web will go into a spin. This will be a continuous cycle of improvement where user analytics will feed the personalization information sources, which will lead to better content and better tagging, which in turn leads to better personalization, which in turn will lead to a satisfied end user who can give better analytics. This cycle will continuously strive towards a more satisfied customer. Content producers will win if they make their content easily consumable by personalization engines. Portal providers and search providers will win if they can get the personalization game right. 

The core of the attention management cycle will be the accuracy of personalization. This accuracy will rely on analytics information. As time goes by, information about you will not be free. You can protect this information and chose to share it or not. Some ideas like Attention Trust are a strong indication of a move in that direction. What this means is that if you can protect information about you, both explicit and implicit (analytics), you could also monetize it by charging direct money or points which you can redeem at the site you share your attention with.

 Thoughts?