Comments on Personalization is not Technology: Using Web Personalization to Promote your Business Goal

ji kim — Fri, 26 Jan 2007 14:51:50 GMT

I recommend reading two following books, they are packed with good information and will help start people new to personalization/recommendation technology systems (plus you don’t need PHD in computer science or psychology to read these):

1. Designing Information Spaces: The Social Navigation Approach
(expensive book, but if you are familiar with sites like “friendster” or “tribe”, this book will provide good ideas/tips. Pay special attention to chapter on collaborative filtering).

2. Word of Mouse: The Marketing Power of Collaborative Filtering
(written by guys who started/associated with a company called Netperceptions and University of Minnesota Group Lens project guys, easy to read book on marrying personalization technologhy for the the real world…it’s about time they wrote this.)

Christian Ricci — Fri, 26 Jan 2007 14:51:50 GMT

Mark,

Thanks for the comments. Personalization (PZN) in intranet sites, particularly Human Resource-type, service sites can be a very valuable tool. You want to focus on the natural progression (stages) and cycles of an employee’s experience.

Stages: Employees start out as “new hires”, become “mentors”, eventually “pre-retiree” and “retiree”. That’s a super simple progression; you can probably come up with other stages (for example optional stages like maternity leave or sabatical). For each stage, what an employee requires of an HR site may change. The new hire needs help filling out their benefits forms, may need keys to the office, a badge, etc. The employee about to retire needs help designating a successor, transitioning their 401K or pension plan, understanding their stock (option) plan, etc.

Cycles: Over the course of a year, employees need different things from your HR site. At the beginning of the year they may need help compiling their goals and filling out self/peer-reviews. At the end of the year, they may require information for open-enrollment/benefit submissions. Just before summer, they may need help filling out vacation requests.

The information you need to implement this type of PZN may be implicitely or explicitely derived… for example, you may know (or have access to) each employee’s hire date, recent promotions, etc. So, you know how long they’ve been with the company, if they are a recent management promotion, whatever. You’ve also got a corporate calendar…

Mix this with explicitely provided data—prompt the user mid-summer about whether they plan on altering their healthcare benefits this year… if yes, you may start introducing healthcare service documentation to the site, send out reminder emails, etc.

Carefully applied PZN can take into account the stage of an employee and natural cycles in their job, adapting the user interface, and presented content, to the set of potential tasks.

....but, make sure you don’t place boundaries around the information your employees need to do their work and feel “taken care of” as members of your organization—PZN, especially if it alters navigation or visible links, can hide or orphan content making the experience confusing.

Chris

Mark — Fri, 26 Jan 2007 14:51:50 GMT

My companies goal is to provide a personalization experience that collects information based on either web usage habits or life events i.e.) leave of absence, pay range, management training, retirement investments, etc…(retrieved from an HR records database), and responds to the employee with actions or content to review prior to the event. This could even be done through email. This maybe outside the scope of traditional web personalization however.

I think this is a good articule to help define goals and issues to consider before considering a solution.

Brett Lider — Fri, 26 Jan 2007 14:51:49 GMT

I think the modest approach to personalization is a good tact both for individual organizations and the concept as a whole. Nice article. I think it would have been helpful to have additional case studies or examples to really illustrate how personalization can be applicable in small contexts, pervasively, and in different ways that many of us assume.

Christian Ricci — Fri, 26 Jan 2007 14:51:49 GMT

Max,

If analysis paralysis doesn’t kill a project, it certainly has the potential to kill its momentum. I think you can often attribute an organization’s hesitancy to make “stake-in-the-ground” decisions on raised expectations when expensive software (and a costly implementation) is involved.

I also realize that many companies treat guidance by consultants and vendors with caution—finding ways to overcome this kind of barrier is important but frequently difficult.

Ji,

Thanks for contributing the URL.

ji kim — Fri, 26 Jan 2007 14:51:49 GMT

I agree with max that lot of businesses do not have very good understanding of applying web personalization technology to increase their customer’s user experence…...I also worked at a software company that provided personalization technology platform (collaborative filtering), and underspending on design from customer’s side was always a problem.
In side note, I rememer back then, it was hard to find research and papers that focused on personalziation and UI design…..although there are some good articles on published on recommender systems.

check out Rashmi Sinha’s paper:
http://www.rashmisinha.com/recommenders.html

Max — Fri, 26 Jan 2007 14:51:49 GMT

I was a UI designer for a leading vendor of an enterprise-level ‘personalization’ platform, back when that was the rage. I never felt that any customers even scratched the surface of what the platform was capable of.

I strongly believe that this was due to their inability to figure out exactly “what to do” rather than their inability to do it. In many cases, these businesses were probably overspending on technology and underspending on design.

Christian Ricci — Fri, 26 Jan 2007 14:51:49 GMT

Ji,

Thanks for the comments. You make a good point about the complexity associated with personalization of customer interfaces. Retail website personalization requires carefully balancing what implicitly (versus explicitly) derived data you leverage in manipulating the experience.

We’ve all heard stories of Amazon’s early (an ongoing) work in this area—Your purchase of a baby shower gift two years ago doesn’t necessarily imply that you have a child. Many factors are involved—perhaps one approach would be to give greater value to explicitly collected information or “scoping” implicitly collected data.

Each app is different. And many solutions require revision as time and consumer patterns vary.

Thanks again,

Chris

ji kim — Fri, 26 Jan 2007 14:51:49 GMT

Thanks Christian for the article, it’s been awhile since I read an article about web personalization – especially web personalization. As in your article, it’s really hard to implement out-of-the box personalization technology without carefully defining your business requirements to use cases and user profiles. Once solutions that worked on one particular industry will not work on the other.

Applying personalization also become espeically tricky on retail websites and web applications. Personalization in retail websites have to satisfy both the retailer and customers.
For e.g. retailers want to promote the latest electronic gadgets to increase their profit, but at the same time provide a way to dispaly the right products that customers are looking for.

So taking retail website as an example, when you apply a personalization technology to provide better user experience for the customers, think about personalization can be easily managed by the retailers and provide the right content for the customers.