---

Personalization is the process of deciding – given a large set of possible choices – what has the highest value to an individual. These choices can range from a customized home page look and feel to product recommendations, from banner advertisements to news content. As we will see, personalization is a very broad term and must be further qualified.

Personalization:

Personalization is the process of deciding – given a large set of possible choices – what has the highest value to an individual. These choices can range from a customized home page look and feel to product recommendations, from banner advertisements to news content. As we will see, personalization is a very broad term and must be further qualified.

Objectives

• Personalization as a concept
• Uses of personalization
• Different types of personalization

What can be personalized?

Personalized content may be advertising, recommended items, screen layout, menus, news articles, or anything else accessed via a web page or software application.

Business benefits

Personalization contributes to a variety of e-business goals:

Measuring success

Personalization should lead to measurable increases in sales, items per sale, site traffic, and conversions. Ideally, site tracking tools, such as IBM’s Site Analyzer, can be used to develop this data.

Privacy issues

In any personalization scenario, privacy must be balanced against increased user benefit. Tracking users’ buying habits, of course, requires collecting some fairly sensitive information. As a counter to these privacy issues, though, personalization provides benefits, in terms of increased usability and less unwanted content

Types of Personalization

We’ve established the benefits of personalization—it helps your site attract, keep, and re-engage customers and increase the quality and quantity of customer sales. In fact, personalization is one of the primary advantages provided by Internet-enabled businesses.

	Personalization	Advantages	disadvantages
User-profile based	These allow users to determine the content on a personalized page, e.g. “sports,” “news,” “entertainment,” and so on.	Users can select exactly what they want to see. They control their own profile and the information in that profile.	This process is extremely manual. users may often miss new content if profile gets stale etc.
Rules-based	Another form of personalization implements rules based on a customer’s demographics, past purchases, or product attributes.	This method can be implemented consistently with other marketing programs This method is great when you know your prospective customer base really well.	Rules-based personalization involves an extensive, research-intensive setup, which can be difficult to manage, particularly with a large inventory. Essentially, new rules must be added for each new product.
Market-basket analysis	Also mathematically generated, market-basket analysis provides “item to item” recommendations, based on items purchased together by past users.	Market basket analysis improves on rule-based recommendations, in that identifies some less obvious combinations of items. Since it can be used with less data, but doesn’t require rules configuration.	Market-basket analysis assumes that people tend to think alike—that is, that since user x bought two items in combination, user y will also desire these products in combination.
Collaborative filtering	This method involves gathering data on user preferences and behavior, and then using that data to produce recommendations for new users. These similar users are often known as “mentors” or “neighbors.”	Once the system is in place, it automatically generates new recommendations, without any additional input besides a continual stream of new users.	Because CF requires a body of data before it can make recommendations, it has difficulty making recommendations for new items. This is one half of the problem with “cold-starting” collaborative filtering on a particular site.