Limits Of Personalization

The

other day on a visit to Amazon.com to order additional books, I was happy to be

welcomed by name and the page listed some recommendations based on my previous

purchases. While browsing the list of recommended books. I was disappointed to

find that I had already bought all the listed items, I had purchased most of the

titles from Amazon.com, but the site failed to recognize that simple fact.

If

you have been keeping up with the web and its buzz you must have read or heard

about personalization. It is being touted as the art of marketing to an

individual and delivering content that is focused on you and is targeted for

your eyes only.

What is personalization?

In a

nutshell, personalization is the art of changing your website’s appearance and

behavior so that it is tailored to an individual. For example, you step into a

departmental store and head towards the shoe section. Based on how much time you

spend in that section, rest of the sections transform themselves to display more

shoes. The more time you spend in the shoe sections, the more, the rest of the

store changes and finally the entire store ‘appears’ to you as a shoe store!

In the brick and mortar world, that may not happen but in e-space, this can

happen with the clicks of a mouse.

In traditional methods of market

research, the research company sends out questionnaires, engages focus groups,

performs phone surveys to collect data about consumers and categorizes it into

various demographics, including region, income level, family size and age group.

This data can be collected for a specific audience or sold to consumer goods

corporations that classify and categorize the information to support predictions

and ‘target’ their goods and services to desired ‘market segments’. The

objective of this exercise is to target appropriate customers for products and

services of the vendor.

In the new commerce,

opportunities are different. Now, the business offering the product or service

can reduce the need for specific market research services and collect data

desired directly from the consumer. Generally, the data is current, which

implies that the vendor can target content and the product or service to the

individual in real time! Once the data is collected, the e-store can potentially

cross-sell and up-sell by offering more content and products in line with your

profile, preferences and purchasing patterns.

Different methods of

personalization

Broadly speaking,

there are two types of personalization–usability personalization, also termed

as customization and content personalization.

With usability personalization,

users are given a set of choices to define how they want their view of the

vendor website to appear and behave. A clear example of this is myYahoo. This

type of capability can provide the customer with a personalized experience and

should help in retaining the customer as a regular visitor.

For content personalization, the

intent is to make the customer’s visit to the vendor site meaningful, by

tailoring the content to the customer’s personal tastes. For a specific

example, on the myYahoo site you can select from a list of gadgets that you

would like to see on your homepage. If you select the weather section in your

left channel pertaining to usability, then you can select cities for which you

want the weather data that pertains to content. A second example would be the

Amazon.com site where a very different approach to personalization

implementation is employed. Here you are greeted by your name when you revisit

the site along with a recommended list based on your previous purchases. On this

site, you do not have any choices on usability personalization or content.

However, the site personalizes the content for you.

The Yahoo model represents the

self-service model where the user is responsible for making all the selections,

while the other hand, the Amazon model is automated and based on the software

that comes in many flavors–collaborative filtering, statistical scoring, user

profiles, click-stream and rules-based logic. Both models have strengths and

weaknesses should be mixed and matched based on the target audience and business

purpose of your site. In the self-service model, the software shows no

intelligence and relies entirely upon user interaction, while the automated

model tries to anticipate user behavior without really understanding real world

situations.

Inadequacy in personalization

The

personalization logic proved incomplete during the holiday shopping period. If

you place an order for family members and friends with divergent tastes, the

site will assume that all the products are being bought for the logged on

customer’s personal use. So, even though you would not normally purchase ‘acid

rock’ music, the site will keep offering similar music CDs to you. The other

question currently is–how will the personalization software adapt to change in

user’s tastes with passage of time?

On the other hand, if the

self-service model were to be adopted, it goes to the other extreme and relies

completely on user input, providing little or no automation.

One of the good site designs and

personalization concepts is used by the home depot’s site–www.homedepot.com.

It gathers data about your objectives and goals as you browse. You can avoid

filling any forms if you wish. The content is categorized, based on a user’s

goals for coming to the site. Once you identify your goals by selecting one of

the tabs, the site pushes content in a seamless way and it never even appears

that the content is being pushed.

Beyond personalization

Personalization

is becoming the current hype of the web industry today. There are many first

generation players with products that will let you apply some sort of automated

content personalization. The most well known companies in this market segment

are–Andromedia, BroadVision and Net Perceptions. As always, in the race of ‘time

to market’, the first solutions were not tech-savvy but merely automated some

aspects of personalization. Most of the first generation tools still required

significant programming support, editorial staff and other members to make it

work. With many players involved in defining a solution based on personalized

content, the user is not guaranteed a great experience on the site that uses

these tools. With the potential of new tools attacking the same problem in a

different manner, it will be interesting to see the new solutions.

In a nutshell

Current

deployments of personalization represent a solid start. Care must be taken to

not just automate old methods of market analysis with the hope that it will give

better or different answers. In the quest to provide unique experience to every

individual visiting the site, one ends up intimidating the customer by collecting substantial quantities of data but is unsure of what to do with it.

With increased concern regarding customer privacy and the potential for

legislation, greater sensitivity to methods of collecting profile information

will be required. 

Rajeev

Nanda