Continue in 2 seconds

Getting to Know You ... Getting to Know All About You

  • Claudia Imhoff, Joyce Norris-Montanari
  • August 01 2001, 1:00am EDT

The Internet has grown and matured to such an extent that most organizations are now compelled to include it in their strategic plans. You can't pick up marketing, sales or technology magazines without seeing at least one article on how to make the most of your Internet exposure to your customers. The Internet has become the most pervasive technology to accelerate your ability to gather information on your customers. What's more, your Web-savvy customers are some of the most demanding and capricious ones yet, causing marketers to reach for antacid on a daily basis.

However, as with most business tools, a Web site's value to an organization is less intrinsic and more a function of how it is used or applied. More than ever, your ability to acquire and retain these shrewd customers now depends on how much you know about them. Increasingly, when customers can get a product from multiple sources, they base their purchase decisions less on the product and more on the personalization of information delivery and content. The more you personalize your product for the Web site visitors and the visits themselves, the more they will expect or demand that treatment next time around.

To achieve the goals of e- marketing, personalization must be customer-centric, not product-centric. A true customer-centric enterprise does not stop with just this form of personal interaction. A true customer relationship management (CRM) enterprise will trace every interaction with the customer including complaints, call center transactions, in-store purchases, etc. – not just Web visits.

Certainly, one approach to customer centricity is to create a detailed profile of each customer's buying habits. The more data you capture about the customer, the better the e-business applications can create personal interactions, provide better service and, ultimately, increase customer loyalty. Many enterprises have arrived at the same conclusion: To more effectively engage and retain customers, you have to personalize customer communications during customer Web visits. For organizations that embrace the principles of CRM, a Web site represents significant opportunities to increase the quantity and quality of customer interactions and improve competitive advantage.

Assimilation of customer data occurs at multiple places within the enterprise. In the case of Internet-sourced customer information, assimilation starts when the customer visits the Web site. From there, customer information is assimilated into individual customer profiles and customer classifications. Finally, it is assimilated into the enterprise's strategic customer knowledge. All of this assimilation requires a sustainable technical architecture, one that facilitates an enterprise customer view, integration and synchronization of operational data and processes, and thorough analysis of enterprise data.

Maximizing the CRM benefits of the Internet often boils down to gathering, analyzing and responding to customer information at three specific points during the process of assimilation (of the data, not the customer!).

POINT 1: Real-Time Customer Interaction Gathering

One challenge with personalization is that it requires large amounts of data to be collected every day for every customer for every Web visit. With increased traffic, this data is growing exponentially. Traditionally, this data was used to monitor Web traffic, determine future bandwidth requirements and assure trouble-free Web-site operations. Now it is being used to tweak campaigns, up-sell or cross-sell customers and redirect site visits. These new uses of Web data have dramatically increased the technological complexity and difficulty of capturing and analyzing the data. Three significant problems must be overcome before useful analysis can occur:

Volumes of Data: Popular Web sites create several gigabytes to terabytes of data per day; however, these sites do not mistake volumes of data for volumes of information. Synchronizing, combining and integrating server logs, sorting and then processing the data is a time-consuming and complex set of tasks that must be accomplished under severe time constraints.

Quality of Data: Prospects and customers do not always complete forms or enter sufficient data. They may start an order and then cancel the transaction, or they may only have a cookie with no other information. Sometimes, prospects and customers may even give false or misleading information. Filtering and sorting out extraneous and unnecessary data and then integrating the remaining data with existing data is difficult at best and may be impossible.

Integration of Data with Other Systems: Your Web site does not exist in a vacuum. It is simply one part of the customer's experience. A Web site's effectiveness must be examined in relation to all other customer touchpoints (other sales channels, CSR interactions, distributors, VARs, etc.). It is foolish to believe that a Web site is profitable just because its revenues exceed its costs because it could be cannibalizing another once-profitable sales channel.

How do you overcome these difficulties? A good start is to define the objectives and critical activities you want to track and analyze. Simply understanding what data you need to collect and how to define it is a significant step in the right direction. However, you must then determine which activities should be measured and how you will measure them. This is important because objectives will overlap, causing conflicts and confusion regarding what is truly needed. Your company will differentiate itself only if it can determine what activities are unique and critical in achieving its objectives and how progress toward these objectives will be measured.


Many Web visits are brief. Given processing constraints and the limited time frame to recognize and interact with the customer, minimal analysis occurs at this point. The primitive analysis that does occur during Web visits is limited to determining if the visitor is a known customer, if that customer already has a profile established and, if so, what content should be displayed. Note that the analysis to determine the content presented to an existing customer was performed earlier using prior Web-visit data.


Web data is used as a basis for immediate analysis where every keystroke or combination of navigational keystrokes may trigger an activity. The triggered activity might be a banner change presenting a new product at a special price or it might be an e-mail coupon sent to the customer after he or she has logged off the Web site. Additionally, the Web site may welcome the visitor by name, inform the visitor of new products similar to those he or she purchased or give recommendations or suggestions for future purchases.

It is important to differentiate between two types of responses – solicited and unsolicited.

Solicited Responses: For solicited responses in which the customer specifically requests something, you must ensure that every customer interaction is closed-loop. That is, there should not be any one-way communications. This means that every query is fully answered, problems are promptly handled and the customer is constantly informed of problems and order status. In addition, the technical architecture must support the Web visit. A customer ordering from the Internet expects the same level of service he or she would receive from a live customer service representative. For example, once an order is placed, the customer should be informed as to whether the item is in stock, when to expect the item and how to track its progress to the customer's door. This implies that the Web site is fully integrated with the operational systems fulfilling the order.

Unsolicited Responses: For unsolicited responses in which your company takes the initiative, you utilize customer information and take an action that you assume will be of use or of benefit to the customer. If your assumption is incorrect, you may be perceived to have violated the customer's trust or misused personalization information in terms of the customer's privacy. This is the one area that threatens e-commerce more than any other. Personalization requires the gathering of as much information about the customer as possible. Most people are still very wary of how and where they provide confidential or personal information. Statistically, you are far safer flying in an airplane at 30,000 feet than you are driving your car, yet no amount of statistics will convince some people that they are safer in the air. You can't always make people feel secure. Customers will gravitate to those enterprises that make them feel secure. You must define and implement policies and procedures that guarantee the maximum amount of security for your customers and monitor these policies and procedures to ensure they are fulfilling their promise.

POINT 2: Customer Profile Assimilation Gathering

The assimilation of customer profile information brings the Web information together with current customer information from operational systems, external customer demographic information and the results from strategic analyses. It should occur as soon as economically possible after the interaction at Point 1. The customer operational data store (ODS) is the component of the Corporate Information Factory that houses the customer profile information. It enables quick response time for queries about the current customer's situation and supports reporting capabilities for the business community. (Refer to The Corporate Information Factory, 2nd Edition, by William H. Inmon, Claudia Imhoff and Ryan Sousa, published by John Wiley & Sons, 2000, ISBN 0-471-39961-2, for a complete description of the Corporate Information Factory.)

Another way to solve the problem of the volume of Web data is to determine the level of aggregation, summarization or data grouping that is needed. Data reduction in the customer ODS can be substantial and make the ODS much more usable and informative. These aggregations or summarizations should be based on the objectives you defined for Point 1. You may determine that customer segmentation is an important objective for your enterprise. On the other hand, perhaps it is the customer's motivation for buying that is important. You then aggregate or summarize data based on these objectives. You want the fundamental nature of the customer's activities but not the extraneous clutter.

Regardless of the method of aggregation, summarization or grouping you use, you must maintain the consistency in your measurements. For example, if you capture revenues for each individual customer, then you must capture the costs for each individual customer as well.


Effective personalization occurs through two primary techniques:

Inference-Based Techniques: Software is used to profile or track a customer's behavior, identify other people with similar behavior and then, with a recommendation engine, create product suggestions to the individual. One drawback of this type of personalization is that the product recommendation cannot occur until someone has purchased the product that will be recommended. Perhaps one of the best examples of a company using this type of personalization is

Rules-Based Techniques: Software is used to provide campaign suggestions governed by rules as defined by the marketer. These are based on business expertise, previous clickstream analyses and prior campaign outcomes. Companies such as E.piphany, NetGenesis and WebTrends offer Web applications to help marketers analyze and fine-tune campaigns.

A personalization engine, along with other input, is used to create a customer profile. The customer's profile consists of a description of buying habits, products likely to be purchased and some demographic information.

The goal of most e-businesses is to understand their customers' buying habits and behaviors as close to real time as they possibly can. Analysis that is less than immediate – occurring 15 minutes to one hour after the Web logs have been collected – uses the defined aggregation and summarization schemes. Time-based analytics are used to gain additional knowledge about the Web site customer. Information gathered during this analysis will be used to generate the content (recommendations, banner ads, welcoming messages, etc.) during the customer's next visit to the Web site and is posted in the customer operational data store as soon as technologically feasible.

Detailed Web log information is propagated into the customer ODS multiple times a day. This can be accomplished using an extract, transform and load (ETL) tool. The profile record found in the customer ODS is updated in the traditional online transaction processing (OLTP) manner. The update to the customer ODS will record information about a customer including any items purchased since the last update. For instance, products that a customer may have purchased throughout the day can be updated in the customer ODS with an ETL process and may eventually be passed on to the data warehouse.

On a time-based interval, the Web logs are processed into higher summarization tables. The roll-ups or summarizations may take place within the customer ODS and are recreated to hold only a day's worth of history. The summarized roll- ups are used for reporting as well as tactical analysis. These summarizations could be placed in a cube or an online analytical processing (OLAP) "oper-mart" for tactical analysis. An ETL process can be used to create the summarized data as well.

With an application that evaluates the customer profiles, specific customers can be targeted for a structured marketing campaign based on the integrated information. The sales channels can quickly become aware of the opportunities in the ODS via mobile or Internet access within the corporation.


Although Point 2 analyses are extensive, Point 2 responses are simple: The recommendations for personalized customer communication are stored in the customer ODS so they are readily available the next time the customer logs onto the Web site.

POINT 3: Enterprise Customer Analytics


Historical customer data is propagated into the data warehouse from the customer ODS as well as from the traditional operational systems. This includes pertinent Web data such as purchased items, information requested, products temporarily put into the customer's shopping basket, etc. Snapshots of historical, strategic data can be captured on a daily, weekly or monthly basis for the data warehouse. Certainly, snapshots of daily log roll-up information are a great source for strategic analysis. The data warehouse team may want to snapshot the number of products, product types, dollars spent and number of visits to the Web site by customer over a month. Customer profile information from the customer ODS is combined with other operational data and loaded into the data warehouse. This data is then delivered into the analytical applications (e.g., data marts, exploration warehouses and data mining warehouses).


Once the data is gathered into the data warehouse and spun out into the various analytical applications, the business users have a plethora of opportunities to better understand their customers. They can use the analytical applications to determine high lifetime value customers, target customer segments for campaigns, determine the optimal sales channels to reach each type of customer, determine the appropriate actions to take in given situations (appropriate banner ads for specific customers, e-mails to others, coupons for returning customers, etc.). Over time, an analyst can look at a customer's buying history and determine likely next purchases for the customer and the best sales channel to use for these individuals (Web site, direct sales, VARs, etc.).


The analyses are used to create or modify marketing campaigns and enhance customer loyalty, interest and inclination to buy products and services. The results of the many analyses can then be populated back into the customer ODS, giving the corporation the ability to use these critical pieces of information in a real-time, quick-access mode (Point 1). For example, customers may be scored in terms of their value to the corporation by analyzing their buying history and behavior. The results of this analysis – the numeric score for each customer – can then be populated into the customer ODS. Now the results of this extensive analysis are available to anyone within the corporation instantly via access to the ODS. Thus, the customer service representative, Web site or sales representative can determine a customer's value before making contact with him or her.

Today, customers have more choices than ever. They are more aware of the possibilities and more demanding of personal attention. They choose their providers on whatever basis they wish – from price to features to service to customized arrangements. This situation shifts the focus from the product toward the individual customer. Convenience, return policies and credit translate into customer service. The more personal this becomes, the more your customers will be loyal to your organization.

If all providers have the same high level of product quality and convenience, the focus shifts again to trust in the relationship. Some customers trust their bookies more than they do a vendor of legal products and services because the bookie's livelihood is based on his/her record of fulfilling promises, protecting customers' privacy and promptly/correctly processing transactions. Perhaps we can learn from the bookie and strive to build systems that generate that kind of trust.

If you have a Web site, your customers and prospects will visit you there. Although their visits will be relatively short, they present a huge opportunity for both of you. Make the most of them.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access