Accumulative E-Business Intelligence

Published in July 2001

Business Intelligence, decision support, analytical capability—it’s all about attempting to predict and influence the future. In general, the more data about the past and present
that’s available to decision-makers, the greater the probability they have of accurately predicting—and influencing–the future. The challenge, of course, is making these volumes of
data available in a comprehensive and comprehensible format.

Probably the greatest challenge posed by e-commerce to data warehousing technologies is the increasing volume and diversity of data available for analysis. Any type of e-businesses is confronted
with these challenges, but they are likely to arise from different directions and according to a different timetable, depending on the type of e-business model on which an enterprise is based.

E-businesses typically fall into one or the other of the following:

  • Single-channel—“pure-play” e-businesses, or dotcoms
  • Clicks and mortar—companies with prior established channels outside their e-business initiatives

Which of these types a given enterprise falls into typically determines its stage of e-business “maturity” at a given point in time. A typical pure-play e-business enterprise probably
entered the digital marketplace earlier than its more conservative bricks-and-mortar competition, and is therefore likely to be farther along in the e-business maturity cycle. Bricks-and-clicks are
discovering that what was previously thought to be a disadvantage—the existence of legacy data assets—is in actuality an advantage. A time-tested technical infrastructure is likely to
constitute another considerable advantage of established businesses.

Pure-play e-commerce businesses, in order to grasp the range of data necessary to compete with more established competition, must span vertical channels to partners’ data. Clicks-and-bricks
e-businesses need “only” reach out across their own value chains to acquire the necessary scope of data assets.

Either position has apparent advantages and disadvantages relative to the other, as described in the table below.

Table 4.1 E-Commerce Business Model Comparison


Within either business mode, as the scope of available analytical data expands, so does the potential value of actionable information produced. But so does the cost of obtaining and compiling the
data. A framework of some type is required for defining and approaching these issues in an orderly fashion.

The figure below illustrates the concept of accumulative e-business intelligence, where each successive component builds on—accumulates—based on the foundation of the component(s) it


Comparative scope of e-BI data (not to scale!)

Clickstream Business Intelligence

The primary vehicle for B2C e-commerce is of course the World Wide Web. The largely standardized software architecture of the Web and the architecture of commercially available B2C e-commerce
server software influence to a great extent the basic types and content of the data created by Web B2C interactions. The primary data entities involved in Web-based e-commerce include

  • End-user gestures, or “clicks”
  • Customers
  • Products
  • Orders

Products, Customers and Orders are of course part and parcel of the business of selling, regardless of what channel is utilized for the selling activity. Clickstreams are a new source of
customer-behavior data made possible by the Web, and are indeed part of the very fabric of this new channel.

Essentially, a clickstream is a record of the path that a given Web site user takes through the e-commerce channel exposed to him or her by the selling enterprise. This path record provides
considerably more detailed behavioral data than that made available by any other selling channel. A company’s Web-based e-commerce site is in many ways analogous to a published product
catalog and order form. Imagine if an equivalent amount and detail of behavioral data was available on a customer’s interaction with this more traditional channel:

  1. Customer receives catalog in mail
  2. Catalog stacked in pile of mail
  3. Catalog retrieved from mail pile
  4. Catalog opened to index
  5. Pages turned to hiking outerwear
  6. Pages turned to clearance items
  7. Pages turned to order form
  8. Pages turned back to clearance items…

And so forth—with the hope of terminating in the mailing of a completed order form. This is exactly the volume and level of detail of the data collected in Web server logs. It is easy to see
why we are becoming increasingly dependent on intermediate software to categorize, summarize, and derive conclusions from data of this nature.

Both types of e-businesses can acquire and maintain clickstream data with equal effort. User clickstreams can be compiled, summarized, and analyzed to determine trends in customer
behavior—most importantly, what behavior most typically leads to a successful sale and results in the generation of revenue for the enterprise. The results of such analysis can be fed back
into personalization, cross-selling, and even pricing rules invoked in subsequent customer visits.

Data at the clickstream level of granularity can be found in mobile and telephone-based digital commerce applications as well.

E-Commerce Business Intelligence

Companies progress at varying rates from initial Web storefronts to adding the “real” functionality required for true Web e-commerce. Real functionality requires the capability for
visitors to input data—registering as buying customers, providing credit card information to enable payment, selecting products and placing orders, and viewing account status and balances.

Log files generated by Web servers are specific to clickstreams coming from Web pages. To begin to provide e-BI, explicit data relationships must be established among diverse data collected in Web
logs, e-commerce, and legacy operational and financial systems. Examples of such valuable data relationships include, among many others,

  • Web purchase transactions connected to the clickstream leading to the purchase
  • Geographies of visitors compared to that of purchasers
  • Busiest Web traffic volume compared to highest purchase days and times

Integration of overall e-commerce data—including digital channels above and beyond B2C clickstreams gives a company a correspondingly broader analytical perspective on its relationships
within the overall digital marketplace. Other typical digital channels include email, Voice over IP (VoIP), chat, XML, and EDI.

Multichannel Business Intelligence

Integrating analytic information from all channels—non-digital, such as physical branches and distribution centers, as well as digital—gives a company the broadest possible view not
only of cost- and revenue-generating activities, but more importantly, of its customers.

Within the past decade, an influential business trend with considerable impact on data warehousing has been the drive for businesses to become increasingly “customer-centric,” or
customer-oriented. Customer-centricity entails a shift in enterprise focus from one of selling products to one of pleasing customers—a focus often summed up in the phrase “Know thy
customer.” This perspective continues to evolve, expanding from just knowing the customers’ current characteristics and past behaviors to attempting to predict what their future
behavior will be, in order to anticipate and proactively fulfill their wants and needs.

From a data perspective, customer-centricity requires that any available customer-related data be consolidated, cross-referenced, analyzed, and internally disseminated to the greatest extent
possible. This requires integration of data on customer characteristics, transactions, and all other activities, gathered from across all channels with which the company’s customers interact.

Widely adopted applications for this data include closed-loop CRM and personalization. In addition, increased depth and breadth of knowledge on customers enables increased capability to anticipate
and/or detect fraudulent activity.

Closed loop customer information management consists of four major function groups connected into a closed loop by data flowing between the function groups. This flow is shown below.


Closed-loop customer information management
  1. Track: Internally available customer-related data (transactions, clickstreams, call center logs, etc.) are merged with customer demographic data available from external providers. The
    most common storage implementation for this merged data is a single database such as a data warehouse. However, in some cases a middleware connectivity infrastructure can allow multiple
    standalone databases to be accessed as a single data source. Each approach has benefits and drawbacks.
  2. Analyze: Calculations based on user-defined metrics are performed using the integrated data as input. Profitability scores are assigned to individual customers and/or groups of
  3. Model: Campaigns and promotions are devised and targeted to groups of customers (clusters) with similar profitability scores and/or behavior characteristics.
  4. Adjust/Act: Campaigns and promotions—cross-selling, up-selling, discounts, permission marketing, etc.—are invoked within customer-facing processes (browsing, shopping, order
    entry, phone contacts), through multiple channels, in an effort to promote and reward profitable customer responses.

Responses from customer interactions—positive, negative, and neutral—are fed back into the integrated data store to be used in determining subsequent promotional activities. This loop
enables customer-interaction processes to “learn” from the success or failure of previous profitability-enhancement efforts, with a resulting increase in success rate and overall
profitability, over time.

Enterprise-Wide Business Intelligence

A wider analytical perspective is made possible through integration of enterprise-wide data across all channels and value chain linkages. Revenue as well as cost information across all channels and
value chain activities can be juxtaposed and interrelated for the benefit of timely and informed decision-making.

Enterprise Information Portals (EIPs) can be said to span whatever gap there may be between the ambiguously bounded realms of BI and knowledge management (KM). EIPs have appeared as a means by
which a “standard” user interface (i.e., data-human interface) can be imposed on diverse data sources and formats as a method of coercing such various data sources into an appearance of
integration. Often, such a veneer is all that is called for. Conversely, in many cases the last step of cross-relating the various pieces and parts is left up to the end user.

Three converging phenomena—widespread adoption of the Java language, e-commerce pressures, and hungry software vendors—have induced a rapid evolution in information delivery for
decision support—from query-and-reporting facilities, through online analytics, to what can fairly be termed business intelligence applications. Applications of this type are typically
designated EIPs, or digital dashboards. Microsoft uses the term digital dashboard in reference to EIP applications built using its software products—specifically Office 2000, Outlook 2000,
and Exchange Server.

A true enterprise information portal can be distinguished from a simpler Web reporting application by the scope of data it contains: enterprise-wide business intelligence, by definition. In a
nutshell, an EIP typically

  • Contains multiple windows within a single screen.
  • Combines decision-support information with operational information such as email and transaction processing.
  • Has typically been constructed using a “component-based approach,” that is, pulling together diverse, smaller “pieces and parts” of software—such as Java
    applets—and data.
  • Contains diverse information, often represented in different presentation metaphors—charts, spreadsheets, images, videos, graphics, tickers, etc., presenting information from multiple
    sources both inside and even outside the company.

Accumulating and relating external data along with internal data in the same interface provides employees with what amounts to a “market-wide information portal”.

Marketwide Business Intelligence

Combining a company’s internal data with discerning observations about its marketplace—competition, prospective customers, historical trends, and so on—adds a significant level of
analytical capabilities to its arsenal. The more a company can discover, retroactively and proactively, about where it stands relative to its current and future competitive marketplace, the more
astute and advantageous can be its responses to the marketplace.

A representative concept leading to enablement of marketwide data availability and analysis is that of Web farming. Originated and promoted by Richard Hackathorn, Web farming describes techniques
for searching, compiling, linking, and analyzing data from across the World Wide Web, with the objective of gaining and maintaining a clear understanding of the activities of a company’s
marketplace. Such marketplace data—detailed and summary information on the company’s competitors and partners—is freely available in the content of Web documents such as annual
reports, news releases, and stock market reports.

A challenge above and beyond merely acquiring such data is the ability to integrate and interrelate this data with internal data within the four scopes previously described. Data warehousing
disciplines and tools in data mapping, reconcilement, and transformation can be applied to make significant progress in connecting these types of disparate data sources and formats.

Actionable Strategy

Today’s multi-channel enterprises are faced with significant challenges when consolidating decision-making data from internal and external environments, as well as from the channels with
which their environments are interconnected. The concept of accumulative business intelligence can help companies construct an actionable and comprehensive business intelligence strategy which
addresses these challenges.

This article is an excerpt from Data Warehousing and E-Commerce,
Copyright 2001, Prentice-Hall PTR.

Share this post

William Lewis

William Lewis

Wiliam has more than 20 years’ experience delivering data-driven solutions to business challenges across the financial services, energy, healthcare, manufacturing, software and consulting industries. Bill has gained recognition as a thought leader and leading-edge practitioner in a broad range of data management and other IT disciplines including data modeling, data integration, business intelligence, meta data management, XML and XSLT, requirements structuring, automated software development tools and IT Architecture. Lewis is a Principal Consultant at EWSolutions, a GSA schedule and Chicago-headquartered strategic partner and systems integrator dedicated to providing companies and large government agencies with best-in-class business intelligence solutions using enterprise architecture, managed meta data environment, and data warehousing technologies. Visit William can be reached at

scroll to top
We use technologies such as cookies to understand how you use our site and to provide a better user experience. This includes personalizing content, using analytics and improving site operations. We may share your information about your use of our site with third parties in accordance with our Privacy Policy. You can change your cookie settings as described here at any time, but parts of our site may not function correctly without them. By continuing to use our site, you agree that we can save cookies on your device, unless you have disabled cookies.
I Accept