Business Intelligence Impact: the Assimilation of Business Intelligence into Core Business Processes
Leading proponents of business-driven business intelligence (BI) design and development methods have recognized that the BI payoff is created through improved or reengineered core business processes from BI investments (Williams and Williams, 2006), but until now there has been no rigorous empirical data to validate that perspective. As a result, companies in a wide range of industries have struggled to make the business case for BI, and have failed to drive BI use into the core business processes that help create business value.
Abstract
Leading proponents of business-driven business intelligence (BI) design and
development methods have recognized that the BI payoff is created through improved
or reengineered core business processes from BI investments (Williams and Williams,
2006), but until now there has been no rigorous empirical data to validate that
perspective. As a result, companies in a wide range of industries have struggled to
make the business case for BI, and have failed to drive BI use into the core
business processes that help create business value.
Two empirical studies were conducted at the University of Melbourne (Elbashir,
2007; Elbashir, Collier, and Davern, 2007) that broaden our understanding of how BI
creates business value. The studies found that:
- The assimilation of business intelligence systems (BIS) into business
strategies and activities of organizations delivers business benefits, or BI
impact.
- Improved business process performance, enabled by BIS, can improve
organizational performance.
We call these improvements in business process performance and organizational
performance BI impact. In this article, we examine the management
implications of BI impact on how BI is designed, justified, and managed.
Introduction
Organizations around the world are exhibiting a keen interest in BI, as is
clearly reflected in their increased spending to acquire the enabling technologies.
Despite large investments in BI and the potential benefits to organizations, there
is a distinct lack of rigorous, cross-industry empirical evidence that demonstrates
just how BI creates business value. As a result, BI teams struggle to make the
business case for BI to business executives, and they struggle to drive the use of
BI down into core business processes. They struggle to explain and achieve BI
impact.
BI used in core value-adding processes can directly improve business
performance and valuecreation. This is a far stronger foundation for making the
business case for BI than general assertionsabout 'making better decisions.'
A recent empirical study of 347 organizations conducted at the University of
Melbourne provides insight into how BI creates business value and how to measure
that business value (Elbashir 2007; Elbashir, Collier, and Davern, 2007). The study
found that BI enables organizations to generate business value at both the business
process level (customer-related processes, internal operations, and supplier-related
processes) and at the organizational level.
The study also reported that organizations will be able to generate business
value from their BI investments only when they use these tools to support their
business strategies and value chain activities. We refer to this process as BIS
assimilation. While the need for BIS assimilation might seem obvious, we
find in practice that many organizations do not take the steps necessary to ensure
that BI is leveraged within the business processes that make a difference.
These findings raise an important question. How can organizations that are
successful with business intelligence assimilate BIS into their business strategies
and activities? To begin to address this question, this article will:
- Describe a comprehensive measure of BIS assimilation that captures the
deployment of BIS in organizations that have used BIS to achieve business
benefits.
- Explore the practical implications regarding how BI is designed, justified,
and managed.
We define BIS as “interactive, computer-based systems for data
analysis and reporting which support business analysis and
decision-making.†After a brief description of the research, we will
explain the impact of BIS assimilation on business value creation.
The Research
The research findings confirm that BIS assimilation is a high-order performance
outcome—meaning the effects occur within the core business
processes. We identified three dimensions of BIS assimilation: (1) business
production and operations, (2) customer relations, and (3) marketing and sales.
Together with the measurement items used to capture each dimension, each of these
dimensions includes measurement items that represent generic strategies and related
business activities. For instance, items used to measure the
“customer relations†dimension of BIS assimilation
include delivery of products/services (business activity) and providing value-added
goods/services to customers (business strategy). The complete list of these
activities can be found in the following list:
BIS Assimilation Measures for Three Dimensions
1. Customer Relations
- Customer services (e.g., improving customer satisfaction)
- Delivery of products/services
- Enhancing customer relations
- Enhancing existing products/services
- Providing value-added goods/services to customers
- Creating new products/services
2. Business Operation
- Supplier management (e.g., inbound logistics or purchasing)
- Manufacturing and/or internal operations
- Being a low-cost producer/provider
- Creating flexible manufacturing/operations processes
3. Marketing and Sales
- Marketing (e.g., targeting customers and tailoring offers)
- Sales (e.g., sales-force automation, revenue management)
- Entering new markets
The research also found that BI used in core value-adding processes can directly
improve business performance and value creation. This is a far stronger foundation
for making the business case for BI than general assertions about
“making better decisions.†In practice, business
leaders have a hard time using “better business
decisions†to justify large BI investments. How does one measure whether
better business decisions are being made?
The research results allow us to move beyond weak business cases and imprecise or
nonexistent measures for BI assimilation/deployment. We can define BI investments as
business process improvement investments, and we can measure current and projected
improvements in business process performance and organizational performance.
Further, the research makes it clear that we have to manage BIS assimilation in
order to achieve BI impact. The practical implications of these findings are
presented in the following sections.
Managing for BI Impact
To achieve BI impact, the BIS components must be aligned with, and integrated
into, an organization’s business strategies and core value-adding
processes. In effect, BI and BI-driven improvements to targeted business processes
must be managed holistically. This has specific implications for the requirements,
design, business justification, program/project management, development, and the
locus of implementation responsibility.
More generally, we must manage BI and business process
convergence—or BI assimilation—which requires
an effective partnership between the business and IT. This too has specific
implications.
From a lifecycle perspective, the key tasks that must be accomplished to achieve
BI assimilation are illustrated in Figure 1.
Figure 1. Managing for BI impact across the system development
lifecycle
Essentially, achieving BI impact requires the effective management of BI
assimilation across the systems development lifecycle. This conclusion was borne out
by our survey results. There are other prerequisites, such as deployment of
appropriate and specialized IT infrastructure, but the focus here is on managing the
key steps and tasks required to integrate BI into core business processes that
impact organizational performance.
Accordingly, Figure 1 shows six lifecycle steps and the key tasks to be
accomplished. At each step, achieving BI assimilation as a means to realize BI
impact requires explicit consideration of business process as a central focus
for management attention.
Step 1: Define the Business Requirements
The first step in a business-driven approach to BI is to develop a comprehensive
and concrete understanding of the business opportunities to leverage BI within core
business processes that make a difference in organizational performance. In addition
to using appropriate BI opportunity analysis and portfolio management techniques, we
must give explicit and detailed attention to current-state and target-state business
processes and associated process performance metrics and goals. Essentially, we need
to answer four key questions at this stage:
- What business information do we need?
- To perform what business analyses?
- In support of which decisions?
- In the context of what business processes?
An overview of the basic analytical flow for identifying BI-driven business
process improvement opportunities is presented in Figure 2.
Figure 2. Business requirements analysis—BI
assimilation into business processes
In addition to identifying BI-driven opportunities to improve process and
organizational performance via analysis of the factors in Figure 2, we need to
capture the specific BI requirements. This can be done using proven BI
requirements-gathering methods.
In the context of business-driven BI design and development methods, BI
requirements are specified in terms of analytical subject areas, the business case
for the specific analyses, business questions, a fact-qualifier matrix, and a
definition of terms.
Building on the BI-driven process improvement opportunities identified above and
the specific BI requirements, we also need to characterize the current state and
target state of the business processes that are targeted for BI-driven improvement
or reengineering. For example, a leading financial services firm in the U.S. has
launched a major effort to improve its ability to cross-sell and up-sell to the
millions of individuals who are participants in employer-sponsored benefit plans. By
applying business-driven requirements analysis techniques, this firm identified
BI-driven process improvement opportunities in the areas of market segmentation,
campaign management, customer service, distribution, and cross-channel/cross-segment
message coordination.
The targeted processes reflected a typical mix of manual processes, custom data
extracts, spreadsheet applications, desktop databases, and coordination via ad hoc
e-mails. The vision was to combine BI, an enterprise data warehouse, business
process management technology, campaign management applications, improved and/or
reengineered business processes, and measurable process performance improvement
goals. This comprehensive and concrete view of business requirements provides a
solid foundation for BI assimilation and for making the business case for
investment.
Step 2: Make the Business Case
As the findings reported in Elbashir (2006) demonstrate, BI impact requires BI
assimilation, which means that BI must be integrated into organizational business
strategies and business processes.
The practical importance of this is clear: business
managers—not IT managers—should accept
responsibility for defining a BI-driven concept of operations for the targeted
business processes, developing business process change plans and cost estimates, and
developing ROI and sensitivity analyses. IT can provide cost estimates for the BI
portion of the business case, but BI impact can be achieved only via business
process change, which can be driven only by business managers. Accordingly, it makes
sense that business managers should lead the charge for building the business
case.
The difference between having IT develop a BI-centric business case around
“better decision-making†and having the business
develop a process-centric business case around leveraging BI is potentially profound
on several levels. First, it would reverse a common approach to BI capital
budgeting, whereby IT takes the lead on developing a business case with input from
the business about its requirements. The result is an IT-centric business case where
the nature and extent of business process change is not fully explored.
Business managers—not IT managers—should
accept responsibility for defining a BI-driven concept of operations for the
targeted business processes, developing business process change plans and cost
estimates.
Now that we have empirical data linking BI assimilation to business-value
creation, we have a defensible, nonpolitical argument for developing business cases
that are driven by a process-centric view.
The process-centric business case has several benefits, aside from the obvious
benefit of ensuring tighter linkage between targeted business processes and the BI
to improve or reengineer those processes. By requiring a process-centric business
case, organizations increase the likelihood of substantial business involvement in
the process. In effect, the business would own the process improvement project from
the outset, and the effort required to develop the business case would leave the
organization in a much better position to:
- Build consensus around the need for process change
- Build business buy-in for the targeted, BI-driven changes
- Identify a process performance baseline and/or process performance
benchmarks
- Establish process performance goals
- Build ROI analyses that business leaders believe
Step 3: Design the Business Process
For traditional BI projects, the design step of the systems development lifecycle
is well characterized and widely understood. A range of options, methods, and tools
are available for focusing on data architecture, data models, data presentation, and
metadata.
With the exception of business-driven design methods, most of these legacy BI
design methods pay inadequate attention to BI assimilation and the business-value
capture mechanism by which an investment in BI will result in a payoff for the
company. The findings reported in Elbashir (2006) suggest that this omission is a
potentially fatal flaw. If we don’t know up front how BI is going
to improve process performance or how we would measure such improvement, how can we
justify the investment?
A key implication of Elbashir’s results is that legacy BI
design methods need to be improved to incorporate business-driven methods that
explicitly consider business process design. Specifically, the design stage of the
lifecycle must focus on the envisioned target state of the business process to be
improved or reengineered, including specifications for:
- The organizations that will be involved in the target process and the
associated upstream and downstream interfaces
- The process roles that are needed and the departments or individuals to
whom they will be assigned
- The envisioned steps and activities within the process, such as the
horizontal and vertical work flows
- The BI elements needed by the process, including business information,
analytical tools and applications, and structured decision processes
- The content needed by the process, including reports, documents, e-mails,
forms, and so forth
- The business rules for the process
- Process control flows and process BI
- User interfaces and screen paths
The effort that must go into business-process design depends on the specific
BI-driven business-process improvement opportunity. In some cases, an existing
process may stay largely unchanged with the exception of adding BI at a specific
step. In other cases, a process must be designed from scratch. To illustrate, we
revisit the case of the leading financial services firm that has launched a major
effort to improve its ability to cross-sell and up-sell to the millions of
individuals who are participants in employer-sponsored benefit plans. As noted, the
firm used business-driven techniques to identify the need to improve:
- Market segmentation
- Customer service
- Campaign management
- Distribution
- Cross-channel, cross-segment message coordination
For the market segmentation process, only a handful of people are involved, and
they already have sophisticated data analysis software. The primary process change
required to leverage new BI is to have the same analysts point their tools at a
dependent data mart off of an enterprise data warehouse—instead
of using a hodgepodge of custom data extracts. In this situation, the business
process design effort is minimal.
At the other end of the spectrum, the process change associated with campaign
management will be substantial, and thus the process-design effort will be more
extensive. The firm currently operates on a business-to-business model, selling
benefit plans through distributors to businesses in a range of industries.
Accordingly, they conduct few direct marketing campaigns aimed at individuals, and
the current business processes are largely manual and based on custom data
extraction per selection criteria specified by marketing. As a result, campaigns
take weeks to plan, coordinate, and execute, and campaign performance measurement
requires time-consuming ad hoc analysis of custom data sets.
In the future, data about individual benefit-plan participants and their actions
will be integrated from multiple touchpoints into the enterprise data warehouse,
pushed to campaign-management applications, and then pushed out to a
campaign-analytics data mart. Accordingly, the process of selecting campaign lists
will be entirely different and highly automated, eliminating the need for custom
data extraction and ad hoc, cross-departmental coordination. Further, standard
campaign performance metrics will be delivered via the campaign-analytics data mart,
eliminating the need for ad hoc analysis of custom data sets.
These changes will require a comprehensive business process design with careful
attention given to all of the fundamental process design factors, including the need
for BI in the form of business information and analytical applications. The effort
will ensure that BI assimilation occurs.
Step 4: Develop the Business Process
For traditional BI projects, the development step within the systems development
lifecycle is well defined and widely understood. A wide selection of books,
educational opportunities, and proven methods cover development tasks such as
source-to-target mapping, data acquisition, data profiling, data transformation and
integration, data movement, data quality, metadata acquisition and movement, and
data delivery.
While new tools come into play from time to time— such as
master data management tools used to enhance data quality—the
basics of BI development have been encapsulated within mature technical processes.
That being said, we need to extend traditional BI development methods to include the
technical and non-technical methods associated with business process development if
we are to achieve BI assimilation and BI impact.
When we consider development in the business-process world, the scope and meaning
of the term depends on the approach being used to achieve business process
improvement and/or business process reengineering. For our purposes, we can
characterize the approaches along two key dimensions: (1) the degree of business
process change and (2) the degree of process automation. The process change approach
selected by a given company determines where it is positioned on these dimensions,
which then drives the key development tasks. This is depicted in Figure
3.
Figure 3. Business process development tasks depend on the scope of
change and degree of automation.
As a general framework, the matrix in Figure 3 presents a simplified view of what
are actually complex trade-offs, the results of which can place organizations in
differing positions within the four quadrants. (The quadrants
“define†basic strategies that dictate the key
business process development tasks that must be accomplished in the context of
achieving BI assimilation.) We will use an example to explore the development
implications of the four strategies.
Business Process Change at a Federal Government Agency
By law, by policy, and/or by choice, a federal government agency
we’ll call FGA enters agreements with external entities for which
it provides and/or receives certain business services. Within this general context,
approximately 2,500 agreements have been made, some dating back to the 1970s. These
agreements define the business and economic terms under which services are to be
provided. They fall into four primary categories:
- Programmatic agreements—free exchanges of information,
often initiated by FGA to serve its operational and/or statutory purposes
- Reimbursable agreements, where FGA provides services for monetary
compensation
- Interagency agreements, whereby FGA pays for services received, such as
payroll services received from the Department of the Interior
- Quid pro quo agreements, which are essentially barter agreements with
external parties
Today, these agreements are managed and executed via business and IT processes
that FGA has determined are insufficient to meet its business requirements. To
rectify this situation, FGA developed a comprehensive program plan for reengineering
the legacy agreements management business processes into a modernized agreement
process, or MAP. The plan incorporates the use of all four strategies to accommodate
the highly autonomous operations of the various organizational units that interact
within the current agreement management processes and that will continue to interact
under the MAP.
Strategy #1: Aggressive Leveraging of Business Process
Management
At the agency level, the MAP must achieve a higher degree of process control and
status visibility, and it must capture and deliver a richer selection of process BI
for regulatory compliance and process-improvement purposes. Given the scope and
complexity of the process, the FGA plan is to aggressively leverage modern business
process management (BPM) technology and methods at the agency level.
The implications of this approach for business process development are that (1)
process design will be done in a process modeling environment; (2) business rules
are specified and managed within the business rules engine; and (3) process
development occurs simultaneously with process design because the process modeling
tools automatically generate executable processes. In the BPM environment, BI needs
are specified within the flow of the process model, thus ensuring effective BI
assimilation. Overall, leveraging BPM in the context of BI-driven business process
improvement or reengineering is an excellent way to achieve the convergence of BI
and business processes that is necessary for BI impact.
Strategy #2: Incremental Leveraging of Business Process
Management
At the organizational unit level, aggressive leveraging of BPM does not make sense
for certain units. This is because those units already have effective business
processes to perform their role within the larger MAP. Accordingly, those units will
incrementally leverage BPM where it is necessary to interface with the agency-level
MAP and where they can benefit from improved BI regarding process costs and cycle
times.
They will interface their workflow control points with the MAP as the means of
receiving work assignments and passing back status information. They will also be
the beneficiaries of the process performance BI that will be generated by the
agency-level MAP. Given this incremental leveraging of BPM, the business-process
development tasks are less extensive, though they will still take place within the
BPM technical environment.
Strategy #3: Incremental Improvement with Traditional
Tools
For those organizational units where aggressive leveraging of BPM does not make
sense, there are also opportunities for incremental process improvement using
traditional tools. Such opportunities include:
- Business process improvement techniques such as Six Sigma, Lean, and
Quality Circles
- Standardized spreadsheet templates for financial estimating
- Standardized agreement forms
- Improved communication of process steps and responsibilities using tools
such as Visio
- Improved training
- Proven change management techniques
Under this strategy, the business process development tasks are driven by the mix
of tools used. In general, development consists of preparing documents,
spreadsheets, process maps, standard e-mail notifications, approval forms, policy
amendments, and so forth.
Strategy #4: Radical Improvement with Traditional Tools
For some of the organizational units at FGA, current business processes under the
broader agreement-management process will require a comprehensive overhaul, but
given the nature of their part of the work, such an investment in BPM does not make
sense. Rather, they will radically improve the business process by using traditional
tools. From a development perspective, the tasks are the same as for the incremental
improvement strategy.
Regardless of which business process change strategies are used, the development
stage of the BI-driven process improvement lifecycle requires attention to both BI
development and business process development. This is different than traditional BI
development projects where IT does the technical BI work and it is assumed that
the business will figure out the business process part. This is another
instance where business-driven design and development methods transcend the more
limited traditional methods.
Steps 5 and 6: Implementation and Continuous Improvement
During the implementation and continuous improvement phases of the systems
development lifecycle, BI assimilation again requires moving beyond the scope of
traditional BI lifecycle methods. In addition to the usual BI tasks, implementation
and continuous improvement of the improved and/or reengineered target business
processes must be aggressively managed. At this stage, business value will be
created from the BI and business process change investments. This is also the stage
where collecting process performance data is critical so that actual performance can
be compared with baseline performance, benchmark performance, and/or process
performance targets.
Summary
Achieving BI assimilation as a means to BI impact requires explicit consideration
of business process as a central management focus. Our discussion has
illustrated how the integration of BI and business process can be managed at each
stage of the systems development lifecycle. This approach means that those who
manage BI-driven process and profit improvement programs must expand the scope of
their management attention beyond traditional BI design and development
concerns.
It is also clear that many of the key tasks within stages and across the
lifecycle can be performed only by business people. More broadly, the task of
managing BI assimilation across the lifecycle has implications for the composition
and skills of the management team, the methods used, and the business and technical
skills required for success.
Traditional BI approaches have long argued for “business
involvement†but have stopped short of defining what that means. We
believe that the findings reported in Elbashir (2006) provide convincing empirical
evidence that we need more than “business
involvement†if BI investments are to pay off. Rather, we need BI-driven
process improvement initiatives to be owned and driven by the operational managers
who own the critical customerfacing, operational, and supplier-oriented processes
that are the key value-added processes of the firm.
Only these managers have the power and influence to drive new ways of doing
business that leverage BI. Only these managers can change the culture of information
usage within their units. Accordingly, these are the people who need to be held
accountable for the business value of business intelligence.
References
Elbashir, M. [2007]. Strategic Information Systems and Organizational
Performance: Exploring the role of Assimilation and IT infrastructure, 3rd
Asia/Pacific Research Symposium on Accounting Information Systems, Brisbane.
Elbashir, M., M. Davern, and P. Collier [2007]. “Business
Process Versus Organizational Performance Effects of Business Intelligence
Systems,†International Journal of Accounting Information
Systems, forthcoming.
Williams, Steve, and Nancy Williams [2006]. The Profit Impact of Business
Intelligence, Morgan Kaufmann.
This article originally appeared in the issue of TDWI.