Description
The Top 10 Critical Challenges for Business Intelligence Success
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Atre Group, Inc. Written by: Shaku Atre
303 Potrero Street, #29-303 Published in Computerworld
Santa Cruz, CA 95060
[email protected]
www.atre.com
.
The Top 10 Critical Challenges for Business Intelligence Success
More than half of all BI projects fail – make sure yours isn’t one of them.
Let’s start with the bad news: More than half of all Business Intelligence projects are either never completed or fail
to deliver the features and benefits that are optimistically agreed on at their outset. While there are many reasons
for this high failure rate, the biggest is that companies treat BI projects as just another IT project. Face it: Business
Intelligence, or BI, is neither a product nor a system. It is, rather, a constantly evolving strategy, vision and
architecture that continuously seeks to align an organization’s operations and direction with its strategic business
goals.
With BI, business success is realized through rapid, easy access to actionable information. This access, in turn, is
best achieved through timely and accurate insight into business conditions and customers, finances and markets.
Complex stuff, but worthwhile. Successful BI brings greater profitability, the true indicator of business success. And
success is never an accident; companies achieve it when they do the following:
• Make better decisions with greater speed and confidence
• Streamline operations
• Shorten their product development cycles
• Maximize value from existing product lines and anticipate new opportunities.
• Create better, more focused marketing as well as improved relationships with customers and suppliers alike.
Organizations must understand and address these 10 critical challenges for BI success. BI projects fail because of:
1. Failure to recognize BI projects as cross-organizational business initiatives, and to understand that as such
they differ from typical stand-alone solutions.
2. Unengaged business sponsors (or sponsors who enjoy little or no authority in the enterprise)
3. Unavailable or unwilling business representatives.
4. Lack of skilled and available staff, or sub-optimal staff utilization.
5. No software release concept (no iterative development method)
6. No work breakdown structure (no methodology)
7. No business analysis or standardization activities.
8. No appreciation of the impact of dirty data on business profitability.
9. No understanding of the necessity for and the use of meta-data.
10. Too much reliance on disparate methods and tools (the dreaded silver bullet syndrome).
In this white paper, we examine each of these challenges.
1. Cross-Organizational Collaboration
Traditionally, any business initiative, including a decision-support project, was focused on a specific goal that was
limited to a set of products or an area of the business. Due to this narrow focus, organizations were unable to
analyze the project’s impact on business operations as a whole. As Organizations became more customer-focused,
these initiatives began to integrate customer information with product information.
It is critical to realize that customers and markets, not manufacturing plants and product managers, must drive the
business. It is also optimal to correct any customer problems before the customer realizes the problem existed.
Enterprises have a better chance to achieve high customer loyalty if customers can pay when their problem is
solved – not when the product is shipped. Initially, the integration occurred in regional or departmental databases,
with no cross-regional collaboration.
®
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Enterprise data warehouses were the next step in the evolution toward cross-organizational integration of
information for decision-support purposes such as sales reporting, key performance indicators (KPIs) and trends
analysis. Customer relationship management (CRM) followed, bringing the promise of increased sales and
profitability through personalization and customization.
BI is the next step in achieving the holistic cross-organizational view (Figure 1). It has the potential to deliver
enormous payback, but demands unprecedented collaboration. Where BI is concerned, collaboration is not limited
to departments within the organization; it requires integration of knowledge about customers, competition, market
conditions, vendors, partners, products and employees at all levels.
To succeed at BI, an enterprise must nurture a cross-organizational collaborative culture in which everyone grasps
and works toward the strategic vision.
2. Business Sponsors
Strong business sponsors truly believe in the value of the BI project. They champion it by removi ng political
roadblocks. Without a supportive and committed business sponsor, a BI project struggles for support within an
organization – and usually fails.
Business sponsors establish proper objectives for the BI application, ensuring that they support the strategic vision.
Sponsors also approve the business-case assessment and help set the project scope. If the scope is too large,
sponsors prioritize the deliverables.
Specifically for BI projects, business sponsors should also launch a data-quality campaign in affected departments.
This task goes to business sponsors because it’s business users who truly understand the data.
Finally, business sponsors should run a project review session at assigned checkpoints to ensure that BI
application functionality maps correctly to strategic business goals, and that its return on investment (ROI) can be
objectively measured.
3. Dedicated Business Representation
More often than not, the primary focus of BI projects is technical rather than business-oriented. The reason for this
shortcoming: most BI projects are run by IT project managers with minimal business knowledge. These managers
tend not to involve business communities. Therefore, it’s not surprising that most projects fail to deliver expected
business benefits.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
It’s important to note that usually 20% of the key businesspeople use BI applications 80% of the time. Therefore, it’s
vital to identify key business and technical representatives at the beginning of a BI project – and to keep them
motivated throughout the project. A BI project team should have involved stakeholders from the following areas:
Business executives are the visionaries with the most current organizational strategies. They should help make key
project decisions and must be solicited for determining the project’s direction at various stages.
Customers can help identify the final goals of the BI system. After all, their acceptance of products or service
strategies is what matters most.
Key business partners provide a different view of the customer and should be solicited for information at the start
and on an ongoing basis.
The Finance department is responsible for accounting and can provide great insight into an organization’s
efficiencies and improvement areas.
Marketing personnel should be involved during all phases of the project because typically, they are key users of BI
applications.
Sales and customer support representatives have direct customer contact and provide customer perspective during
a BI project. They must have representation on the team.
IT supports the operational systems and provides awareness about the backlog of BI requests from different
groups. In addition to providing technical expertise, the IT staff in the BI project team must analyze and present BI-
related requests.
Operations managers and staff make tactical business decisions. They provide the link between strategic and
operational information, making them important during some key phases of a BI project.
4. Availability of Skilled Team Members
BI projects differ significantly from others because at their outset, they tend to lack concrete, well-defined
deliverables. In addition, the business and technical skills required to implement a BI application are quite different
than other operational online transaction processing (OLTP) projects. For example, while operational projects
normally focus on a certain area of the business, such as enterprise resource planning (ERP), customer
relationship management (CRM) or supply chain management (SCM), a BI project integrates, analyzes and
delivers information derived from almost every area of the business as a whole.
The required technical expertise varies as well; typically, for example, a database administrator’s focus is efficient
retrieval of data using OLTP systems. By contrast, where BI systems are concerned, it’s vitally important to focus
on data storage in addition to data retrieval.
A BI project team lacking BI application implementation experience will most likely fail to deliver desired results in
the first iteration. Since most BI projects have aggressive timelines and short delivery cycles, an inexperienced and
unskilled team is a risk that must be avoided.
Mandatory BI project skills include:
• BI business analysts who can perform cause-and–effect analysis to develop business process models for
evaluating decision alternatives. These individuals should also be able to perform what-if analysis by following
proven BI methodology.
• A KPI expert experienced in creating balanced scorecards. These experts must be able to identify the KPIs
that meet business needs, calculate and report them and monitor performance. They also should iteratively
re-evaluate KPI effectiveness and must integrate these KPIs into the balanced scorecard.
• Balanced scorecard experts to continuously develop and fine-tune scorecards. Measuring success in a
dynamic business environment requires an effective toolset. With a balanced scorecard, an organization’s
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
vision and strategy can be translated into objectives, targets, metrics – and incentives to meet those
objectives and targets.
• Data warehouse architects with experience developing BI-related logical and physical data models, including
both star schemas and OLAP. Ideally, these people might also have experience with such technologies as
statistical tools and data mining algorithms.
• Cube developers and implementers with experience implementing BI-specific data models, OLAP servers and
queries. These individuals must be able to develop and deploy complex and intelligent cubes to conduct
multi-dimensional OLAP analysis for different users.
• Personalization experts experienced at developing Web-based generic BI applications that can not only meet
the reporting needs of many users, but also provide a personalized view to each user.
5. BI Application Development Methodology
To succeed, BI projects must adhere to a plan with clearly defined methodologies, objectives and milestones. In
this respect, they are hardly unique. However, unlike other undertakings, BI projects are not limited to a confined
set of departmental requirements. Rather, their purpose is to provide cross-organizational applications. Therefore,
BI methodologies and deliverables differ.
Like any project, BI starts out by answering some basic questions, such as: What will be delivered? What are the
benefits and expected ROI? What is the total cost? When will it be delivered? Who will do it? The answers
collectively define the BI project as follows:
Project deliverables map goals to strategic business objectives. These deliverables should be measurable in
business terms. For example, “In order to increase sales 20%, the sales data merged with pipeline data must be
available to sales teams within three days of month’s end”.
Project scope aligns deliverables with BI application deployment phases and timelines. Unlike traditional
OLTP applications, the number of transactions the system will perform cannot measure BI project scope.
Transactions usually represent an organization’s processes, which in turn represent functions. Since BI projects are
data-intensive, not function-intensive, their scope must be measured by the data they will transform to the target BI
databases, and how quickly this data can be available. This focus on data is necessary because almost 80% of the
effort in a typical BI project is spent on data-related activities.
ROI for a BI project must be derivable from project deliverables. Project sponsors must measure the effectiveness
of delivered BI applications after the completion of each phase to determine whether the project is delivering the
promised ROI. If it isn’t, improvements must be made.
6. Planning BI Projects
Due to the nature of the beast, BI projects tend to hit more unknowns than OLTP projects that implement the
processes of organization, which in turn represent the functions. By contrast, BI projects are supposed to provide
data, which will be transformed into information, which in turn is transformed into action. Therefore, BI project
planning is not a one-time activity, but rather an iterative process in which resources, timelines, scope, deliverables
and plans are continuously adjusted (Figure 2).
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Although it’s an iterative process, the initial project plan must be created with as much detail as possible (Figure 3).
BI project planning activities include:
Determining project requirements. As part of this activity, existing high-level data, functionality and infrastructure
requirements must be reviewed and revised to include more detail and remove ambiguity.
Determining the condition of source files and databases. Before completing the project plan, operational data stores
must be reviewed to account for any issues that may surface during the data-analysis phase.
Determining or revising cost estimates. During this activity, the organization performs detailed analysis to determine
purchase and maintenance cost estimates for hardware, software, network equipment, business analysts, IT staff
members, implementation, training and consultants.
Determining or revising risk assessment. Enterprise must perform a detailed risk assessment in order to accurately
determine and rank BI project risks (based on severity and the likelihood of their occurrence).
Identifying critical success factors. Here an organization determines what conditions must exist in order for the
project to succeed. Factors include supportive business sponsors, realistic time frames and the availability of
resources.
Preparing the project charter. This is a detailed memorandum of understanding that should be prepared by the
project team and approved by the business sponsor and key business representatives.
Creating a high-level project plan. These are detailed breakouts of tasks, resources, time lines, task dependencies
and resource dependencies mapped on a calendar.
Kicking off the project. On completion of the plan, the project is kicked off in an orientation session at which all team
members, business representatives and the BI sponsor are present.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
7. Business Analysis and Data Standardization
By now it’s clear that BI projects are data-intensive and that “data out” is as important as “data in”. It’s crucial that
the source data be scrutinized. The age-old saying, “Garbage in, garbage out”, still holds true.
In most BI projects, business analysis issues are related to source data, which is scattered around the organization
in disparate data stores and in a variety of formats. Some of the issues include:
Identifying information needs. Most business analysts have challenges when it comes to identifying business issues
related to BI application objectives. They must evaluate how addressing these issues can help in obtaining answers
to business questions such as, “Why is there a decrease in sales revenue in the fourth quarter on the West Coast?”
Once the issues are identified, business analysts can easily determine related data requirements, and these
requirements can in turn help identify data sources for the required information.
Data merge and standardization. The biggest challenge faced by every BI project is its team’s ability to understand
the scope, effort and importance of making the required data available for knowledge workers. That data consists of
fragments in disparate internal systems and must be merged into a common data warehouse – not a trivial task.
Data requirements normally extend beyond internal sources, to private and external data. Therefore, data merge
and standardization activities must be planned and started at the beginning of the BI project.
8. Impact of Dirty Data on Business Profitability
Inaccurate and inconsistent data costs enterprises millions. It’s imperative to identify which data is important, then
find our how clean it is. Any dirty data must be identified, and a data-cleansing plan must be developed and
implemented.
The business objectives of any BI project should be tied to financial consequences such as lost revenue and
reduced profit. The financial consequences are usually the result of a business problem related to inaccuracies in
reports due to reliance on invalid, inaccurate or inconsistent data. However, most BI projects fail to tie financial
consequences to dirty data through monetary expressions (such as losing $10 million in quarterly revenue due to
the enterprise’s inability to up-sell)
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Even the best BI application will be worthless if driven by dirty data. Therefore it is important for every BI project to
employ knowledgeable business analysts who understand the meaning of source data and can ensure its quality.
Underestimating the data cleansing process in one of the biggest reasons for BI failure. Inexperienced BI project
managers often base their estimates on the number of technical data conversions required. Project managers also
fail to take into account the overwhelming number of transformations required to enforce business data domain
rules and business data integrity rules.
For some large organizations with many old file structures, the ratio of a particular data transformation effort can be
expected to be as high as 85% effort in data cleansing and only 15% in enforcing technical data conversion rules.
Therefore, even if estimates appear realistic at the project’s outset, you must factor in data-cleansing efforts. Note
that full-time involvement from the right business representatives is mandatory for data-cleansing activity.
9. Importance of Meta-Data
Clean data is worthless to knowledge workers if they do not understand its context. Valid business data, unless tied
to its meaning, is still meaningless. Therefore, it is imperative for all BI applications to consciously create and
manage the meaning of each data element. This data about data is known as meta-data, and its management is an
essential activity in BI projects.
Meta-data describes an organization in terms of its business activities and the business objects on which they’re
performed. It helps transform business data into information. It is imperative for every BI environment. For example,
what is profit? Does every businessperson have the same understanding of profit? Is there only one calculation for
profit? If there are different interpretations of profit, are all interpretations legitimate? If there are multiple legitimate
versions of profit, then multiple data elements must be created, each with its own unique name, definition, content
rules and relationships. All this information is meta-data.
Meta-data helps businesspeople navigate BI target databases and helps IT manage BI applications. There are two
types of meta-data:
• Technical meta-data provides information about BI applications and databases, and assists IT staff in
managing these applications.
• Business meta-data provides business users with information on data stored in BI applications and
databases.
Both types are crucial to success and should be mapped to each other and stored in meta-data repositories.
10. The Silver Bullet Syndrome
There is neither a single technology nor a technique that will resolve all the challenges to reach the goal of a
successful BI environment. That is to say, there is no silver bullet.
BI projects have an enormous scope and cover multiple environments and technologies. At a minimum, a BI
environment comprises:
• A tool for extracting, transforming and loading data from disparate source systems into the BI target data
warehouse.
• A data warehouse that stores historical and current business data, as well as an OLAP server that provides
analytic services.
• Front-end BI applications that are used to provide querying, reporting and analytic functions to the
organization’s knowledge workers.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
In most organizations, these BI components are implemented in different phases and by project teams. Each team
implements the product that meets most of its functional requirements. More tools create greater complexity and
increased interoperability issues, and require more administration involvement.
BI project teams must always consciously strive for the lowest possible number of tools. This will allow different BI
activities to map to the same overall roadmap.
Conclusion: Maximizing ROI
BI applications, if implemented efficiently and properly, have tremendous payoff. They can help an enterprise
increase its business agility, decrease operating costs and improve its customer loyalty and acquisition.
And in most cases these improvements bring a host of tangible benefits (better customer satisfaction, increased
revenue and profits, cost savings and higher market share). Bottom line: a successful BI project is a genuine, often
dramatic, improvement to any organization.
Ah, but there’s that word again: successful. As we’ve seen, many complex factors go into the successful BI project.
By paying attention to the 10 critical challenges for BI success, your enterprise has a great chance to complete and
deliver the features and benefits agreed upon at the beginning of the project.
About the Author:
doc_968103760.pdf
The Top 10 Critical Challenges for Business Intelligence Success
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Atre Group, Inc. Written by: Shaku Atre
303 Potrero Street, #29-303 Published in Computerworld
Santa Cruz, CA 95060
[email protected]
www.atre.com
.
The Top 10 Critical Challenges for Business Intelligence Success
More than half of all BI projects fail – make sure yours isn’t one of them.
Let’s start with the bad news: More than half of all Business Intelligence projects are either never completed or fail
to deliver the features and benefits that are optimistically agreed on at their outset. While there are many reasons
for this high failure rate, the biggest is that companies treat BI projects as just another IT project. Face it: Business
Intelligence, or BI, is neither a product nor a system. It is, rather, a constantly evolving strategy, vision and
architecture that continuously seeks to align an organization’s operations and direction with its strategic business
goals.
With BI, business success is realized through rapid, easy access to actionable information. This access, in turn, is
best achieved through timely and accurate insight into business conditions and customers, finances and markets.
Complex stuff, but worthwhile. Successful BI brings greater profitability, the true indicator of business success. And
success is never an accident; companies achieve it when they do the following:
• Make better decisions with greater speed and confidence
• Streamline operations
• Shorten their product development cycles
• Maximize value from existing product lines and anticipate new opportunities.
• Create better, more focused marketing as well as improved relationships with customers and suppliers alike.
Organizations must understand and address these 10 critical challenges for BI success. BI projects fail because of:
1. Failure to recognize BI projects as cross-organizational business initiatives, and to understand that as such
they differ from typical stand-alone solutions.
2. Unengaged business sponsors (or sponsors who enjoy little or no authority in the enterprise)
3. Unavailable or unwilling business representatives.
4. Lack of skilled and available staff, or sub-optimal staff utilization.
5. No software release concept (no iterative development method)
6. No work breakdown structure (no methodology)
7. No business analysis or standardization activities.
8. No appreciation of the impact of dirty data on business profitability.
9. No understanding of the necessity for and the use of meta-data.
10. Too much reliance on disparate methods and tools (the dreaded silver bullet syndrome).
In this white paper, we examine each of these challenges.
1. Cross-Organizational Collaboration
Traditionally, any business initiative, including a decision-support project, was focused on a specific goal that was
limited to a set of products or an area of the business. Due to this narrow focus, organizations were unable to
analyze the project’s impact on business operations as a whole. As Organizations became more customer-focused,
these initiatives began to integrate customer information with product information.
It is critical to realize that customers and markets, not manufacturing plants and product managers, must drive the
business. It is also optimal to correct any customer problems before the customer realizes the problem existed.
Enterprises have a better chance to achieve high customer loyalty if customers can pay when their problem is
solved – not when the product is shipped. Initially, the integration occurred in regional or departmental databases,
with no cross-regional collaboration.
®
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Enterprise data warehouses were the next step in the evolution toward cross-organizational integration of
information for decision-support purposes such as sales reporting, key performance indicators (KPIs) and trends
analysis. Customer relationship management (CRM) followed, bringing the promise of increased sales and
profitability through personalization and customization.
BI is the next step in achieving the holistic cross-organizational view (Figure 1). It has the potential to deliver
enormous payback, but demands unprecedented collaboration. Where BI is concerned, collaboration is not limited
to departments within the organization; it requires integration of knowledge about customers, competition, market
conditions, vendors, partners, products and employees at all levels.
To succeed at BI, an enterprise must nurture a cross-organizational collaborative culture in which everyone grasps
and works toward the strategic vision.
2. Business Sponsors
Strong business sponsors truly believe in the value of the BI project. They champion it by removi ng political
roadblocks. Without a supportive and committed business sponsor, a BI project struggles for support within an
organization – and usually fails.
Business sponsors establish proper objectives for the BI application, ensuring that they support the strategic vision.
Sponsors also approve the business-case assessment and help set the project scope. If the scope is too large,
sponsors prioritize the deliverables.
Specifically for BI projects, business sponsors should also launch a data-quality campaign in affected departments.
This task goes to business sponsors because it’s business users who truly understand the data.
Finally, business sponsors should run a project review session at assigned checkpoints to ensure that BI
application functionality maps correctly to strategic business goals, and that its return on investment (ROI) can be
objectively measured.
3. Dedicated Business Representation
More often than not, the primary focus of BI projects is technical rather than business-oriented. The reason for this
shortcoming: most BI projects are run by IT project managers with minimal business knowledge. These managers
tend not to involve business communities. Therefore, it’s not surprising that most projects fail to deliver expected
business benefits.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
It’s important to note that usually 20% of the key businesspeople use BI applications 80% of the time. Therefore, it’s
vital to identify key business and technical representatives at the beginning of a BI project – and to keep them
motivated throughout the project. A BI project team should have involved stakeholders from the following areas:
Business executives are the visionaries with the most current organizational strategies. They should help make key
project decisions and must be solicited for determining the project’s direction at various stages.
Customers can help identify the final goals of the BI system. After all, their acceptance of products or service
strategies is what matters most.
Key business partners provide a different view of the customer and should be solicited for information at the start
and on an ongoing basis.
The Finance department is responsible for accounting and can provide great insight into an organization’s
efficiencies and improvement areas.
Marketing personnel should be involved during all phases of the project because typically, they are key users of BI
applications.
Sales and customer support representatives have direct customer contact and provide customer perspective during
a BI project. They must have representation on the team.
IT supports the operational systems and provides awareness about the backlog of BI requests from different
groups. In addition to providing technical expertise, the IT staff in the BI project team must analyze and present BI-
related requests.
Operations managers and staff make tactical business decisions. They provide the link between strategic and
operational information, making them important during some key phases of a BI project.
4. Availability of Skilled Team Members
BI projects differ significantly from others because at their outset, they tend to lack concrete, well-defined
deliverables. In addition, the business and technical skills required to implement a BI application are quite different
than other operational online transaction processing (OLTP) projects. For example, while operational projects
normally focus on a certain area of the business, such as enterprise resource planning (ERP), customer
relationship management (CRM) or supply chain management (SCM), a BI project integrates, analyzes and
delivers information derived from almost every area of the business as a whole.
The required technical expertise varies as well; typically, for example, a database administrator’s focus is efficient
retrieval of data using OLTP systems. By contrast, where BI systems are concerned, it’s vitally important to focus
on data storage in addition to data retrieval.
A BI project team lacking BI application implementation experience will most likely fail to deliver desired results in
the first iteration. Since most BI projects have aggressive timelines and short delivery cycles, an inexperienced and
unskilled team is a risk that must be avoided.
Mandatory BI project skills include:
• BI business analysts who can perform cause-and–effect analysis to develop business process models for
evaluating decision alternatives. These individuals should also be able to perform what-if analysis by following
proven BI methodology.
• A KPI expert experienced in creating balanced scorecards. These experts must be able to identify the KPIs
that meet business needs, calculate and report them and monitor performance. They also should iteratively
re-evaluate KPI effectiveness and must integrate these KPIs into the balanced scorecard.
• Balanced scorecard experts to continuously develop and fine-tune scorecards. Measuring success in a
dynamic business environment requires an effective toolset. With a balanced scorecard, an organization’s
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
vision and strategy can be translated into objectives, targets, metrics – and incentives to meet those
objectives and targets.
• Data warehouse architects with experience developing BI-related logical and physical data models, including
both star schemas and OLAP. Ideally, these people might also have experience with such technologies as
statistical tools and data mining algorithms.
• Cube developers and implementers with experience implementing BI-specific data models, OLAP servers and
queries. These individuals must be able to develop and deploy complex and intelligent cubes to conduct
multi-dimensional OLAP analysis for different users.
• Personalization experts experienced at developing Web-based generic BI applications that can not only meet
the reporting needs of many users, but also provide a personalized view to each user.
5. BI Application Development Methodology
To succeed, BI projects must adhere to a plan with clearly defined methodologies, objectives and milestones. In
this respect, they are hardly unique. However, unlike other undertakings, BI projects are not limited to a confined
set of departmental requirements. Rather, their purpose is to provide cross-organizational applications. Therefore,
BI methodologies and deliverables differ.
Like any project, BI starts out by answering some basic questions, such as: What will be delivered? What are the
benefits and expected ROI? What is the total cost? When will it be delivered? Who will do it? The answers
collectively define the BI project as follows:
Project deliverables map goals to strategic business objectives. These deliverables should be measurable in
business terms. For example, “In order to increase sales 20%, the sales data merged with pipeline data must be
available to sales teams within three days of month’s end”.
Project scope aligns deliverables with BI application deployment phases and timelines. Unlike traditional
OLTP applications, the number of transactions the system will perform cannot measure BI project scope.
Transactions usually represent an organization’s processes, which in turn represent functions. Since BI projects are
data-intensive, not function-intensive, their scope must be measured by the data they will transform to the target BI
databases, and how quickly this data can be available. This focus on data is necessary because almost 80% of the
effort in a typical BI project is spent on data-related activities.
ROI for a BI project must be derivable from project deliverables. Project sponsors must measure the effectiveness
of delivered BI applications after the completion of each phase to determine whether the project is delivering the
promised ROI. If it isn’t, improvements must be made.
6. Planning BI Projects
Due to the nature of the beast, BI projects tend to hit more unknowns than OLTP projects that implement the
processes of organization, which in turn represent the functions. By contrast, BI projects are supposed to provide
data, which will be transformed into information, which in turn is transformed into action. Therefore, BI project
planning is not a one-time activity, but rather an iterative process in which resources, timelines, scope, deliverables
and plans are continuously adjusted (Figure 2).
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Although it’s an iterative process, the initial project plan must be created with as much detail as possible (Figure 3).
BI project planning activities include:
Determining project requirements. As part of this activity, existing high-level data, functionality and infrastructure
requirements must be reviewed and revised to include more detail and remove ambiguity.
Determining the condition of source files and databases. Before completing the project plan, operational data stores
must be reviewed to account for any issues that may surface during the data-analysis phase.
Determining or revising cost estimates. During this activity, the organization performs detailed analysis to determine
purchase and maintenance cost estimates for hardware, software, network equipment, business analysts, IT staff
members, implementation, training and consultants.
Determining or revising risk assessment. Enterprise must perform a detailed risk assessment in order to accurately
determine and rank BI project risks (based on severity and the likelihood of their occurrence).
Identifying critical success factors. Here an organization determines what conditions must exist in order for the
project to succeed. Factors include supportive business sponsors, realistic time frames and the availability of
resources.
Preparing the project charter. This is a detailed memorandum of understanding that should be prepared by the
project team and approved by the business sponsor and key business representatives.
Creating a high-level project plan. These are detailed breakouts of tasks, resources, time lines, task dependencies
and resource dependencies mapped on a calendar.
Kicking off the project. On completion of the plan, the project is kicked off in an orientation session at which all team
members, business representatives and the BI sponsor are present.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
7. Business Analysis and Data Standardization
By now it’s clear that BI projects are data-intensive and that “data out” is as important as “data in”. It’s crucial that
the source data be scrutinized. The age-old saying, “Garbage in, garbage out”, still holds true.
In most BI projects, business analysis issues are related to source data, which is scattered around the organization
in disparate data stores and in a variety of formats. Some of the issues include:
Identifying information needs. Most business analysts have challenges when it comes to identifying business issues
related to BI application objectives. They must evaluate how addressing these issues can help in obtaining answers
to business questions such as, “Why is there a decrease in sales revenue in the fourth quarter on the West Coast?”
Once the issues are identified, business analysts can easily determine related data requirements, and these
requirements can in turn help identify data sources for the required information.
Data merge and standardization. The biggest challenge faced by every BI project is its team’s ability to understand
the scope, effort and importance of making the required data available for knowledge workers. That data consists of
fragments in disparate internal systems and must be merged into a common data warehouse – not a trivial task.
Data requirements normally extend beyond internal sources, to private and external data. Therefore, data merge
and standardization activities must be planned and started at the beginning of the BI project.
8. Impact of Dirty Data on Business Profitability
Inaccurate and inconsistent data costs enterprises millions. It’s imperative to identify which data is important, then
find our how clean it is. Any dirty data must be identified, and a data-cleansing plan must be developed and
implemented.
The business objectives of any BI project should be tied to financial consequences such as lost revenue and
reduced profit. The financial consequences are usually the result of a business problem related to inaccuracies in
reports due to reliance on invalid, inaccurate or inconsistent data. However, most BI projects fail to tie financial
consequences to dirty data through monetary expressions (such as losing $10 million in quarterly revenue due to
the enterprise’s inability to up-sell)
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
Even the best BI application will be worthless if driven by dirty data. Therefore it is important for every BI project to
employ knowledgeable business analysts who understand the meaning of source data and can ensure its quality.
Underestimating the data cleansing process in one of the biggest reasons for BI failure. Inexperienced BI project
managers often base their estimates on the number of technical data conversions required. Project managers also
fail to take into account the overwhelming number of transformations required to enforce business data domain
rules and business data integrity rules.
For some large organizations with many old file structures, the ratio of a particular data transformation effort can be
expected to be as high as 85% effort in data cleansing and only 15% in enforcing technical data conversion rules.
Therefore, even if estimates appear realistic at the project’s outset, you must factor in data-cleansing efforts. Note
that full-time involvement from the right business representatives is mandatory for data-cleansing activity.
9. Importance of Meta-Data
Clean data is worthless to knowledge workers if they do not understand its context. Valid business data, unless tied
to its meaning, is still meaningless. Therefore, it is imperative for all BI applications to consciously create and
manage the meaning of each data element. This data about data is known as meta-data, and its management is an
essential activity in BI projects.
Meta-data describes an organization in terms of its business activities and the business objects on which they’re
performed. It helps transform business data into information. It is imperative for every BI environment. For example,
what is profit? Does every businessperson have the same understanding of profit? Is there only one calculation for
profit? If there are different interpretations of profit, are all interpretations legitimate? If there are multiple legitimate
versions of profit, then multiple data elements must be created, each with its own unique name, definition, content
rules and relationships. All this information is meta-data.
Meta-data helps businesspeople navigate BI target databases and helps IT manage BI applications. There are two
types of meta-data:
• Technical meta-data provides information about BI applications and databases, and assists IT staff in
managing these applications.
• Business meta-data provides business users with information on data stored in BI applications and
databases.
Both types are crucial to success and should be mapped to each other and stored in meta-data repositories.
10. The Silver Bullet Syndrome
There is neither a single technology nor a technique that will resolve all the challenges to reach the goal of a
successful BI environment. That is to say, there is no silver bullet.
BI projects have an enormous scope and cover multiple environments and technologies. At a minimum, a BI
environment comprises:
• A tool for extracting, transforming and loading data from disparate source systems into the BI target data
warehouse.
• A data warehouse that stores historical and current business data, as well as an OLAP server that provides
analytic services.
• Front-end BI applications that are used to provide querying, reporting and analytic functions to the
organization’s knowledge workers.
©Copyright 2003 by Atre Group, Inc.
Published in Computerworld – June 30, 2003
In most organizations, these BI components are implemented in different phases and by project teams. Each team
implements the product that meets most of its functional requirements. More tools create greater complexity and
increased interoperability issues, and require more administration involvement.
BI project teams must always consciously strive for the lowest possible number of tools. This will allow different BI
activities to map to the same overall roadmap.
Conclusion: Maximizing ROI
BI applications, if implemented efficiently and properly, have tremendous payoff. They can help an enterprise
increase its business agility, decrease operating costs and improve its customer loyalty and acquisition.
And in most cases these improvements bring a host of tangible benefits (better customer satisfaction, increased
revenue and profits, cost savings and higher market share). Bottom line: a successful BI project is a genuine, often
dramatic, improvement to any organization.
Ah, but there’s that word again: successful. As we’ve seen, many complex factors go into the successful BI project.
By paying attention to the 10 critical challenges for BI success, your enterprise has a great chance to complete and
deliver the features and benefits agreed upon at the beginning of the project.
About the Author:
doc_968103760.pdf