Chapter 6

Intelligent business processes

with embedded analytics

INTRODUCTION

Businesses are made up of business processes. These business processes are

a series of activities with inputs, outputs, roles, and goals. Business pro-
cesses provide value externally (e.g., customers) and internally (e.g., staff).
Business optimization (BO) aims to achieve the process goals efficiently and
effectively by embedding artificial intelligence (AI). BO is a business initia-
tive that starts with the evaluation of the desired business outcomes. These
outcomes are described in the strategy and planning effort (discussed in
Chapter 3). The evaluation of business outcomes is followed by the devel-
opment of a set of process maps or models. These process models help in
understanding where and how to embed AI; the models also assist change
management. A model-based, performance-driven approach eases the
understanding and application of handset, dataset, toolset, and mindset
(discussed in Chapter 2) to business processes.
Processes are modeled keeping the goal of the user in mind. Serving a user
efficiently, providing an enhanced user experience, and having an efficient
and optimized supply chain are examples of user value. AI analyses data
and provide timely and accurate insights for optimized decision-making. In
turn, these business processes also provide value for the organization.
A process is the “manner in which” a suite of activities and tasks are car-
ried out. Processes are thus the “how” aspect of an organization’s functions.
The manner in which a bank account is opened, a patient is admitted to a
hospital, or a sales and marketing campaign is undertaken are examples of
processes. Business processes can become highly complex and intertwined
with processes from other collaborating organizations. Processes use tech-
nologies and people to achieve their goals. Embedding AI in business pro-
cesses makes them efficient and effective.
Intelligent business processes are flexible as they adjust to the needs of
their users quickly and interactively. This flexibility of a process provides
for business agility. Business agility provides accuracy, speed, and distrib-
uted decision points.1 Agility is enhanced by the speed and accuracy of
AI-based analytics to provide insights.

151
152 Artificial Intelligence for Business Optimization

Intelligence is also a factor of sensitivity of analytics. Insights are needed

within a given time beyond which they lose their agile value. This sen-
sitivity in analytics is AI’s ability to analyze increasingly finer granular
data in shorter time spans. Finer, granular analytics enhance the ability
of decision-makers to respond quickly and effectively to stimuli. Agility is
also the ability of these analytics to experiment with variables in real time.
An agile approach to embedding AI in business processes enables experi-
menting, showcasing, accepting feedback, and modifying the analytics to
suit the changing user needs. Agile principles and practices together with
analytics help the organization respond rapidly and accurately. Agile as a
culture anticipates issues before they arise. This is because of trust, vis-
ibility, and iterations in an Agile culture. The explicit (i.e., objective) and
tacit (i.e., subjective) aspects of decision-making are judiciously combined
in intelligent business processes.
Enterprise systems encompass subsystems, large amounts of heteroge-
neous data, information, people, technologies, diverse applications, and
processes. AI is applied to analyze Big Data in such enterprise systems.2
Big Data mining is similar to data mining but the scale of Big Data is much
larger.3 Challenges in the use of Big Data include storing and analyzing
large, rapidly growing data; handling diverse data sources and stores; and
deciding how to utilize that data.4
Excellence in business process modeling includes their unification. This
unification results in ease of use for the customer as a single point of contact
is created. Figure 6.15 shows a comparison between (a) knowledge workers
and customers using multiple silos of data and (b) empowered knowledge
workers and customers through self-serve analytics. Embedding Big Data
analytics in business processes enables business agility.
Enhanced knowledge sharing of insights occurs with formal process
models. Insights are generated and recorded within the business processes
which, in turn, are shared by other users. Process optimization reengineers
processes that change the way the organization operated. For example, the
cash withdrawal process uses digital technology to eliminate the printing of
a physical receipt. Analytics can reduce the time spent by a customer in the
queue for physical cash withdrawals by applying the algorithms to predict
the size and type of withdrawal transactions. Data analytics provide rich
insights that initiate and enable these changes.
Intelligent business processes also support a Lean-Agile organizational
structure in which the organization has a flattened hierarchy and decision-
making happens at the place of action. As a result, individuals at all levels of
the organization are empowered to take decisions based on their responsibili-
ties. This localized and decentralized decision-making is a major change in
the style of working resulting from AI.
Appropriate level of insights and creation of audit trails frees the knowledge
workers to make decisions. Intelligent processes support decision-making at
 Intelligent business processes 153

Knowledge workers accessing Big Data enabling collaborations

multiple silos and using through technologies and
disparate applications to satisfy across platforms thereby
business processes and perform empowering customer self-
routine transaction processing. services and value-addition to
Lack of Flexibility in Changing business processes of knowledge
Processes workers Resulting in AGILE
Business Processes

­Figure 6.1 Embedding AI in business processes resulting in agile business processes.

the lowermost rung of the organization. As both internal and external busi-
ness processes become optimized, a sustainable organization with reduced
environmental footprint emerges. Process models provide not only cost and
time benefits to the organization but also advantages in terms of environ-
ment and sustainability.
Business analysis is the modeling of these processes to identify oppor-
tunities to embed data analytics for optimization. BO is a fundamental
rethinking and radical redesign of business processes6 to achieve dramatic
advantages of business agility. Organization-wide, holistic, and dynamic
approach to optimization provides these dramatic business advantages.
Optimization of business processes requires the creation of process mod-
els using standards such as the BABOK of the IIBA.7 Process models are
created by business analysis (as against analytics) in order to examine the
activities and tasks that can be (a) made redundant due to analytics and (b)
enhanced in terms of their accuracy and speed. Business analysis deals with
changes to the way in which business is carried out.
Each business function and its processes are investigated by business
analysts for opportunities to embed data analytics. Process transformation
and change management follow these investigations. These changes include
embedding A I-based methods within business processes. Examples of these
methods as discussed by Dhar and Stein8 are use of genetic algorithms,
neural networks, expert reasoning, fuzzy logic, and case-based reasoning.
154 Artificial Intelligence for Business Optimization

BUSINESS PROCESS MODELING

Business processes are integral to BO. Business is organized around these

processes. A value-generating business focuses on the users, their goals in
interacting with the organization, and their perception of a good user expe-
rience. Business process models enable business optimization.

Business process modeling ( BPM) in BO

Embedding data analytics within a process requires modeling its goals, roles
(stakeholders), inputs, activities, and outputs. Each process model has the
end-goal of providing value to the user. For example, customer, teller, and
branch manager form the starting point for modeling a banking business
process. The end goal of this process is, say, a successful cash withdrawal.
The inputs, activities, and outputs are modeled as a process flowchart or an
activity diagram.9
Business processes can be external and internal to the business. Some
business processes form the core, others are the supporting processes of an
organization. External facing business processes provide value to an external
stakeholder, such as a customer in a bank or a passenger in an airline. Use
cases10 for external-facing processes include “Customer withdraws Cash,”
“Passenger Checks-In a flight,” and “Patient admits to a hospital.” Internal
business processes satisfy the needs of an internal stakeholder, such as a branch
manager in a bank or a scheduler in an airline. Use cases for internal-facing
processes include: “End-of-Day
­­­­ ­­ Cash-in
­­ and Cash-out,”
­­ “from
­­ the branch,”
“Scheduling
­­ of flights,” and “Re-order
­­­­ antibiotics.”
Internal business processes supporting the business include operations
and management (OSS), production (parts of ERP), HR and people (HRM),
legal and compliance (GRC), and Carbon Emission Management Systems
­­
(CEMS).
External business processes are facing the customers and suppliers and
deal primarily with the growth of the organization. Typical examples include
CRM, SCM, ERP, and BSS. The entire suite of processes is studied and mod-
eled as part of BO.
The analytics within these processes can remain at a higher macro level
or, alternatively, drill down to a fine granularity. Coarse, granular ana-
lytics work on large volume, relative static historical data. This is called
descriptive analytics which provides a historical understanding of process
performance. Fine granular analytics provide insights for micro actions
and snap-decisions with pinpoint accuracy – and are usually predictive in
nature. Fine granular analytics need current data and the present business
context in which decisions are made.
Analytics embedded within properly modeled business processes is the key
to BO. Process models help in deciding which type of analytics (descriptive,
 Intelligent business processes 155

predictive, or prescriptive) will provide the highest value to the user. Process
models also assist in determining the level of granularity. Optimization of
business processes is an ongoing activity that improves performance with
each iteration.

Change management processes

Processes also enable, aid, and support the transition of an organization to
AI-enabled optimized business. These are change management processes that
are not an internal or external business process. Instead, they are “supporting”
processes that help the organization with its BO effort. These adoption pro-
cesses impact, change, and transform the internal and external business
processes. BABOK 3.011 contains a detailed description of “The Transition
Requirements” for the business processes and associated documentation.
The organizational structure (hierarchy) also plays a part in the agility of a
process. The primary stakeholder in a process is involved in “obtaining” value,
and stakeholders from within the organization are involved in “providing”
value. In a data-driven business, processes are executed based on insights
from data analytics. Each process is also extensively supported by technolo-
gies, systems, IoT devices, and security and privacy. These aforementioned
support elements need to collaborate in order to optimize processes.
Successful collaboration requires the process modeler to pay attention to
the stakeholders, their activities, the dependencies between activities, and
the end deliverables. Involvement of these key external stakeholders earlier
in the adoption helps undertake collaborative arrangements (and agree-
ments) for business processes executed across multiple organizations. The
end result is a suite of lean and agile business processes with embedded ana-
lytics. Supporting processes enable further sharing of problems, solutions,
and knowledge across the organization. Finally, training the staff to use the
process is a mindset requirement.

Composite agile method and strategy (CAMS)

The techniques (practices) of Agile play an important role in modeling
the new digital business processes with embedded analytics. Composite
Agile Method and Strategy (CAMS)­­12 combines Agile techniques together
with the formal, planned methods for process modeling. Composite Agile
Method and Strategy (CAMS) in BO provides the benefits of visibility and
transparency, change management, integration of solution, and quality
through continuous testing and showcasing.
This use of composite Agile enables iterative and incremental approach
to process modeling. The Agile iterative approach also helps an organi-
zation build its technical and analytical skills and capabilities for BO.
Iterations and showcasing of process models (following Agile) also involve
156 Artificial Intelligence for Business Optimization

collaborating business partners and supporting technology vendors. These

iterations help reduce the risks in BO. Agile as a method also provides an
understanding of the capability and maturity of the organization. Changes
are continuously showcased to the users. Initial experimentation, alignment
of analytics with business processes, and alignment of technologies with
Enterprise Architecture (EA)
­­ ­­13 reduce the risks in BO.

Process maturity of an organization is another important factor in reduc-

ing risks. This maturity of the organization undertaking BO is ascertained
by examining the technical capabilities, economic strength, people skills
and attitude, and the business processes of the organization.

BUSINESS PROCESS AGILITY

Agile is important in the discussion on business processes and their opti-

mization. Optimized business processes are agile. Agile started an itera-
tive and incremental approach to developing software solutions. Agile has
evolved into a business characteristic.

­­L ean-agile processes

An optimized business is both lean and agile. Analytics enable these busi-
ness characteristics to flourish. Lean approaches are also used for process
optimization, outsourcing strategies, and greening an organization14 – and
their impacts studied on enterprises, government, and society. Process reen-
gineering considers “Lean” as a means to reducing and/or eliminating wast-
ages within the organization processes. Large and global organizations
benefit immediately through “Lean” as they tend to have substantial wast-
ages within their inventory and supply chains. For example, large-scale pro-
cesses and their corresponding value-streams (such as those from mining,
agriculture, and airlines sector) can be reengineered by introducing AI-based
analytics in them, applying metrics and measurements, and training their
users. Agile as a culture also helps the business processes of an organization
to become Lean.15 This leanness of business processes is the result of ana-
lytics embedded within the processes. This is so because analytics improve
accuracy, reduce time, and decentralize decision-making. As a result, many
redundant activities are identified and removed from a business process.

Visibility and transparency

While Agile can help produce faster and higher-quality solutions, these solu-
tions need to be made visible to the end user while they are getting devel-
oped and not at the end of their development. Agile makes the analytical
models and the corresponding business processes immediately visible. The
 Intelligent business processes 157

Kanban16 boards in agile enable the users to see progress of the develop-
ment and provide immediate feedback. This visibility also provides trans-
parency of the progress of BO to all stakeholders.

Change management
Agile methods welcome change rather than resist it. This attitude incul-
cated by Agility is most helpful in transforming business processes. This is
because the change in the business processes due to embedding AI in them
is incremental and transparent to all stakeholders.

Integration solutions
­­

Quality through continuous testing and showcasing

Testing includes functional accuracy and nonfunctional or operational
quality (e.g., performance, security, and usability). Quality of business pro-
cesses needs verification and validation as they change through embedded
analytics. Such comprehensive testing and continuous showcasing are facili-
tated by Agility in developing AI solutions. Testing the business processes
in an “operational” environment is vital for satisfying user experience, and
that happens incrementally through Agile.

DATA ANALYTICS AND BUSINESS AGILITY

Agility is the flexibility within the organization’s processes. Agile business

changes rapidly in response to changing business situations. Data analytics,
embedded with the business processes, are able to provide both descriptive and
predictive insights to the decision-makers. As a result, businesses can antici-
pate challenges and opportunities and improve their responses accordingly.
Lean and Agile are business characteristics that are enhanced by the applica-
tion of analytics to the business processes. Business processes are supported by
analytics, technology, infrastructure or applications, and skilled users.
Embedding analytics within business processes changes the shape of the
process, allows a business to focus on its core offerings, and “off loads”
158 Artificial Intelligence for Business Optimization

noncore processes to supporting partners. This offloading results in lean-

ness of business processes, which also makes them agile.17

Decentralized ­­decision-making
Embedding analytics in decision-making at all levels of a business decen-
tralizes the decision-making process. This makes a business more respon-
sive as many decisions, guided by analytics can be taken immediately.
Organizations are better able to adapt to changes in their environment due
to decentralization.

Finer granularity in business response

Fine granular analytics improve agility and accuracy of a business response.
The finer the granularity, the more personalized is the response. This is so
because finer granularity provides pinpoint accuracy for decision-making.

Elimination of redundancies
Removal of business processes that have otherwise become entrenched
within the organization and may be redundant occurs due to formal model-
ing and examination of those business processes. Activities and processes
that do not add value to their users and which are duplicated elsewhere in the
organization can be removed as part of reengineering of the organization.18

Enhancing sustainability in operations

Analytics are instrumental in making the processes lean. This is so because,
with the help of analytics, processes eliminate wastage and redundancy.
This reduction in wastage and improved monitoring and control of the pro-
cess also reduce the carbon footprint of the organization. AI plays a role in
improving the sustainability of the organization.19

Risks, compliance and audit requirements

­­Governance-risk-compliance
­­ (GRC),
­­ quality, and audit support processes
ensure that Big Data analytics embedded in the internal and external busi-
ness processes are compliant with the many legal and privacy requirements,
are able to manage security and privacy of the data within them, and are
subject to traceability and audit.
Complex regulatory requirements include security, compliance, audit, and
risk management – and they consume substantial business resources. An orga-
nization has to specifically consider and satisfy these requirements. CAMS
makes provision for a detailed documentation to satisfy these requirements.
 Intelligent business processes 159

Disaster recovery (DR)

­­
Data analytics enable predictions of disasters based on (a) past historical
data of an organization, (b) external third party from the industry, and
(c) alternative data sourced from vendors. DR also includes elements of
prescriptive analytics as it recommends actions to be taken in the event of
a disaster.

BUSINESS ANALYSIS & REQUIREMENTS MODELING

Business analysis (BA) is a discipline in its own right. The IIBA 20 provides a
framework for BA work that extends beyond requirements modeling (RM).
Creating process models of the business processes is important in the BO
exercise. This is because BO is business oriented and not limited to tech-
nologies of AI. A few important BA techniques that help BO include critical
thinking, the art of questioning, and mind mapping are described here.

Critical thinking in BPM

Critical thinking 21 is undertaken to arrive at the root cause of an issue and
is a core activity of business analysis. Challenging current thinking and
beliefs to determine what is true, partially true, or false is important during
process modeling. For example, a business analyst thinking critically will
question the purpose of a process, the relevance to the stakeholders, and the
analytics to be used. Critical thinking underlies the other business analyst
­­problem-solving thinking modes. 22
System thinking accompanies critical thinking as it views problems as a
whole system. It is holistic and, therefore, more strategic. Through systems
thinking, the business analyst views the organization and its business pro-
cesses in its entirety to truly understand the impacts of the use of A I-based
analytics and of the changes to the organization.
Critical and system thinking enable BA to handle the long-term impacts
of AI adoption.

Art of questioning
Business analysis starts by asking the right questions related to a business
process. Questions are asked in the right way and at the right time to the right
people.23,24 Questions typically include What? Why? When? Where? How?
Who? BA’s base their questions on clarity, precision, accuracy, relevance, and
depth. Such questions help understand ambiguity and complexity within the
business processes. Business analysts ask questions, listen to the answers,
document and model them, and use the results for appropriate actions.
160 Artificial Intelligence for Business Optimization

The purpose or goal in asking questions can be divided roughly into two
categories:
Understanding the Problem – to ascertain the business needs, capabilities,
requirements, and expected levels of quality in a business process. Needs
analysis, root cause analysis, SWOT25 and PESTLE26 analysis, and criti-
cal thinking are examples of techniques used in exploring a business prob-
lem. These techniques benefit from a suite of questions that are themselves
exploratory and subjective in nature. Techniques in requirements modeling,
such as user stories and use cases, also require the business analyst to employ
the art of questioning to arrive at functional models of the requirement.
Providing the Solution – the purpose of these questions is to arrive at an
appropriate solution to the problem explored. Solutions encompass archi-
tecture, design, developing, testing, and deployment in a typical software
project. Examples of standards used in a software solution production
include UML, 27 agile (methods), TOGAF, 28 development environments test
standards (e.g.,
­­ ISTQB), 29 and deployment approaches. Solution space ques-
tions are far more objective and narrower than in problem exploration as
they focus on a particular technology used in the solution.

Machine learning to frame questions

The art of questioning undergoes some modifications in the AI model.
Traditionally, businesses have a rough idea of what the problems are and
what they do not know. With Big Data, however, both the problems and
the possible hidden answers in the data mass are unknown. For example,
business leaders can ask the data scientists (or equivalent roles) whether it is
worth exploring customer demographics to figure out customer attrition or
potential growth or another potential topic. Can the data itself give an indi-
cation of what can be asked of it? This is precisely where machine learning
(ML) has a role to play.
The size and complexity of Big Data is such that building the models of
solutions from Big Data is not enough; tools are needed to handle the answers
and also frame new questions. With Big Data, business decision-makers
struggle to figure out what questions to ask. With the exponential growth
in data, it becomes important to not only provide better answers to business
questions, but also help businesses understand what types of questions they
must ask of their data.
ML algorithms can dive deeper into data in order to identify patterns that
are not possible to discern with traditional analytical approaches. This is
because of the multilayered or tiered nature of hidden, vast amounts of data.
Algorithms need to be created in a manner that creates learning through
the execution and provides insights. Thus, ML is the dynamic use of algo-
rithms, which provides a range of opportunities ranging from learned user
behavior to dynamic cyber defense.
 Intelligent business processes 161

ML algorithms search through massive amounts of data to find relevant

and interesting questions that the collaborative business can ask to improve
its overall value proposition. Such crucial questions are the result of iden-
tifying insightful similarities and differences within a large, multidimen-
sional dataset that is related or connected with each other in a collaborative
enterprise system (CES).
­­ 30

Mind mapping
A mind map is an important BA tool. Mind map is used in BO to show the
relationship between data, concept, user, and the process. Mind map is an
excellent tool for visual thinkers who are trying to establish the business
value of data. Business analysts think, analyze, model, and present processes
and their elements in images and pictures rather than words. Business ana-
lysts use the graphical representations of mind maps to visualize the prob-
lem and the corresponding solution. Mind map provides the opportunity for
communication very well.
A mind map starts with the central idea of a data item, a concept, or a
business process. Concepts are added as extensions to the central concept
developed. Process models and requirements documents are better organized
in BO by drawing the mind map. This is so because mind maps shift the
focus of BO to business and away from AI technologies. Figure 6.2 shows
the example of a mind map. This particular example starts with a central
theme: currency options predictive analytics. Such predictive analytics needs

Risk, aptitude Descriptive

and strategy Emotion Historical
Exchange
Rate Economy
Indicators

Trades and
other users
(clients)

Currency Path of
Options Pandemic - Predictive
Predictive (Covid19)
Analytics

Daily?
Election news Hourly?
Automated? Commodities
and pointers and stock
indicators

Latest tweets Predictive

­Figure 6.2 Mind map of a data-driven business process in financial market prediction.

162 Artificial Intelligence for Business Optimization

data such as historical exchange rates, the economic indicators, path of pan-
demic, commodities and stock indicators, election news and pointers, trades
and other users and risk, aptitude, and strategy. Note how the central theme
or concept is expanded based on an extension of the previous node. This
expansion of a mind map is also iterative and incremental, following the
Agile approach to developing BO solutions.

Comparison of processes for gaps

Use cases, activity diagrams, and mind maps are the tools to create a model
of the business processes. These process models can be created for the cur-
rent and future processes. The future processes are the ones that are embed-
ded with A I-based analytics. The gaps are identified by comparing process
models. Metrics are used in BO in order to (a) justify the initiative to embed
data-driven decision-making in the business, (b) prove the success of the ini-
tiative, and (c) reduce risks. Metrics also help in comparing process models.
Figure 6.3 shows an example of a basic comparison of a patient check-in
process with and without AI. Process models help in identifying gaps in the
value proposition of the process.
Process models keep the focus of the BO initiative on customer value.
Treating AI implementation as a technology project is a risk. This risk is
ameliorated by use of the business analysis (as against analytics) capabilities
in BO.

Patient check in process

without data analytics

Identify Actions and

Verify Details Diagnosis Insights
Patient Consequences

Customer

Staff

Actions and
Pre-verified AI Diagnosis Insights Value
Consequences

Patient check-in AI
enabled Auto Feedback

­Figure 6.3 Comparison of current process with a digitally enabled smart process.

 Intelligent business processes 163

Managing business system changes

While the business processes, represented by the use cases, activity diagrams,
and mind maps, change with embedded analytics, the various systems to
support the business functions have to absorb newer and finer granular ana-
lytics that support changes to the business processes. Furthermore, collabor-
ative business processes expose an organization’s systems and applications to
those of the collaborating partners. This embedding of analytics and exten-
sion of electronic collaboration changes the systems and their interfaces, and
requires additional focus on the cybersecurity aspects of the systems.
Each analytic embedded in the above processes leads to further levels of
analytics. Examples of this further analytical work include enabling cus-
tomer segmentation and understanding their attrition rates, working out the
formation of communities in order to improve product acceptance and use,
running targeted marketing campaigns through multiple channels, develop-
ing accurate strategies for cross- and up-selling, and optimizing business
processes.
Following business functions and systems undergo a change due to busi-
ness optimization using AI and data analytics:

• Customer relationship management (CRM) – Changes the way prod-

ucts are priced based on a detailed understanding of customer buying
patterns and profiles. Customer analytics search for data based on a
customer footprint on social media and mobile to improve customer
understanding. This understanding includes the customer demo-
graphics, thereby improving the organization’s ability to provide such
customers with a holistic view of their own requirements and enabling
the organization to strategize for the satisfaction of its customer base.
• Marketing and promotions are based on fine granular analytics and
targeted at specific customers. Data analytics provide a strategy to mar-
ket a product to a fine cross section of customers and then replicating
it over a large cross section of other customers. Learnability of ana-
lytical solutions is an important aspect of agility in marketing. Small,
experimental introduction of marketing effort, receiving instantaneous
feedback, and incorporating that feedback in subsequent iteration of
marketing effort is the way this analytics is leveraged for business agil-
ity. Customer sentiments are based on Net Promoter Score (NPS)­­31 and
can help ascertain what the customers think and feel about the product.
As a result, products and services can be changed quickly to accommo-
date the changing sentiments. Assessing customer sentiments using AI
and ML automated tools and presenting appropriate products to them
are major changes.
• Analyzing prices indicates how to prioritize and sell products based on
discriminatory pricing. The creation of such fine, discriminatory pric-
ing that can change from moment to moment and from one individual
164 Artificial Intelligence for Business Optimization

to another is a vital element of business agility. Since customers can

also undertake self-service analytics, the face-to-face contact with the
business is further reduced.
• Supply chain management (SCM) – AI helps in optimizing the inven-
tories and understanding the logistics and deliveries. This optimiza-
tion changes the way the SCM works. Supply chain processes are
based on data associated with quality, fault reports, and also product
recalls should a product fail a compliance need. These analytics can
also assist in process automation as they enable changes to the product
directions in an automated or semiautomated manner depending on
the feedback being received. This optimization also reduces wastage
due to poor inventory management. Collaboration and partnership
with suppliers are established through negotiations and contracts as a
result of use of analytics electronic implementation of these political
speeds the SCM processes.
• Enterprise resource planning (ERP) – AI-based analytics improve the
understanding, creation, launching, management, and withdrawal of
products and services. Insights generated by Big Data analytics help
in optimizing these ERP functions by ensuring their timelines and
accuracy. Products are added or removed in an agile, iterative manner
with the help of analysis.
• Sustainability applications – usually under the umbrella of Carbon
Emission Management System (CEMS) – These functions help handle
corporate responsibility and sustainability requirement by accurately
calculating and controlling the carbon footprint of the organization.
Data-driven insight enables justification for environmentally conscious
decisions by the business. Data is able to show to the decision-makers
the carbon footprint of each business activity and the organization.
Data analytics also correlates the carbon footprints to the risks of
fines due to noncompliance.
• Billing support system (BSS) – Billing, finance, and accounting systems
change with Big Data as the analytics are able to spot trends and advise
the accounting functions in terms of anticipated changes, trends in bill-
ing errors, and where to focus the resources to improve billing efficien-
cies. These changes have a direct impact on delivering customer value.
• Operations and support system (OSS) – Applications, networks, data-
bases, scheduling, security, and other operational support provided
by the systems are analyzed with the help of data – typically collected
through IoT sensors. Data analytics provide a formal, data-driven
approach to maintenance and upkeep of production equipment. This
can be an important type of analytics with applications in aircraft main-
tenance, bridge repairs, and large residential and industrial structures.
• HR and people (HRM) – Organizational structure, motivation,
morale, rewards, KPIs, and collaboration and partnership with external
 Intelligent business processes 165

organizations are all affected by the use of data changes. Recruitment

itself gets shifted to a sophisticated analytical filter that goes way
beyond the keyword searches.
• Governance, risk, and control (GRC) – Through accurate analytics
that enables the organization to comply with regulation (such as with
SOX legislation). Risks are identified much earlier than they occur due
to predictive analytics. Systems are also able to imitate autorecovery
functions on some processes in the organization.

EMBEDDING ANALYTICS IN BUSINESS PROCESSES

Embedding analytics in business processes requires (a) modeling of pro-

cesses and (b) understanding of the analytics. The range and types of ana-
lytics and corresponding supporting technologies create opportunities for
BO. Data analytics are traditionally divided into descriptive, predictive,
and prescriptive analytics. These analytical categories are based on their
purpose and use in business. For example, predictive analytics plots a trend
and attempts to show the position of a point in time (future). This type
of analytics would require assembling large historical data, integrating it
with current data, developing experimental analytical models, and incor-
porating the feedback from showcasing the results. As a result, this type of
analytics may not be ideal for situations requiring real-time, fi ne-granular
insights. Alternatively, descriptive analytics will be more ideal for the afore-
mentioned scenario.

Preparing the data

Most practical analytics are performed on both structured applications (e.g.,
existing CRM, ERP) and unstructured information sources from Web pages,
blogs, and contact centers. Social media and mobile (SoMo) are rich sources
of alternative data. Such unstructured data requires preparation before it can
be analyzed. MapReduce, “R,” Python, and similar tools process unstruc-
tured information, add structure to it, and set a stage for its analysis. The
extent to which the unstructured data needs to be “prepared” for analyt-
ics is an important consideration in BO. This section discusses categories of
analytics in order to identify their appropriateness in practice. These catego-
rizations also indicate how the analytics will benefit the business. The same
analytical tool or application can be used in multiple analytical categories.

Data analytic types and relevance in BO

Analytical categories help in understanding how they can be applied in
practical BO. Analytics helps an organization become and remain Agile.
166 Artificial Intelligence for Business Optimization

Analytics themselves help the business by embedding agility with its busi-
ness processes and organizational structure. The Agile methods also help
in the solutions space in developing the analytics. Agility for the business is
thus supported by developing analytics with an agile approach.
This list of analytics (Table 6.1) helps understand the possibilities of rich-
ness in analytics as well as their application in optimization. For example,
if the purpose of an analytics is to describe a past (historical) situation, then
it’s an informative-descriptive type analytics. Alternatively, if a trend plot is
used to anticipate a change in weather pattern next week, it is a predictive
(futuristic) analytic. Table 6.1 summarizes these analytical categories and
corresponding business optimization strategies.
Each of the analytics summarized in Table 6.1 incrementally builds on
the previous type of analytics. The following is an additional description of
the three important categories of analytics and how they provide value in
business optimization.

­Table 6.1 Data analytics types and corresponding strategies for BO

Strategies for business process
Analytics (Type)
­­ What it implies optimization
Informative Available as easily (publicly) Being static output, has minimal
available information that can be impact on processes. Speed of
broadcast. processing on a large data set is
the minimal cybersecurity
requirement.
Descriptive Describe and Broadcast Provides information to a large
Information/insights
­­ based on group of customers depending
past, historical large data analysis. on their need.
Requires preparation of data that Customer contact points are
may be in different unstructured increased. Business users can
format. May take a long better understand the situations
processing time. in their business processes.
Diagnostic Define analysis as highly Enables identification of pre-
parameterized and focus on a emptive problems. For example,
well-defined function to diagnose equipment failures in a
errors, gaps, and failures. manufacturing unit.
Predictive Analyze the data based on an Provide decision-makers
­­ with
event and provide predictions to trends and patterns of the future.
enable immediate response. Key Time and accuracy sensitive.
Stakeholder focus in order to Depends on the granularity of
help their decisions. business processes. Innovative risk
Helps in risk management as many reduction by using continuous,
risks are not visible. iterative analytics to explore
options; multiple “what-if”
­­­­
scenario generation using data.
Prescriptive Decision-maker
­­ combining with Decentralize decision-making
­­
tacit intelligence. based on recommendations
combined with experience.
 Intelligent business processes 167

Descriptive analytics
Descriptive analytics examines large amounts of historical data to describe
the current scenarios. This is a relatively static data. For example, the
describing process of sales and inventories channels for a salesperson to
understand the areas to focus on. A description of the performance of an
individual salesperson or team is also part of this analysis. Description of
current and past performance can be analyzed to determine the reasons for
success or failure. Data can be classified depending on similar characteris-
tics (e.g., targeted sales campaign). Examples of visualizations in descrip-
tive analytics include scatter graphs and bubble charts.

Predictive analytics
Predictive analytics is meant to identify and determine futuristic trends such
as sales, expected sales, or changes to market behavior. Predictions can
be used in capacity planning and customer retention strategies. Statistical
models used in forecasting are part of predictive analytics. The future of
customer behavior is predicted by examining vast data sets on social media
and mobile data to ascertain customer sentiment. Predictive analytics com-
bine unstructured data (e.g., Web logs, blogs, Facebook, and Twitter feeds),
together with structured transactional data. Predictive analytics are more
fine granular than descriptive analytics.

Prescriptive analytics
Prescriptive analytics goes beyond predications and into the realms of sug-
gested decisions. Decisions using prescriptive analytics are optimized based
on the discovery of trends and patterns within the data. Business rules
are understood and coded in prescribing actions based on these analytics.
Prescriptive analytics make use of past data and models together with the
most current data to create a system that can be immediately applied and
reevaluated across numerous instances. Thus, prescriptive analytics have a
role to play in process automation.

COLLABORATIVE DIGITAL BUSINESS PROCESSES

Digital technologies enable businesses to connect, exchange data and infor-

mation, execute contracts, and increase the overall value to their customers.
This connectivity is usually based on “services.” The tools and technologies
of electronic collaboration support globally dispersed workforces within
the same organization. Collaboration focuses on integrating and extending
corporate operations and services beyond a single enterprise.
168 Artificial Intelligence for Business Optimization

BO is an ongoing and, perhaps, an unending activity. Collaboration of

business processes is a major opportunity for BO. BO is “agile” because it is
continuously changing. Business agility is a customer-centric effort that ben-
efits by collaborating with other enterprises in their space to enhance the user
experience.
The EA’s mechanism to measure, compare, and contrast technology
investments provides the technical foundation for the organization. Such
comparison also provides strategic input into the comparison of processes,
justification of BO, and opportunities for collaboration.
Collaborations require frameworks, patterns, and government and indus-
try regulations as well as best practices, corporate policies, and previously
adopted standards. These strategic choices make collaborative enterprises
agile and flexible enough to adapt to and manage changes in the corporate
IT environment. Ease of integration and the portability of key elements in
existing disparate platforms of the partnering organizations are important.
Teamwork as a part of collaboration is necessary to successfully integrate
enterprises. Hence, evolving business requirements and corporate strategies
together make a strong business case for developing collaborative enterprise
architectures.

Collaboration advantage in a digital world

The impetus for deploying a collaborative enterprise is manifold.
First, it reestablishes a basis on which to do business with partners and
customers in a distributed world, enforcing a set of ground rules such as
­­service-level agreements (SLAs)
­­ or ­­trading-partner agreements (TPAs).
­­
Second, it gives the enterprise architecture and infrastructure the flexibil-
ity to handle growing business needs and to connect globally with prospec-
tive partners, customers, or vendors.
Third, the collaborative enterprise also lays the foundation for establish-
ing communities of practice and promotes reuse of the available knowledge
base without reinventing the wheel.
Business collaborations participate in a mutually beneficial venture.
Connecting, communicating, coordinating, and making commitments between
two or more businesses widen the offerings range. Depending on established
roles and responsibilities in an organization agreements such as SLAs, TPAs,
subcontracting, or sourcing agreements are electronically implemented.
Collaborations share appropriate information, technology, and techniques
using a common electronic medium. Decision-making data and transaction
processing information are exchanged in e-collaborations.
A collaborative environment establishes this operational conduit to
enable optimization of product and service offerings. A collaborative envi-
ronment provides common services, a set of information exchanges, and a
delivery mechanism. Knowledge management (KM), content management,
 Intelligent business processes 169

user experience and relationship management, and security are all intercon-
nected through interfaces. Data analytics offer greater changes of accurate
and timely insights if they are conducted over large and wide-ranging data
sets. This is the biggest advantage offered by collaborators in BO.

Collaborative digital business

A collaborative business enterprise involves people interacting with one
another using processes, policies, best practices, and a collaborative environ-
ment. Technologies and techniques across the enterprise facilitate collabora-
tion. The human component of the collaborative enterprise emphasizes the
complexity of interactions due to the cultural diversity across the organiza-
tion. Collaborative processes include ­­human-to-system,
­­ ­­system-to-system,
­­
or system-to-human forms. Collaborative enterprise systems include inte-
gration of business applications, business processes, frameworks, and ref-
erence models inside or beyond the boundaries of a specific enterprise to
provide appropriate access to the participants.
Digital business uses electronic medium to collaborate electronically with
customers, partners, staff, and regulators. Therefore, every digital busi-
ness is essentially a collaborative digital business. Collaborative technolo-
gies enable a digital business enterprise to explore many opportunities to
provide additional customer value. Collaborative digital business enterprise
becomes a complex and challenging entity because its boundaries are fuzzy
and they are continuously changing. “Collaborative EA” enables handling
the complexity and dynamicity of collaborative digital business.32 In addi-
tion to ­­business-to-business
­­ collaboration, even ­­knowledge-customers and
knowledge-workers associate electronically. A collaborative enterprise archi-
tecture (EA) also offers major benefits in enabling safe collaboration among
customers and workers. Collaborations create and synergize intelligence
within and across multiple organizations to produce actionable insights for
users or end customers.
When business processes are executed across multiple organizations
based on electrically defined policies and contracts, they result in automatic
collaboration.

Complexities of collaborative digital business

The complexity of collaborative processes arises not only due to multiplici-
ties of applications, but also by the fact that they are serving dynamically
changing business processes. Multiple, collaborating applications that are
widely dispersed over physical servers (or in the cloud), many different
input/output points over fixed and mobile devices, and dynamically chang-
ing data comprising text, audio, video, and graphics – all result in a melee
of services that are daunting to decipher for sensible decision-making.
170 Artificial Intelligence for Business Optimization

Business processes become complex as they traverse the boundaries of

multiple organizations in order to collaborate. Assessment and evaluation
of collaborative business processes or workflows, modification to business
interactions, and understanding the dependencies processes is an essential
part of their modeling. Process management also includes collecting such
metrics as user access logs and user fulfillment surveys or tracking the vol-
ume of collaborative activities.
Ranging from simple instant messaging and chat sessions to real-time
audio- and videoconferencing, from video streaming to e-learning to online
communities of practice – the complexity is enormous.

Optimized collaborations
Data, information, process, and knowledge are shared by multiple organiza-
tions across their electronic boundaries and result in optimized processes.
Analytical insights are made available at the right time and place for the par-
ticipating organizations. Collaborative Intelligence (CI) aims to handle the
challenge of sharing across multiple organizations. Integrated and intelligent
business and operational processes result from optimized collaborations.
Automated and optimized collaboration among a group of businesses is
based on CI. Collaboration leverages information sharing across multiple
organizational boundaries and in a dynamic manner.
These electronic collaborations are enabled through tools and technologies
(typically
­­ Web Services and also, increasingly, Analytics-as-a-Service).
­­ ­­ ­­ 33 CI

enhances collective value reduces overall costs. The intelligence within the
processes is used in decision-making in real life by collaborating organizations.
Table 6.2 documents the key concepts of collaborations in enterprise
systems. These concepts form the basis of the mechanisms for embedding
intelligence within processes. Intelligence within these systems facilitates col-
laboration between their processes. Big Data and ML for decision-making
result in improvement in collaborative enterprise systems.
Intelligence can be garnered through information technologies that gen-
erate new and dynamic knowledge within and across the organization.
Collaborative organizations interact with each other, their customers, and
suppliers in real-time through web services. In addition to the technical
capabilities of software, these collaborations also require strong business
relationship-building skills. These business relationships include people
skills and forming electronic policies that can be used in creating and exe-
cuting electronic collaborations. Building relationships and collaboration
also leads to closer scrutiny of the inner workings of member companies, 34
resulting in a need for a greater level of trust and mutual understanding
between these “clustering” member companies.
­­
 Intelligent business processes 171

­Table 6.2 Key concepts of collaborations in enterprise systems2

Key concepts Description
Collaborations These are essential interactions between businesses (and their
systems) in order to carry out business actions and achieve business
goals which cannot be achieved by a single business.
Enterprise These are software systems supported by corresponding data that
systems enable business organizations to carry out their key functions (e.g.,
sales, marketing, inventory management, accounting, HR). Enterprise
systems are continuously evolving to incorporate the Cloud, data
science, AI, and mobility – together with cybersecurity.
Artificial Concepts and algorithms that are implemented in order to capitalize
Intelligence and on the abilities of software systems to “learn” in both supervised and
Machine unsupervised manners and continue to learn and correct themselves.
Learning
Big Data High volume, rapid velocity, and differing varieties of data that has the
potential to be analyzed using data science techniques in order to
provide insights.
Business Includes processes and procedures within business organizations that
decision-
­­ have the potential for ongoing improvement through the use of
making intelligence based on (typically) Big Data.

among those applications also need to be facilitated within the Collaborative

Enterprise Systems (CES). These information exchanges evolve into
ontology-based collaborations among multiple applications and databases.
Furthermore, an organization needs to collaborate among its people and
processes. Business value is derived by enabling people to make productive use
of applications that go beyond the specific transaction they are engaging in with
the organization. Sharing data and information across organizational bound-
aries can produce imaginative new pieces of knowledge that the organization
can creatively use. Collaborative Intelligence (CI) is the extension and applica-
tion of BI together with collaborative business process engineering,35which is
built on AI. CES, equipped with ML, can optimize organizational resources
by using current cloud computing and SOA capabilities.36

VISUALIZATION AND BUSINESS PROCESSES

Visualization is the presentation aspect of BO. Business decisions rely

on data analytics. The velocity, volume, and speed of analytics have an
impact on the comprehensibility of analytics. Without due consideration
to visualization design, making sense of the hidden patterns is impossible
no matter how good the analytics are. Presenting the results of analytics in
a human-comprehensible way is crucial for the success of BO. How much
information is comprehensible? While the metrics vary depending on the
user, the classic “span of control”37 is a good metric to keep in mind. An
172 Artificial Intelligence for Business Optimization

aesthetically pleasing model will have no more than seven elements. The
end users need much less information than the business decision-makers.
Effective visualization is designed in a way to provide the right amount of
information, at the right time to the right person, keeping in mind, limitations
of the device.
Visualization tools interact with the analytical tools to present the results
of the analytics. Ineffective visuals can also be counterproductive as they can
reduce the efficiency and effectiveness of decision-making. User-centered
designs ensure users are able to learn the interface intuitively and that the
interface grows with the growing expertise of the user. Educating the user
in the use of visuals adds to its quality.
Workflow models become important in designing group visuals. Visuals
are not just analytical results. The movement of a patient in a hospital, for
example, is a powerful visual to help the staff.
Presentation of the data results analytics consider three aspects of qual-
ity: syntax, semantics, and aesthetics. Syntactically quality deals with
factual correctness, whereas semantics deal with the correctness of mean-
ing. Aesthetics implies style and representation of insights from analytics.
Aesthetics also represent the ease of creating and reading an analytical
model. Although the analytic may be accurate (syntactically) and meaning-
ful (semantically), its useability depends on its visualization or the style of
representation.
Designing visualization quality considers the following factors:

• Type and size of devices. Each device has its limitations that will limit
the presentation. Visual design includes as many possible devices with
varying screen sizes and different operating systems and environments.
• Rate of change of underlying data and corresponding speed of analyt-
ics to keep up with the changing data. This will impact the perfor-
mance of the visualization because the moment the data is updated, the
analytics, and therefore the presentation, has to change. Presentation is
thus related to the underlying analytics. Flexibility in visuals is enabled
through parameters and rules.
• The context dependency of the results being presented in which busi-
ness scenarios with the results will be used. Thus, the underlying
theme or purpose of a visual is crucial. The purpose of a business
process has to be reflected by the visual. Thinking and designing the
visual starts with the nature and purpose of the business process.
• Amount of information that needs to be shown to the user. This infor-
mation can change from user to user. Displaying the maximum pos-
sible information on one screen is not always a good idea. Aesthetics
includes showing relevant and limited information to the user.
• Currency or time duration for which the visualization is relevant.
Understanding this currency is critical as it influences the quality of
 Intelligent business processes 173

presentation and also has security implications. If a visual is relevant

only for a few seconds, a large, colored, graphic format (rather than
numbers and texts) is advisable. Other pieces of information that may
be relevant for a longer period of time can be placed in a summarized
form on a dashboard to provide both graphics and numbers and text.
The time to display and the amount of information to display are both
functions of multitiered and multilayered presentations. Workflow
between the presentation layers and their interdependencies are also
considered in a group (team) visual design.
• Use of sensors and audio cues. Customization of visual presentation
is providing users with options to configure. Automation of presen-
tation without configurability option reduces the quality of the user
experience. Since each user has her own preference in terms of how
she wants to visualize the information, that information should be
customizable. Testing of every permutation and combination may not
always be possible because of their large numbers.
• Testing – at least initially – can follow the 80–20 rule; that is, 20% of
the features and preferences of visuals will be used by 80% of the users.
Once those tests are accomplished, the rest of the features can be tested.
• Similar to analytics, visuals also have a level of granularity. Visuals
can represent the three middle layers of the data to decision pyramid.

Device and performance

consideration in visualization
The end user devices play an important role in visualization. The manner
in which visuals are created and displayed depends on the characteristics
of the user devices. Following are some important device considerations in
designing visualization and presentation of AI analytics to the user:

• Size – Miniaturization has played a great role in the progress of device

technology. Miniaturization and consequently portability of devices
need due consideration. The art of visualization in analytics has to
constantly adjust and rearrange itself to reflect the advances in the size
of ­­hand-held devices.
• Security – Security in all aspects of the devices requires due attention.
Making the visualization appealing cannot come at the expense of
security. Security of devices includes security of the software, applica-
tion, visual as well as access to the device and its physical security.
• Privacy – Safeguarding user privacy is of paramount importance in
visualization. Which visuals to show and for how long should they be
shown is a factor of privacy characteristic.
• Functionality – Hardware devices like cell phones and ipads contain
large storage and processing capacities. This gives them the ability to
174 Artificial Intelligence for Business Optimization

process colored graphics, photos, and videos. The graphic processing

power of devices should be fully utilized in visualization.
• R isk – Risk mitigation is another factor in device consideration of
visualization. In addition to security and privacy risks, devices also
present the risks of inappropriate use and are subject to malware
(Chapter
­­­ 9).
• Speed – As apps get heavier and bulky, their performance on handheld
devices can degrade. Slow speed of processing on handheld devices
results in one of the most annoying features - the cursor going round and
round in circles. Users do not have the patience, hence features such as
off-line processing are useful. Apps are also lightweight so that they can
be accessed, and their processing interchanged without waiting times.
• Updates to devices – Visuals need to keep pace with the device ver-
sions. Older versions of apps should retire and be replaced by newer
ones. Users find value in renewal of apps, visualization contents, and
interfaces.

CONSOLIDATION WORKSHOP