Unit 4:

ETL, OLTP, OLAP & KPIs

By:
Nargis Mohapatra
 Extraction

➢ Extraction is the first and foremost aspect in business analytics

➢ As the data come from various heterogeneous source systems, it is very important
to extract the relevant data.

➢ Moreover, not all the available data are meaningful. Hence, extraction of relevant
data, is very important for building appropriate data warehouse.

➢ The extraction process is further segmented into two sub phases:

• Data Discovery Phase

• Data Analysis Phase

 Data Discovery Phase
Identification and Selection
➢ Recognizing and selecting the relevant data sources based on business needs.
➢ This includes structured (databases, spreadsheets) and unstructured data (emails, social media).
➢ Ensuring that selected data aligns with analytical goals.
Vigilance (Data Quality & Governance)
➢ Assessing the accuracy, consistency, and completeness of data.
➢ Identifying any anomalies, inconsistencies, or missing values.
➢ Applying data governance policies to ensure compliance and security.
Determining the Origin (Data Lineage & Provenance)
➢ Tracing data back to its original source to verify authenticity and reliability.
➢ Understanding how data has been transformed or manipulated along the way.
➢ Ensuring data integrity through proper documentation and metadata tracking.
Representation (Data Structure & Format)
➢ Understanding how data is stored and represented (tables, JSON, XML, flat files, etc.).
➢ Converting or transforming data into a usable format for analysis.
➢ Ensuring consistency in data representation across systems.
 Data Analysis Phase
➢ After the discovery phase, the next step is to analyse the content of the data.

➢ In the data analysis phase, the main emphasis is on maintaining the right content of the data. If the
data are of poor quality, then it would result in inappropriate design of data warehouse.

➢ The process of detecting and removing or correcting (depending on the quality of data) the
corrupt files in the data warehouse is known as data cleansing or data scrubbing.

➢ Data coming from different source systems might be of poor quality, i.e., there can be various
mistakes in the data. These mistakes can occur due to several reasons, most common being errors
committed by data entry operators while entering record in the source system, or wrong type and
format of data, such as inappropriate date formats, or missing fields in data such as names left
blank, etc., can result in corrupt data.

➢ It is very important to cleanse or scrub the data so that clean data can be extracted for building the
required data warehouse.
 Some Common Types of Data Corruption
➢ As the organisations are working in different countries, it leads to several regional and
cultural variations such as sometimes the name of a person in one country can be very
difficult to understand in another country. This can result in the wrong entry of names
and even addresses.

➢ The wrong entry of address also results in a mismatch between address code and area
code or pin code.

➢ Sometimes, a non-permissible data value is entered into a particular field. For example,
in the field for age, the data entry operator enters telephone number.

➢ There may be instances where the same data are duplicated and present at various
places in the same file. For example, there can be three records for the same person
Mark Andrew, one with the name Mark Andrew, another with only Mark in the name
field and a third one with the name Andrew Mark.
➢ The most common error occurs in the date field as date data can be
entered in a number of different ways or formats, for example, 26-Dec-
2008, 26/12/08, 12/26/08, Dec-12-2008 and so on.

➢ There may be the problem of inconsistency in the data as sometimes the

same data are stored at different levels of detail such as address. At one
place, the address can be Punjab, in another place it can be Punjabi
University, Patiala, Punjab.
Data Cleansing Stages

Parsing

Correcting

Standardising

Reformatting

Merging

Final Correction
Parsing: In the data cleansing process, first the confusing fields or error-prone fields are
parsed or broken to identify each component in detail, for example, name field is broken into
first name, middle name and last name.
Correcting: Applying algorithms, the data cleansing phase corrects the misspelled fields
such as name and address.
Standardising: Following the organisational rules, the data are cleansed so that data can be
stored in a standardised format, for example, to store the designation of a senior manager a
standardised format is used as SR. Mg.
Reformatting: In this part, the data are stored in proper data types to avoid any future
mistakes.
Merging: In the merging, first separately extracted parts are matched and merged into one
table so as to avoid the duplicity.
Final Correction: Once the data cleansing is complete, the data records which were in error
are logged separately in some file and not passed to output for final storage. These records
lead to logging errors. So final correction deals with deleting or correcting logging errors.
Data Quality Aspects

Accuracy

Explicit

Reliable

Compressive

Consistency
Accuracy: It is very important that the data should provide a correct description of the entity. For
example, if the name of the employee is Harry and designation is Assistant Manager, then the stored
data should also show exactly the same record in the same format with proper data type.
Explicit: The values in the data should convey only one meaning and that too without creating any
ambiguity. If there are ten employees working with the name Akash, but only one manager with the
name Akash Sharma, then the stored data should provide the exact information about the employee
and his designation.
Reliable: The values in the data should be consistent throughout the entire data warehouse. For
example, if for Punjab, PJB abbreviation is used, then at all places, the same abbreviations should be
used.
Compressive: The data should be compressive at both individual and aggregate levels. At the
individual level, there should be proper values for each instance, i.e., there should not be any null
values and at aggregate level the data should provide the total number of records without losing any
individual records.
Consistency: The data should be synchronised throughout the enterprise. Consistency implies a single
version of truth, that is the same data should be available to all users, irrespective of their designation,
location, department, etc.
 Types of Extraction
Full Extraction: Complete data are extracted from the source system. It does not take into account the
subsequent changes that occur in the source system. Its application is also very limited because most
times it is very important to track the changes in the source systems and to update those changes.
Incremental Extraction: In this process, current tables are extracted from the source systems and then
they are compared with the previous tables to identify the changes. But this process is lengthy. It is also
very costly as specialized operations are applied to track the changes. This process only captures two
values, i.e., start value in the previous table and the end value in the current table but do not cover the
intermediary values.
Oracle Change Data Capture (CDC): CDC overcomes the limitations of incremental approach. CDC
uses synchronous and asynchronous CDC. The synchronous CDC tracks changes when specific
operations are used on the data. These operations signify that some changes has occurred in the data.
In asynchronous CDC, the changes are tracked from database redo files. Redo files are those files
which have a record of all the changes made in the database.
Offline Extraction: Data are not directly extracted from the source systems, i.e., the data are extracted
from outside source systems such as redo files, flat files and dump files.
 Transformation
➢ It is very important to convert the data into one generalized format.
➢ The process by which the format of the source data is converted into EDW (enterprise
data warehouse) format is known as data transformation.
➢ Data transformation specifies rules and regulations for further loading of the data.
➢ The data after scrubbing enters the data transformation phase and then it goes to loading
phase.
➢ Data transformation is a very critical phase as for proper loading of data it is very
important that data should be present in a desired format. For Ex: Organisation wants to
create a data warehouse and the data are coming from Marketing, Finance and
Production Departments.
➢ To integrate the data, it is very important to change the format of data, as the data are
coming from various heterogeneous sources.
 Levels of Transformation
Transformation is performed at two levels:
Record Level: The records, i.e., field and tables are transformed into the desired format. The
data transformation uses various functions to accomplish this task:
Select: If the source system has three columns, i.e., Name, age and Address, the organisation
only needs two columns for EDW. Then by applying the SELECT function one can choose
desired columns.
Joining: Joining helps to combine different fields from different source tables into one table.
Joining helps to consolidate the scattered data.
Summarisation: This helps to convert the detailed information into summarised information.
For Ex: daily transactions can be summarised as total amount received and dispatched.
Sorting: It helps to arrange the data in a specific sequence, for example, salary can be sorted
as in the highest to lowest order.
Pivoting: Pivoting helps to transform rows into columns and vice versa. This facility is very
helpful in the proper analysis of data.
 Field Level
Field level transformation helps to transform format of field according to the desired
format. There are two types of the transformations:

Single Field
Here, the same field which is
recorded differently from
various source systems is
transformed according to EDW.
For Ex: same field
measurement is recorded in
different formats as cm, m, yd,
Sq. With the help of single field
transformation, the field is
transformed into one format.
 Multi - Field
In this type, transformation takes
place on more than one field, for
example, two fields are transformed
to create one field, i.e., Stu ID in
the target field. This type of
transformation is known as many to
one transformation, i.e., two source
fields are transformed in one field.

Similarly, there can be one to many

transformation, i.e., one source
field is transformed into multi-
fields, for example, address field in
the source is transformed into three
fields in the target.
 Loading
The loading is the last phase in the ETL process. After extraction and
transformation, finally the data are loaded into the data warehouse.

Modes of loading Types of Loading

Load Mode Initial Load

Append Mode
Incremental Load
Destructive Mode
Full Refresh Load
Constructive Mode
 Modes of Loading
➢ Load Mode: Here, there can be two situations:
If the target source is empty and data exist in the incoming tables, then the target source is populated with
the incoming tables.
If the target source is full and data also exist in the source tables, then in the load mode existing data are
erased and new data take the place of old data.
➢ Append Mode: Append mode is a slight modification of load mode. In the append mode, if the target tables
are full and data exist in the source tables, then in append mode the incoming data are added to the target
tables but at the same time the previous data are not eased. It depends on the situation how to handle the
source data. If the source data are copies of already stored data, then the source tables can be marked as
duplicate or even it can be rejected.
➢ Destructive Merge: Here, primary key is matched with the key of existing records and there can be two
situations.
If the primary key matches, then in the destructive merge, the record is updated.
If the primary key does not match, then the new record is added in the target table.
➢ Constructive Merge: Constructive merge is also based on matching of primary key, but with slight
modification of the destructive merge. If the primary key matches the key of the target table, then existing
record is not updated rather new record is added and it is marked as new record which is added against old
record.
 Types of Loading

To accomplish the above modes of loading process, there are three types of loading:

• Initial Load: In this application, the target table is empty and the new data warehouse is to
be created. For this purpose, load mode is applied. Even in load mode, case (a) is
applicable.

• Incremental Load: As the organizations are working in a highly dynamic environment, it

is very important to constantly upgrade the existing data warehouse. In the incremental
load, changes occurring in the source system are updated in the existing target system.
This updating can be done by applying either destructive or constructive merge.

• Full Refresh Load: In this type of application, all the previous data are erased and new
data are added. For this purpose, load mode case (b) is applicable.
 OLTP (Online Transaction Processing System)
• OLTP is a class of applications which provides transaction- oriented applications, i.e., data entry, update
facility and quick retrieval of information.
• OLTP applies client- server processing and is used for routine transactions. It is generally applied by lower
level of operational- control level staff. Ex: in banks, the clerks use OLTP applications for quick update of
information and quick retrieval of information.
• It is applied for those routine transactions, which requires an instant update, for ex: in case of ATM machines
when you with withdraw money instantly, your account balanced is updated and you get the update
information.
• The OLTP applications are designed where,
a. Same database is being referred to fetch information by many users.
b. When the users want online input and output of information.
c. The users are not always technical persons.
d. When the fast retrieval of information is required.
e. When fast updating of information is required.
Case Study
PayPal’s business growth is exponential per year due to the addition of new users, transactions, features and new
channels such as mobiles, etc. Database workloads grew 50- 100% every year. To meet the demands of the
growing business, it implemented fast paced OLTP environment which supports 500+ database instances, 10- 130
TB of data, 5000- 14000 concurrent processes and 10 GB redo/minute.
OLTP applications can be segmented as:

Transaction Processing

Batch Processing

Interactive Processing
 Transaction Processing

This application processes a single item or

information as required. In this application,
multiple users can fetch the information
simultaneously. The user log is very user-
friendly. The response rate in this application
is also very fast.

For example, in a website of the particular airline company, the users do not need to write any technical
program, rather this application "How many seats are available on Monday from Delhi to Mumbai"? and
can quickly get the information. Moreover, multiple users may want to know same vacancy, and this
application provides answers to all the multiple users. To make the things more clear, the diagram enlists
the OLTP model of the ATM machine. Multiple users can operate different ATM machines of the same bank
at the same time. The user can get the mini statement; withdraw money, deposit money, etc. It is important
to note that this application processes single transaction at a time, i.e., one application at a time.
 Batch Processing
As opposed to the transaction processing, batch
processing is applied where the information (or
items) is required in batch. Whereas in the
transaction processing the requirement is the
processing of single information, sometimes
there are situations where the information is
required in batches. For those situations, batch
processing systems are applied. In the batch
processing system, the records are collected in a
batch and the most important aspect is that this
batch is made by presorting of the items.
This presorting is done by applying some specific rule; for example, payroll generation is a type of batch
processing system where the information is required in batch (say, after 30 days or monthly) and the
presorting is done, ie, the information is sorted date wise starting from the first of every month or any other date
depending on the policy of the organisation. One of the main differences between transaction processing and
batch processing is that while transaction processing is online based, the batch processing has generally offline
access to information because batch processing systems are applied in those situations where there is no urgency
of getting information regarding each and every transaction.
 Interactive Processing
➢ Interactive processing is just like transaction processing, but there is a thin line which divides the
interactive processing and transaction processing.
➢ In the transaction processing, the user interacts with the system for a short period of time, but in
case of interactive processing, there are multiple interactions between the user and the system.
➢ In the interactive processing, the system takes considerable time to update the information; as a
result of which for that time the user cannot access the system.
➢ In the technical terms, this is known as lock period, i.e.. the system gets locked and there is no
interaction between the user and the system.
➢ As a result of this, the interactive processing is not suitable where the users want instant access to
the information.
➢ The interactive processing is generally applied in case of project development where limited
number of users are working and there is no urgency in getting the information.
➢ Another area where the interactive processing is applied is the editing and decision
support systems.
 Characteristics of OLTP
Online Connectivity: One of the most important characteristics of OLTP is that users can
access the system online. This connectivity can be done using LAN or WAN depending
on the situation.
Availability: In the present situation, organisations cannot afford to neglect the customers.
The customer can desire to access the information at any time of the day. for example, a
customer might want to see his bank record using the Internet banking facility at 11 p.m.
Even employees may need to access or update the information at any time of the day;
hence, OTP must be available 24 hours a day. For this, the organisations need a robust
OLTP applications which are available 24 hours a day.
Response Rate: Yet, another important feature of OLTP has been that in OLTP there is a
short interaction between the user and the system, which means that OLTP must have a
rapid response rate. OLTP should provide proper load balancing by giving priority to the
selected transaction over other transactions.
Cost: OLTP is being extensively applied by organisations. It will be not wrong to say
that OLTP is a necessity for organisations in the present time. Hence, it is very
important that the cost of the processing of transactions should be low as in any
organisation there are thousands of transactions which are processed every day. This
cost of processing will add to the running expenses of an organisation, and hence, it is
very important that OLTP transaction cost should be less.

Update Facility: OLTP should provide the facility of instant update of the information.
As there are many users who access same Database, there should not be any delays or
lock periods. For this purpose, it is very important that OLTP must use the full
potential of the hardware and software so that an instant update facility
can be provided.
 OLAP (Online Analytical Processing)

➢ OLAP is that application which focuses on analysis of data rather than on transactions
such as OLTP.

➢ OLAP is a specific technology, which enables the managers to view the data from
different dimensions.

➢ OLAP (Online Analytical Processing) is a powerful technology used in business

analytics to analyze large volumes of multidimensional data quickly and efficiently.

➢ It helps businesses make data-driven decisions by enabling fast retrieval, aggregation,

and comparison of data across various dimensions.
 Benefits of OLAP
Multidimensional Analysis: OLAP organizes data into cubes, allowing users to analyze
information across multiple dimensions (e.g., time, geography, product). The dimension
modeling provides the facility of representing the elements in a hierarchical manner. For
example, a year can be broken down into a lower hierarchical level of quarters. So, this
hierarchical representation helps managers to easily view and analyse the complex data
for each year by breaking the data and analysis into quarters
Centralized Approach: OLAP consolidates data from multiple sources into a
centralized repository, ensuring consistency and accuracy. It improves security, access
control, and data integrity by maintaining all business data in one place. Centralized
storage avoids duplication of data and ensures uniformity across different departments.
Fast Data Retrieval: OLAP stores pre-processed and indexed data, allowing quick access to complex
queries. Multidimensional structures enable rapid computations compared to traditional databases. Users can
perform ad-hoc analysis and get real-time insights without waiting for lengthy processing.
Reporting: As OLAP is applied for analysis and it is very important that manager should be able to report
the findings to his seniors OLAP provides the facility of creating reports, i.e., the managers can create reports
using OLAP. The tools required to create reports in OLAP are very easy to understand and operate, Even a
non-technical person applying the tools of OLAP technology can easily create the reports. A simple click of
the button creates reports based on the selected dimension. For example, reports can be generated region-
wise for sales of the product, i.e., in which region a particular product is preferred.
Special Features of OLAP: There are special features in OLAP which help the managers to analyse the data in
an efficient manner. These features are:
Roll-Up: As the data are stored hierarchically applying dimension modelling in OLAP, roll-up feature allows
the analyst to reduce the dimensions. Roll-up feature calculates total for a particular hierarchy of dimension,
i.e, it performs aggregation. Roll-up operation helps to summarise the data, for example, when the data are
stored in the form of location dimension and location is divided as Punjab, Haryana, UP. In case of Punjab,
the data are stored in the form of 10 districts of Punjab. Roll-up facility allows summarising the data from
different districts.
Drill-Down Feature: The drill-down is just opposite to roll-up feature. Drill-down feature helps the analyst
to have the detailed view of the data. For example, for one element Punjab, the drill-down feature helps to
break Punjab into, say, five districts so as to view the sales in each district.
Slice and Dice Feature: Managers need to analyse different dimensions for performing analysis of data. The
slice and dice feature helps the managers to select one dimension (SLICE) or multiple dimensions (DICE)
for in-depth analysis of data.
Pivoting: Pivot operation helps to rotate the data. This rotation of the data helps to analyse the data from
different perspectives. Pivoting can be done by:
• Changing rows into columns
• Changing one or multiple rows into one or multiple columns
• Different dimensions can be selected from different tables to make one table and then these
dimensions can be rotated.
OLAP Architecture
OLAP can be divided as:

ROLAP (Relational OLAP)

MOLAP (Multidimensional OLAP)

HOLAP (Hybrid OLAP)

ROLAP: In ROLAP, the data are stored in the form of relational tables. In this type of
application, data are not loaded in a cube. The ROLAP application is best suited when the
manager needs to have a detailed and in-depth analysis of data. Although ROLAP can handle
large data, but the processing of ROLAP is quite slow, i.e., to get an answer the manager has
to wait as the data are stored in relational tables and not in a cube.
MOLAP: In MOLAP, the data are stored in the form of multidimensional cube. Data are
stored in the form of pre-calculated aggregations which leads to a faster response rate as
compared with the ROLAP. As with the help of pre-calculated aggregation, all possible
combinations are made and then the data are stored in the cube which results in a faster
response rate. But as the whole data are stored in a cube there is always a limitation up to
which limit MOLAP can handle the data. Moreover, the organisation has to spend a
substantial amount of resources to run MOLAP applications.
HOLAP: HOLAP combines the advantages of ROLAP and MOLAP. For example, HOLAP
can handle large amount of data and at the same time it provides faster accessibility and other
key features of cube technology i.e., summary reports.
 KPIs (Key Performance Indicators)
➢ Key Performance Indicators (KPIs) are measurable values that help businesses track progress
toward specific goals.
➢ They provide insight into an organization’s efficiency, performance, and overall success.
➢ KPIs vary based on industry, department, and strategic objectives.
➢ Common categories of KPIs include financial, customer, operational, sales, and marketing metrics.
➢ Effective KPIs should be specific, measurable, achievable, relevant, and time-bound (SMART).
➢ Businesses use KPIs to make data-driven decisions and improve overall performance.
➢ Regular monitoring and analysis of KPIs help organizations identify strengths and areas for
improvement.
➢ Examples of KPIs include revenue growth, customer retention rate, and employee productivity.
➢ Organizations often use dashboards and reports to visualize KPI trends and patterns.
➢ Continuously optimizing KPIs ensures alignment with business goals and long-term success.
 Characteristics of KPIs
Metric-Based: KPIs are always based on numerical data, making them objective and measurable.
They help businesses track progress using concrete figures such as revenue growth, customer
retention rate, or employee productivity. By relying on metrics, organizations can eliminate
guesswork and focus on data-driven decision-making. Without metric-based KPIs, performance
evaluation becomes subjective and inconsistent.

Reflection of Top Management Thinking: KPIs should align with the strategic vision and goals set
by the leadership team. They serve as a bridge between executive objectives and operational
execution, ensuring all departments work toward common targets. For example, if a company aims to
expand its market share, a relevant KPI might be customer acquisition rate. This ensures that
performance measurement directly supports the organization’s long-term success.

Information-Oriented: KPIs should provide actionable insights that help businesses make informed
decisions. Instead of just collecting data, organizations must analyze KPI trends to identify strengths,
weaknesses, and opportunities for improvement. For instance, if the churn rate is increasing,
management can take proactive steps to improve customer engagement and retention. Well-defined
KPIs enable companies to adjust strategies and drive continuous growth.
Simple: A good KPI should be easy to understand and interpret by all stakeholders in the
organization. Overly complex KPIs may lead to confusion and misalignment between teams.
For example, "Customer Satisfaction Score" is a simple KPI that quickly communicates how
customers feel about a product or service. When KPIs are straightforward, employees can
focus on achieving measurable goals without unnecessary complications.
Quantifiable: KPIs must be based on measurable values that can be tracked over time.
Abstract or vague indicators make it difficult to assess actual performance. For example,
"Increase sales revenue by 15% in the next quarter" is a quantifiable KPI, whereas "Improve
business performance" is too vague. Measurable KPIs allow businesses to compare results
over different periods and evaluate success effectively.
Limited in Number: Organizations should focus on a few key KPIs to avoid data overload
and ensure clarity in performance tracking. Too many KPIs can dilute focus and make it
difficult to prioritize strategic actions. For example, a retail business might track revenue
growth, customer retention rate, and inventory turnover as its top three KPIs. Keeping the
number of KPIs limited ensures that employees and managers remain focused on the most
critical success factors.
 Process of Setting KPIs
Define Business Objectives: The first step in setting KPIs is identifying the core business objectives.
Organizations must understand their long-term vision, strategic goals, and specific areas that need improvement.
For example, if a company aims to increase customer satisfaction, a relevant KPI might be "Customer Satisfaction
Score (CSAT)." Aligning KPIs with business objectives ensures they contribute to the company’s overall success.
Communicate with Relevant Stakeholders: Once objectives are defined, it is crucial to communicate them
across the organization. Management, department heads, and employees should clearly understand what is being
measured and why. Transparency in KPI selection fosters alignment and ensures that all teams work toward
common goals. Effective communication prevents misunderstandings and helps employees take ownership of their
performance.
Take Opinions and Feedback: Gathering feedback from employees, managers, and stakeholders helps refine KPI
selection. Since different departments have unique perspectives, their input ensures that the KPIs are realistic,
achievable, and relevant. For instance, the sales team may suggest using "Lead Conversion Rate" instead of a
general "Sales Growth" KPI. Considering multiple viewpoints helps create well-rounded KPIs that accurately
measure performance.
Make the Final Decision: After evaluating feedback, the management finalizes the KPIs that best align with
business objectives. The selected KPIs should be SMART (Specific, Measurable, Achievable, Relevant, and Time-
bound). Once finalized, they should be documented and integrated into performance tracking systems. Regular
reviews and adjustments ensure that the KPIs remain relevant as business needs evolve.
 Balanced Scorecard
➢ A Balanced Scorecard (BSC) is a strategic performance management framework that helps
organizations measure and improve their performance across multiple perspectives.
➢ It goes beyond financial metrics by incorporating four key perspectives: Financial, Customer,
Internal Processes, and Learning & Growth.
➢ The Financial Perspective focuses on profitability, revenue growth, and cost efficiency.
➢ The Customer Perspective measures customer satisfaction, retention, and market share.
➢ The Internal Processes Perspective evaluates operational efficiency, quality control, and innovation.
➢ The Learning & Growth Perspective assesses employee training, organizational culture, and
knowledge management to ensure long-term success.
Customer Perspective: The Customer Perspective measures how well an organization
meets customer needs and expectations. It answers the question: How do customers
perceive us? Key indicators include customer satisfaction score (CSAT), net promoter score
(NPS), customer retention rate, and market share. High customer satisfaction leads to repeat
business, positive referrals, and brand loyalty, which ultimately drive revenue growth.
Companies must continuously adapt to changing customer preferences and improve service
quality to maintain a competitive edge. A strong customer focus ensures long-term success
by fostering deeper relationships and trust with consumers.
Internal Processes Perspective: The Internal Processes Perspective evaluates the
efficiency and effectiveness of a company's operations. It answers the question: How
efficient are our internal processes? Metrics such as process efficiency, cycle time, defect
rates, and innovation rate help businesses identify bottlenecks and areas for improvement.
Streamlining internal processes reduces costs, enhances productivity, and improves product
or service quality. Continuous process improvements, automation, and innovation help
businesses stay agile and responsive to market changes. A well-optimized internal process
system ensures smooth operations and better customer experiences.
Learning & Growth Perspective: The Learning & Growth Perspective focuses on employee
development, innovation, and organizational improvement. It answers the question: How can
we sustain our ability to improve and innovate? Key performance indicators include
employee training hours, workforce engagement, talent retention rate, and knowledge
sharing. Investing in employee development leads to a skilled, motivated, and high-
performing workforce. Organizations that prioritize learning and innovation remain adaptable
in a rapidly changing business environment. A strong learning culture ensures long-term
sustainability by fostering creativity, leadership development, and continuous improvement.
Financial Perspective: The Financial Perspective focuses on an organization's ability to
generate profits, manage costs, and create shareholder value. It answers the question: How do
we look to our shareholders? Key financial metrics include revenue growth, profit margins,
return on investment (ROI), and cost efficiency. A strong financial performance ensures that
the company remains competitive and capable of funding future growth initiatives. However,
financial success alone is not enough; it must be supported by improvements in customer
satisfaction, operational efficiency, and employee development. By tracking financial KPIs,
businesses can make data-driven decisions that enhance profitability and long-term stability.
KPI-based Balanced Scorecard
Example
 Dashboard
➢ A dashboard is a visual representation of key performance indicators (KPIs), metrics, and
data that provide insights into business performance.
➢ It consolidates information from multiple sources into a single interface, allowing users
to monitor trends, track goals, and make data-driven decisions.
➢ Dashboards are often interactive, enabling users to filter, drill down, and customize views
for better analysis.
➢ They can be categorized into operational, strategic, analytical, and tactical dashboards,
depending on their purpose and audience.
➢ Businesses use dashboards in various functions such as finance, marketing, sales, human
resources, and IT to improve efficiency and decision-making.
➢ A well-designed dashboard simplifies complex data, enhances visibility, and helps
organizations respond quickly to changing business conditions.
 Features of Good Dashboard
Understandable: A good dashboard should be simple, clear, and easy to interpret. Users should be
able to grasp the key insights at a glance without requiring extensive training or explanations.
Avoiding unnecessary complexity ensures that decision-makers can quickly extract meaningful
information and take appropriate action. A well-structured layout with logical grouping of data
enhances readability and usability.
Situation-Based: Dashboards should be designed to meet specific business needs and user
requirements. A sales dashboard, for example, should focus on revenue, lead conversion, and customer
acquisition, while an HR dashboard should highlight employee performance, retention, and training
metrics. Customization options allow users to adapt the dashboard to changing business situations,
making it more relevant and useful. The best dashboards provide real-time or near-real-time insights to
help businesses respond proactively to emerging challenges.
KPI-Based: An effective dashboard is built around Key Performance Indicators (KPIs) that align with
business goals. It should display only the most critical metrics to avoid information overload. By
focusing on relevant KPIs, users can track progress, identify trends, and make data-driven decisions.
For example, a marketing dashboard may feature KPIs like customer engagement, website traffic, and
conversion rates to measure campaign effectiveness.
Summarized: A good dashboard presents a high-level summary of data rather than
overwhelming users with excessive details. It should highlight key takeaways, trends, and
performance status in a concise manner. Users should be able to drill down into more
detailed reports if needed, but the main dashboard should provide an at-a-glance
overview. Summarized data helps executives and managers make quick, informed
decisions without getting lost in unnecessary complexity.

Graphical: Dashboards should use visual elements like charts, graphs, gauges, and
heatmaps to enhance data comprehension. Graphical representations make it easier to
identify trends, compare performance, and spot anomalies quickly. Color-coded indicators
(e.g., red for issues, green for success) help users instantly understand performance levels.
Well-designed visualizations reduce cognitive load and make data more actionable.
 Dashboard Design
Hierarchy Based: The dashboards can be designed based on different hierarchies in a
organisation, for example, a dashboard for Chief Manager with the roll-up facility to view
the consolidated view of the performance of all the departments. Similarly for the team
leader, the dashboard displays overall team performance or with the drill-down capabilities
to view the performance of a particular team member.
Functionality Based: The dashboards can also be designed based on functionality like
➢ Marketing based dashboards
➢ Human resource based dashboards
➢ Manufacturing based dashboards
➢ Finance based dashboards
Applying Marketing
Based Dashboards:
The Marketing
Manager can view all
the important
information regarding
marketing like:
➢ New customers
added
➢ New leads
generated
➢ Conversion of
leads
into actual sales
Human Resource Based
Dashboards: The HR
manager can view all the
important information
regarding the critical HR
metrics like:
➢ Headcount
➢ Payroll breakdown
➢ Attrition rate
➢ Training schedule
Finance Based Dashboards:
The Finance Manager can
view all the important
information regarding critical
financial metrics like:
➢ Revenue
➢ Earnings per share
➢ Dividends
Manufacturing-Based Dashboards:
The Production Manager can view all
the important information regarding
critical production metrics like:
➢ Number of orders
➢ Orders manufactured location-
wise
➢ Orders manufactured by units in
a particular location
➢ Number of defects in product at
a particular location
➢ Total number of orders shipped
from different locations