Professional Documents
Culture Documents
Definitions
“A complete end-to-end set of activities that provide value, through the delivery of a
product or service, to the customer of the service”. Sharp and McDermott
“How an organisation does its work – the set of activities it pursues to accomplish a
particular objective for a particular customer.” Thomas Davenport
“Process: an organised group of related activities that together create a result of
value to the customer.” Michael Hammer
“A collection of inter-related events, activities and decision points that involve a
number of actors and objects, and that collectively lead to an outcome that is of
value to at least one customer.” Dumas et al.
Business process is a way to split all these tasks into units so when there is a
problem, you will know where to look
In an organisation you will have several teams (finance, HR, technology etc.), but
none of these produces what the customer wants. Outbound logistics is the people
who deliver the product to the customer.
Example 1 (Shipment)
PO Received Check and Confirm PO Package Products Load Truck Notify
Shipment Issue Invoice Match Payment Payment Made
PO Issued Obtain PO Confirmation Schedule Delivery Unload Truck Issue
Delivery Receipt Check Invoice Schedule Payment Goods Arrived
Example 2 (Restaurant)
Customer Arrived Greet and Seat Take Order Bring Menu Serve Meal
Present Bill Issue Invoice Customer Paid
Kitchen is Dirty Load Dishwasher Clean Kitchen Surfaces Brush Grills
Collect Laundry Sweep and Mop Unload Dishwasher Kitchen is Clean
Typical Processes in Organisations
Order-to-cash
o Performed by a vendor – starts when a customer submits an order and ends
when order has been delivered to customer and customer has paid.
Quote-to-order
o (Quote-to-order + order-to-cash = quote-to-cash)
o Typically precedes order-to-cash
o Starts when supplier receives a Request for Quote (RFQ) from customer and
ends when customer places an order based on quote.
Procure-to-pay (aka Purchase-to-Pay)
o Starts when organisation individual determines something needs to be
purchased, ends when order/service has been delivered and paid for
Application-to-approval
o Starts when someone applies for benefit or privilege and ends when these
are granted or denied
Issue-to-resolution
o Starts when customer raises issue and ends when customer and/or supplier
has resolved the issue.
Value Proposition
Example: “My washing machine doesn’t work”
o Negative outcomes (value- reducing):
Fault not repaired in a timely manner
Fault repaired but customer pays more than expected
o Positive outcomes (value-adding):
Fault repaired immediately with minor intervention
Fault repaired, covered by warranty
Elements of a Process
Event: things that happen automatically (no duration). Events trigger processes.
Activity: a set of tasks that are initiated (triggered) by an event. Simple activities are
tasks.
Decision points: a decision that affects the way the process is executed.
Actors: human actors; organisations; IT systems.
Physical objects: equipment; materials; products; locations; hard copies.
Immaterial objects: electronic documents and records; emails.
Outcome: the execution of a leads to one or several outcomes. Outcomes should
deliver value to actors (customers) involved in the process.
Customer: the actor who consumes the output of the process.
Example of Elements
Event: Appliance owner makes a service call
Activity: Schedule technician visit
Decision points: Is item under warranty?
Actors: Customer service representative
Physical objects: Spare parts
Immaterial objects: Technician availability calendar
Outcome: Appliance fixed under warranty
Customer: Appliance owner
Example 2
Process Performance
If you had to choose between two services, you would typically choose:
o Faster
o Cheaper
o Better
Three dimensions of process performance
o Time
o Cost
o Quality
Improving process performance:
o Customer Arrived Greet and Seat Take Order Bring Menu Serve
Meal Present Bill Issue Invoice Customer Paid
o Can improve by:
Outsourcing to customer
Standardise (daily specials)
Eliminate Cooking
Automate
Eliminate Waiters
Re-sequence
Invest and Build
Why BPM?
If we merely apply technology solutions to poor processes, we may not get better
outcomes for our customers.
“The first rule of any technology used in a business is that automation applied to an
efficient operation will magnify the efficiency. The second is that automation applied
to an inefficient operation will magnify the inefficiency.” – Bill Gates
Every organisation follows a flow of business activities, all of which are directed toward
the completion of a business transaction. The effectiveness of these activities has a
direct and significant impact on the organisation’s outcome.
BPM Lifecycle is the set of activities that constitutes the foundation of business
process management methodology. It lies at the heart of the business process
management discipline and provides us with a strategy for systematically and
continually improving company processes.
This section provides an overview of the concepts, methods, techniques and tools
that compose the BPM discipline through the lens of BPM lifecycle. This lens
provides a structured view of how a given process can be managed.
Process Identification
o Describes the overall process organisation in terms of a process architecture
Process Discovery
o Comes up with the as-is process model to describe the current process,
where you identify weakness and their impact on the performance of the
process.
o This will help you redesign your process to the to-be process model
o You implement the to-be process model into an information system, provide
the IS infrastructure for running the process, and educate the people invoiced
in how it is meant to operate.
o Afterwards, individual cases will be executed so they can be monitored,
providing insight into the conformance and performance of the process
Process Identification
In each organisation, there are enablers which support the business processes (core
processes), which support the strategy of the organisation.
In this course, we focus on workflow design and information systems enablers
Process Improvement in Context: Value Chain Model
Michael Porter introduces the concept of the value chain. This is a collection of all
activity’s performance to design, produce, market, deliver and support any product
line.
The middle row represents the primary activities of the organisation.
E.g., when manufacturing a product, the raw materials are received by inbound
logistics, made available to operations, and then converted to a product. This
product is in strategy with the organisation.
o When manufactured, this product is sent to outbound logistics who distribute
through their channels marketing and sales who promote the product and
service providers to provide support to customer.
o Customer is ready to pay a margin to receive this product
Within the process identification phase, the Business Process Analyst (BPA) will ask the
following questions:
Which processes are we dealing with?
How can we measure the outcome?
A BPA will need to define the start and end of a process to eliminate confusion (e.g., the as-is
process).
If the company has engaged in BPM initiatives in the past, it’s possible that parts of
the process will already be defined and documented. However, if the company has
not engaged in BPM initiatives, it needs to identify the processes that are relevant to
the problem/issue and define which relationships they have.
Measuring value is a crucial step in BPM, because without measuring you cannot see
the improvement that the process has made.
A process can be measured as:
o Cost related - Decreasing costs while keeping the same/better quality.
o Time related - Improving the cycle time (time to do a process), while keeping
the same/better quality.
o Quality related - Reducing error rates (times processes end up in a negative
outcome)
In each organisation, there is a range of stakeholders involved in the process,
however, only a handful of those stakeholders have a full understanding and
overview of the business processes in their organisations.
o Those few stakeholders have an important role in identifying the processes
with high priority in order to closely manage and improve them.
BPM is costly, i.e., not all organisations are able to rigorously analyse, redesign and
control all their processes.
o Some business processes need to receive higher priorities because they
support the strategy of the organisation and help achieve its business goals.
1. Context of Process Identification
case shows a generic manufacturing company where there are strategic and
management processes at the top, support processes in the bottom around
personnel, management, and assets, and core processes in the middle
Example: UNSW
Question: What are core, support, and management processes of UNSW?
o 2025 Strategic vision: UNSW aspires to be Australia’s global university,
improving and transforming lives through excellence in research, outstanding
education and a commitment to advancing a just society.
Solution:
o Management Processes: Vision, Strategy, Goals
o Core processes (groups of): Research, teaching (and advancing society).
o Support processes: Estate Management, HR, IT, student societies etc.
E.g., you work for an IT company and the manager tells you there is a forecasting
problem.
o You assign them to a data scientist, and you can get some improvement by
trying to build a better model given what is currently wrong.
o However, you can help them more by looking globally to make a better
improvement e.g., synchronising the marketing campaign with the purchasing
process so you can use data from one to the other to maximise the value of
your BPM project.
Process Enumeration
“Most businesses have just three core processes:
1. Sell stuff
2. Deliver stuff
3. Making sure you have stuff to sell and deliver” - Geary Rummler
Insurance Company
University
Process Scope
Processes are interdependent → insights into interrelations required
o Horizontal: upstream – downstream processes and their value chains
(sequence)
o Vertical: main processes – sub-processes (Decomposition)
o Specialization: general – special product/ service
Relationships between Processes
3. Process Selection
Designation via Reference Models
A reference model is used as a template to design the process architecture
Examples:
o Information Technology Infrastructure Library (ITIL)
o Supply Chain Operations Reference Model (SCOR)
o Process Classification Framework (PCF)
o Control Objectives for Information Technology (COBIT)
o Value Reference Model (VRM)
o Voluntary Interindustry Commerce Solutions (VICS)
o eTOM Business Process Framework
Recap
Business processes are important because:
o All organisations manage processes
o They execute an organisation’s mission, strategy, goals and objectives
The Value Chain Model (Porter 1985) enables us to view process improvement in
terms of:
o Core vs. Non-core processes
o Sources of competitive advantage (low cost vs. differentiation)
Business Process Management (BPM) focuses on continuous improvement of cost,
time, accuracy, value, customer experience and/or efficiency
Process identification
o Process architecture definition aims at enumerating major processes of
organization.
o Process architecture defines relationship between processes.
o Seven-step method for definition of process architecture including process
landscape model.
o Process selection is concerned with prioritizing processes.
o Priorities upon importance of processes, health, and feasibility of
improvements.
o Assessed by process owners or grounded on process performance measures
and objectives.
o Most common performance dimensions are time, cost, quality, and
flexibility.
o Process portfolios help in selection of processes for improvement.
o Selected processes become subject of remaining phases of BPM lifecycle.
Week 2 – Introduction to Process Discovery
Process modelling is
an essential part of
the process discovery
phase in the BPM
lifecycle.
In this course, we will
be using Business
Process Modelling
Notation version 2.0
(BPMN 2.0) to model
processes.
BPMN 2.0 is an
international standard
that is governed by
the Object Modelling
Group (OMG). Most BPM modelling tools adhere to the BPMN 2.0 standard.
In this course, we will be using Signavio as the BPMN 2.0 tool.
Activity: captures something that is done (we may refer to this as “work” during the
course) in the organisation.
Event: captures when something happens, like the start and end events discussed
above. Represents the process’ triggers
Gateway: captures when the flow goes in different directions, either to split (or
duplicate) the process flow and then to join it back up again. Gateways usually
represent decisions or conditions when a process might work differently in different
situations.
Sequence flow: is the path that the process travels, showing the order of execution,
where the process goes to next after each step.
Events
The start event kicks off a single instance of the process – for example, an order
arrives via email.
o An orders’ clerk may arrive at work in the morning to find that 10 orders that
have come in overnight.
o This means the start of 10 instances of the process.
It is useful to think of each instance as a token that travels along the path until it gets
to the end (i.e. until the token is consumed or when it dies)
The end event shows a token being “consumed” or it can be visualised as a “token
sink.” This signals the completion of a process.
Every process has at least one start and one end event.
Gateways
XOR Gateways
The most common type of gateway is a decision gateway – if the stock is available,
we can accept the order. However, if the stock is not available, we will reject the
order.
This is known as an eXclusive OR (or XOR) – A token can take only ONE path of the
XOR gateway.
XOR gateway can “split” or “join” the sequence flow (we may refer to this as a path).
o When we use an XOR-split in a model, we label the sequence flows
coming out, to show which instance would follow which path,
remembering there can be only one path used for each instance (for
an XOR):
In this example, there are two possible paths, but there can be
more in case there are more than 2 conditions.
For example, an order could be accepted (condition 1), rejected
(condition 2), or placed on hold (condition 3).
o When we use an XOR-join in a model, remember the process is expecting
only ONE token. Once the token passes through the XOR join, it moves on to
the next parts of the process
AND Gateway
AND gateway splits the token and follows two paths simultaneously.
AND split means that the token takes both paths at the same time
AND join can only continue when ALL the paths have been completed and are ready
to re-join
OR Gateway
OR gateway can create 1 or more paths for the token to follow at the same time. So,
the OR split takes 1 or more branches, the OR gateway is a smart gateway because it
knows if one or more branches need to be taken:
Example: Here, OR gateway is required as although we might have split into two
tokens at the end split, it is also possible we lost one after activity E, so we wait for all
active tokens.
Using Gateways for Re-work or Repetition
XOR gateways are frequently used to send a process model back to a previous step
for re-work. In the example below, if the response is not accepted, it needs to be
prepared again (as depicted in the model, we use XOR gateway)
It is often important to show who is doing what activity. To begin with, we want to
be clear about the boundaries of the organisation. This is modelled as a “pool”:
We usually start modelling within the subject organisation, within one pool,
identifying the internal organisational actors as lanes.
o When we create other pools (e.g., for customers and other external parties)
we leave them as a “black box”, that is something that we can’t see into, and
we don’t know exactly what is happening inside.
o In rare instances, we may also model the process as it appears to our
customers. However, we NEVER model a process traversing between two
pools – each pool starts and ends an independent process with a sequence
that is completed with the pool
Example:
Go to the BPMN elements dropdown and select BPMN complete to see more
elements swim lanes collapsed pool. Change start event to “start message
event”
We want to model the customers pool as a black box. We don’t model any activities
or events inside this pool as we don’t have control on what the customer will do.
Note: make sure when dragging message element, there is a green dot or else it will
show up as a warning when saved. Cannot submit assignment tasks with any
warnings – errors are ok.
Syntax
1. The Sequence Flow cannot cross the boundaries of a Pool
2. Both Sequence Flow and Message Flow can cross the boundaries of
Lanes
3. A Message Flow cannot connect two flow elements within the same
pool
Business Objects
Data is also used and altered by the process, and we may need to represent this in
our model. Data can be either physical or electronic.
Data object: represents an artefact used by (input) or created by (output) an activity.
Data store: is a place containing data objects that are persistent (they are
persistent, are stored beyond start and finish boundaries of the process)
Example
BPMN Complete Data Objects start with Data Store then Data Object
o Note: don’t have to add in every data store/object for assignments and exams
because it may become messy.
Example 2
Order-to-cash
The purchase order document serves as an input to the stock availability check
against the Warehouse DB. Based on the outcome of this check, the status of the
document is updated, either to “approved” or “rejected”. If the order is approved, an
invoice and a shipment notice are produced. The order is then archived on the
Orders DB.
Text Annotations
We can add annotations to our model for clarity. These do not affect the flow of
tokens; they are only for extra information.
Example of BPMN
Order-to-cash
o A typical order-to-cash process is triggered by the receipt of a purchase order
from a customer. The purchase order has to be checked against the stock
regarding the availability of the item(s) requested. Depending on stock
availability the purchase order may be confirmed or rejected.
o If the purchase order is confirmed, an invoice is emitted and the goods
requested are shipped. The process completes by archiving the order or if the
order is rejected.
Breaking it down:
o A typical order-to-cash process is triggered by the receipt of a purchase order
from a customer.
o Upon receipt, the purchase order has to be checked against the stock
regarding the availability of the item(s) requested.
o Depending on stock availability the purchase order may be confirmed or
rejected.
o If the purchase order is confirmed, an invoice is emitted and the goods
requested are shipped. The process completes by archiving the order or if the
order is rejected.
Naming conventions:
Event: noun + past-participle verb (e.g. insurance claim lodged)
Activity: imperative verb + noun (e.g. assess credit risk)
Give a name to every event and task
For tasks: verb followed by business object name and possibly
complement
o Issue Driver Licence, Renew Licence via Agency
For message events: object + past participle
o Invoice received, Claim settled
Avoid generic verbs such as Handle, Record...
Label each XOR-split with a condition
o Policy is invalid, Claim is inadmissible
Example 2
Order distribution process
o A company has two warehouses, one in Amsterdam, the other in Hamburg,
that store different products. When an order is received, it is distributed
across these warehouses: if some of the relevant products are maintained in
Amsterdam, a sub-order is sent there; likewise, if some relevant products are
maintained in Hamburg, a sub-order is sent there. Afterwards, the order is
registered and the process completes.
Other Guidelines
Model in blocks
o Pair up each AND-split with an AND-join and each XOR-split with a XOR- join,
whenever possible
o Exception: sometimes a XOR-split leads to two end events – different
outcomes (cf. order management example)
Important Notes
Do not leave too much space in lanes
Make sure the text is readable
Good Model
High level (5
steps max),
which each of
these steps
modelled as a
subprocess
Name
everything
clearly with
gateways and
start/end
event
Resources
http://www.bpmb.de/images/BPMN2_0_Poster_EN.pdf
Recap
Order-to-cash cocktail
Chapter 5 sections 5.3, 5.4 for steps to move from interpreting a business problem to
creating and validating your BPMN models
RAT Test
1. What are the benefits of modelling processes?
a. To increase the understanding of existing processes
b. To help process analysts to analyse the processes
c. To facilitate a shared understanding of how a process works
d. To apply future changes
2. Who might use the supply chain process model for Woolworths?
a. Woolworths supplier
b. Woolworths customer
c. Woolworths manager
d. Woolworths developer
i. In this course, we are just focusing on the internal usage of the
process
3. Of the process names below which one does not meet the guidelines for naming a
process?
a. Accept order
b. Deliver pizza
c. Message customer
d. Enrol student
e. None of the above
4. Which of the statements below describes an action event?
a. A person or an organisation decides to do something; customer decides to
place an order, a company needs to purchase materials from a supplier, a
statutory authority issues a new guideline
5. A process is ____
a. How to understand contribution
b. Sharing in a customer outcome
c. About adding value
d. About passing the parcel
6. In a process-orientated organisation, who is directly accountable for poor
performance of a process?
a. The process participants
b. The executive management team
c. The process owner
d. The process analyst
7. A process is NOT
a. About adding value
b. A low level task definition
c. A functional internal way of doing things
d. What you do to fulfil your job responsibilities
8. Which of the below is NOT a process
a. Placing an online order
b. Paying the bill after receiving the goods
c. Setting a drive through order at KFC
d. Ordering groceries online from the supermarket
i. The rest are all order-to-cash processes but paying the bill after
receiving the goods is a task inside an order-to-cash process.
9. Which of the below gateways should you consider using last?
a. OR Gateway
b. XOR Gateway
c. AND Gateway
i. You should use OR gateway at the end as the last option. You start
using XOR or AND
10. A pool will be used to represent a
a. Department in the organisation
b. Organisation
c. Process
d. Group of people
i. E.g., UNSW as a pool and departments (administration, student
services, admission etc.) as lanes within the pool.
Week 3 – Advanced Process Modelling
BPMN Gateways
Process Decomposition
An activity in a process can be decomposed into a “sub-process”
Example
Subprocesses
increased
understandability in
business process
modelling
In Signavio
Sub-processes
An activity in a process can invoke a separate sub-process
Use this feature to:
o Decompose large models into smaller ones, making them easier to
understand and maintain
SESE = Single Entry Single Exit fragment, i.e. a fragment delimited by a single entry
node and a single exit node (there are no other incoming arcs into the fragment or
outgoing arcs from the fragment)
In this process, if the response is not approved, we have to go back and redo. We can
take what is inside and make this a subprocess.
Example: block-structured repetition
The advantage of using activity loops is that the size of the model can be reduced
and the details of the repetition block are hidden inside the collapsar process
The loop icon means until the response is approved, it will keep repeating the
process
Handling Events
Events Handing
In BPMN, events model something instantaneous happening during the execution of
a process
They affect the process flow:
Start (Start Events)
Intermediate (Within a process)
End (End Events)
BPMN Event Types
Untyped Event – Indicates that an instance of the process is
created (start) or completed (end), without specifying the cause
for creation/completion
Start Message Event – Indicates that an instance of the process is
created when a message is received
Message Events
Events where information is received or sent externally (e.g., to customer,
supplier) are called message events.
Message events are either called “catching” events, where the event
receives in incoming communication (the white envelopes) or “throwing”
events, where the event sends an outgoing communication (the black
envelopes).
o A process can start by receiving a trigger externally; and end with
sending a trigger externally.
All message events will have a message flow outwards to the party we are
communicating with.
o We sometimes use message flows coming directly out of an activity, if one
part of the activity definition is to send a message.
o We only use a message event if the ONLY work to be done is to send the
message.
o This means that our models can look like either of these options:
The second option would be used if there was more work to be done
that just the send (e.g. generating the invoice etc.)
However, an activity will never replace a start message event. If there
is work to be done after the message is received, this will be in the
first activity of the process.
Example: message events
Comparison with sending/receiving tasks
to wait until the next Monday. Or it has to wait for 3pm; or the 3 hours to pass.
Process can be started (triggered) at a particular time, or timer events can occur
within a process, where continuation is triggered at a particular time or day; or after
a specified delay time.
o Start Timer Event – Indicates that an instance of the process is created at
certain date(s)/time(s), e.g. start process at 6pm every Friday
o Intermediate Timer Event – Triggered at certain date(s)/ time(s), or after a
time interval has elapsed since the moment the event is enabled (delay)
Timer events are catching events only
The process does not generate the timer, but reacts to it
Example:
PO handling
o A Purchase Order (PO) handling process starts when a PO is received. The PO
is first registered. If the current date is not a working day, the process waits
until the following working day before proceeding. Otherwise, an availability
check is performed and a PO response is sent back to the customer.
For example, we review a document, and then the XOR gateway can take one path if
it approved, another if not approved.
o But what if we don’t know what will happen next? Perhaps we have send a
message to our customer asking for confirmation. The process will take one
path if the customer confirms; another if the customer cancels.
o We can’t use an XOR gateway after “Request confirmation” because we don’t
know what will happen next. We can use an event driven XOR split (also know
as an event driven gateway; or a racing event, because the first event to arrive
“wins the race” and the process follows that path).
Example
Model the following process.
A restaurant chain submits a purchase order (PO) to replenish its warehouses every
Thursday. The restaurant chain’s procurement system expects to receive either a “PO
Response” or an error message. However, it may also happen that no response is
received at all due to system errors or due to delays in handling the PO on the
supplier’s side. If no response is received by Friday afternoon or if an error message is
received, a purchasing officer at the restaurant chain’s headquarters should be
notified. Otherwise, the PO Response is processed normally.
Example 2:
After a supplier notifies a retailer of the approval of a purchase order, the supplier
can receive:
o An order confirmation
o An order change
o Or an order cancellation from the retailer
It may happen that no response is received at all. If no response is received after 48
hours, or if an order cancellation is received, the supplier will cancel the order.
If an order confirmation is received within 48 hours, the supplier will process the
order normally
If an order change is received within 48 hours, the supplier will update the order and
ask again the retailer for confirmation. The retailer is allowed to change an order at
most 3 times. Afterwards, the supplier will automatically cancel the order.
Order to cash process: Event driven XOR split is used as the other party (retailer) is
making the decision, not the supplier. Whatever the retailer decides to do first, the
supplier will let it execute and proceed with the order.
Immediately after an event based gateway or racing event, you should not use
activities or tasks. Instead, you should use message or timer events as shown above
as you would be waiting for something to happen from the other party.
Terminate end event is used after “cancel order” as it is a negative outcome
The XOR join is used to bring the “seek confirmation” back to the start to keep it not
messy – it can join the event-driven XOR split instead but this will look messy/
overlapping.
Example 3
What is wrong with the
following process?
o The seller process
will never
complete in the
case when they
are already
registered.
Solution:
Handling Exceptions
Process Abortion
Exceptions are events that deviate a process from its “normal” course
The simplest form of exception is to notify that there is an exception (negative
outcome)
This can be done via the Terminate end event: it forces the whole process to abort
(“wipes off” all tokens left behind, if any)
Exception Handling
Handling exceptions often involves stopping a sub-process and performing a special
activity
Types of exceptions for an activity (task/sub-process):
o Internal: something goes wrong inside an activity, whose execution
must thus be interrupted. Handled with the Error event
o External: something goes wrong outside the process, and the execution
of the current activity must be interrupted. Handled with the Message
event (e.g., customer cancelled the order)
o Timeout: an activity takes too long and must be interrupted. Handled
with the Timer event
All these events are catching intermediate events. They stop the enclosing
activity and start an exception handling routine.
The error events are a result of something incorrect or problem result within the
task/process.
o The end version occurs within the sub-process, and
“throws” the exception the intermediate exception which
“catches” the exception and kicks off the exception
handling.
Example: Extending PO Handling Process
PO handling
o A PO handling process starts when a PO is received. The PO is first registered.
If the current date is not a working day, the process waits until the following
working day before proceeding. Otherwise, an availability check is performed
and a PO response is sent back to the customer.
A PO change request may be received anytime after the PO is registered. This
request includes a change in quantity or line items. When such a request is received,
any processing related to the PO must be stopped. The PO change request is then
registered. Thereafter, the process proceeds as it would do after a normal PO is
registered. Further, if the customer sends a PO cancelation request after the PO
registration, the PO processing must be stopped and the cancelation request must be
handled.
Solution:
After we check for availability, we insert XOR gateway and if all the items are
available, then we will continue as normal.
If one item is not available, we will throw a thunder. This will stop the process and we
will catch this error via a boundary error.
o The boundary error has to have the same name for both.
o The above error event is thrown from the subprocess
The customer may send a request for address change after the PO registration. When
such a request is received, it is just registered, without further action.
Evidence-based approaches
Some examples of evidence-based approaches are:
o Document analysis: internal processes and procedure documentation
o Observation: following or participating in individual cases of process
execution (e.g., ordering a product from a company, and following that order
through all the internal steps for fulfilment)
Automated-based discovery
We have also seen in week 3, automated process discovery, which requires the use of
event logs. Event logs (refer to figure below) include data captured from source
systems involved in the process:
1. It is important that the event log contains the Case ID, which is a unique
identifier for each instance of the process being analysed. For example, this
could be an order number, an application number, a request number, etc.
2. An activity descriptor is also needed, to provide information on the step
sequence, as well as a timestamp.
3. Other information might also be captured such as the case’s qualitative or
quantitative data (e.g., order source, order value, or in this case loan type and
loan amount). These data points can help with the calculation of quantitative
analysis. The process participant in each step should also be recorded.
Using event logs, the automated-based discovery helps process
analysts map out potential triggers, steps and exceptions.
For example, in a grocery store, there are multiple different types of
checkout processes that occur. Some of these processes may include
the purchase of one item and others may include errors due to
barcode scanning. All these individual instances are recorded in event
logs. Based on all these different instances, a process model can be
created.
Log Entries
Examples of log entries (elements: label; sequence; time stamp)
o Check Invoice for Invoice No. 4567 completed on 12.11.2010 at 9:19:57
o FunctionStoreCustomerData(„Müller“,c1987,
„BadBentheim“)completedon12.11.2010at9:22:24
o Send Invoice for Invoice No. 4567 completed on 12.11.2010 at 9:23:18
Logs contain valuable information to answer questions like:
o When and how many process instances have been executed?
o Are there recurring patterns in the execution of activities?
o Can process models be derived from the data?
o Which paths of execution are used how often in the process models?
o Are there paths which are never taken?
As a process analyst, what type of approaches might you use to conduct your
investigations?
o Review and existing process model
o Interview customers of the process
o Use data from information system used in the process
o Conduct empirical studies of the process, through observation or document
review
o Interview process participants
You may assume that data is the most important, but often people (interviewees) can
provide you with more insights both on the “why” and where there are expectation
gaps.
Increasingly, as Information Systems participate in a wider variety of processes, we
can use the data from the systems to inform our process analysis.
o The automated tools we use for measuring process analysis may also be used
for process monitoring.
Often the source data that is captured during process execution is not in an easily
translatable and analysable format.
o There are numerous data analysis and presentation tools that can be
employed by business users for collating and reporting.
o To analyse processes, we will be examining a tool called Process Mining.
Process Mining specifically supports process analysis.
o By using Process Mining, in conjunction with a variety of quantitative and
qualitative analysis, we can understand the issues and challenges in our
current process.
Process Measurements
Understanding the key metrics of process performance is critical when targeting
process improvement. The most common dimensions of process improvement will
be recapped in Quantitative Analysis of Week 7:
o Quality
We can measure elements like customer satisfaction, product returns,
customer complaints
o Time
We can measure the time it takes to execute the process as a whole or
a specific process variation.
o Cost
We can measure the cost required for the completion of the process
as a whole or a specific process variation. There can be important
trade-offs with fixed and variable costs, and substantial improvement
investments versus reduced running costs over time.
When we evaluate the as-is process, it is important to consider the baseline metrics
and the potential improvements. The method for measuring the improvements must
also be considered in the business case. This information can be used as the
foundation for ongoing process monitoring.
Celonis Process Mining Lecture
“What method would you use as a manager to understand how a business process
actually runs in your multinational corporation?”
o Interview staff, however this would take a lot of time and won’t give you the
complete, up-to-date information
Process mining is based on real empirical data so it’s very accurate with high
transparency
In the real world you have a bunch of processes inside the company
Most companies have some sort of electronics system of records
o All this generates a massive amount of data
Celonis software connects to all these records, reads the data and reconstructs the
processes
Process Mining vs. Data Mining
Process mining
o Analyses process
o Examines extracts data from event logs
o Discover, monitor, improve real business process
o Insight on process:
performance
optimization
Data Mining
o Analyses large data sets
o Broad application and techniques for data extraction
o Discover previously unknown information
o Insight on various relationships within data in any field
Process Analysis
o Performance
Why are these cases late?
Where are bottlenecks?
Which resources are overloaded?
Important as if your process performance is not optimal, you
are losing money, your employees are unhappy, employees are
doing useless tasks trying to fix inefficiencies in the process
o Conformance
Process Conformance is a technique to analyse the relationship
between log data and process models
Input: Logs and process model
Output: information on the relationship, e.g. fitness
How often are four-eyes principles violated?
Which activities are often skipped?
Which resources cause deviation?
Important as not following processes is unsafe, your company
might get a huge fine for regulations and legal actions
o Enhancement
Where to change activities?
How should the process be redesigned?
Which activities could be automated?
An interdisciplinary science
Research classification: Process mining lies at the intersection between process
science and data science, bridging the gap between model-based process analytics
and data-centred analytics.
Process mining extracts the event log, then goes through discovery, conformance and
enhancement. You can then look at the data and see what’s happening again
Demo
Pre-uploaded data sets on the left
Go to Order-to-Cash and select Order to Cash Demo – EN (USD)
Short Exercise
ANALYSIS OF PROCESS FLOW IN THE VARIANT EXPLORER
o Please open the O2C process and navigate to the “Process” sheet. In this
step, you will learn how to interpret the process graph and how to identify
inefficiencies and bottlenecks. Please note that one or multiple answers
might be correct.
2. What’s the overall average throughput time (in days) for the happy path from
process start to end?
a. 16
3. Display the first 17 variants in the variant explorer. How does the 17th most common
variant look like (use process animation)?
a. The order is cancelled and afterwards, a delivery document is created.
1. Look at sales order items where the goods are returned. Who is the dominant
customer for these positions in terms of number of sales order items?
a. Queen Industries 11044
2. Have a look at customer Fast Section 2560. Observing the number of SO items and
the net value over time, what attracts your attention?
a. The net value of sales orders drops noticeable in March 2018.
Search up Fast Section 2560 and apply the filter.
Introduction
We use the qualitative approach to categorise
how a problem is structured.
As a process analyst, your job is to reduce the defects in the process, to reduce the
cost, to minimize customer complaints, etc…
There are 5 techniques: Value-added analysis, Waste Analysis, Root-cause Analysis,
Pareto Analysis, and Issue Register.
o These techniques allow us to break down the process, pinpointing any
process inefficiencies and weaknesses.
First, we want to cover a set of techniques for analysing the process and determine
what adds value, what does NOT add value and how value is being added.
From an internal perspective, we also want to know if there is any work done in the
process that should not be done, which means waste. We need to find out what we
are doing right and what we are doing wrong in that process.
From an external perspective (e.g. the root-cause analysis), when we know about the
problem from other parties (e.g. customer complaint: very frequently the delivery
arrives to the customer 3 days late and the customer is NOT notified of the delay).
o The root-cause analysis allows us to search for the root cause of the problem.
It allows the process analyst to ask the stakeholder questions that can lead to
the root-cause of the problem.
The issues, both internal and external, that have been identified are recorded in the
“Issue Register”. We can prioritise the issues in this register using a Pareto chart.
Value-Added Analysis
The value-added analysis is a very simple technique that decomposes the process
tasks into steps that are pieces of work and for each piece of work or each step, we
want to know if it adds value to the customer or to the business.
Consider the task of putting an elephant in a refrigerator. If we were to break this
task down into three steps, what would those steps be?
o Open the refrigerator
o Put the elephant in the refrigerator
o And close the refrigerator.
Every task can be decomposed into some steps, e.g., 1) setup, 2) the actual work,
and 3) the completion phase, where the work is transferred to the next person.
o To decompose a task into 3 steps: we ask ourselves:
What do we need to do to start the task?
What do we need to do to the task?
What do we need to do to close the task?
o When we decompose the tasks into steps, we need to consider if there are
any conditional steps.
o Remember the elephant in the refrigerator? Now consider putting another
animal in the fridge in 4 steps.
You open the refrigerator
If there is another animal, you take it out
You put the new animal into the refrigerator
And you close the refrigerator
o When we decompose a task into steps, we consider a conditional step, this
step is the cleanup (e.g. clean the fridge – remove the elephant)
In this Value-Added analysis, we will take the tasks and decompose each task into
steps then label each step. We are working with steps – not tasks. We will label the
steps as: Value-Adding (VA), Business Value-Adding (BVA) or Non-Value-Adding
(NVA).
Value-Adding Activities
A step is Value-Adding (VA) if it produces value or satisfaction to the customer of the
process.
There are criteria for VA steps:
o The step is VA if the customer is clearly willing to pay for it. In a university,
student’s admission process, assessing your application as a student is a VA
step (positive criterion)
o The second criterion: would the customer agree that this step is necessary to
achieve their goals? (positive criterion)
o The third criterion is the negative side, If the step is removed, would the
customer perceive that the end product or service is less valuable? In other
words, if we stop doing this, will the customer scream?
For example: in the order-to-cash process: the customer sends you the
order and you confirm the delivery. This is VA. If you do not confirm,
the customer will contact you to ask if you will be sending the order.
Delivering the products is also important. Let’s take the university
admission process, when you - as a student submit - an application,
the university admission office does lots of steps; they assess your
application, and they notify you of your application outcome. If they
do not notify you, you will be chasing them, so those steps are VA. For
you as a customer, it appears that the steps are only those ones.
Another example: you order something from eBay with a few clicks
and you receive them in a couple of days.
when the order is placed to when payment is received. By scrutinizing every step of
the process and eliminating Non-Value-Adding wastes, such as overproduction,
waiting, and unnecessary inventory, the timeline could be reduced.
Motion
Motion of resources internally within the process
Common in manufacturing processes, less common in business processes
Examples
o Vehicle inspection process, a process worker moves with the inspection forms
from one inspection base to another; in some cases inspection equipment
also needs to be moved around
o Application-to-approval process: a process worker moves around the
organization to collect signatures
E.g., in a travel requisition form, a worker has to move from, one desk
to another to obtain the acquired approvals. Approval for expenses,
approval, travel plan etc.
Hold
Inventory
Materials inventory (more inventory than what is strictly necessary)
Work-in-process (WIP) – Inventory waste shows up in the form is WIP. WIP is the
number of cases that have started and have not yet completed.
Examples
o University admission process: About 3000 applications are handled
concurrently
o Vehicle inspection process: when a vehicle does not pass the first inspection,
it is sent back for adjustments and left in a pending status. At a given point in
time, about 100 vehicles are in the “pending” status across all inspection
stations
E.g., hospitals keeping excessive levels of medicines in stock, knowing
they won’t be able to use them before the expiry date.
Waiting
Task Waiting for materials or input data
Task waiting for a resource
Resource waiting for work (resource idleness)
Examples
o Approval-to-approval process: Request waiting for approver
o University admission process: Incomplete application waiting for additional
documents; batch of applications waiting for committee to meet
o Vehicle inspection process: A technician at a base of the inspection station
waiting for the next vehicle
Over-do
Defects
Correcting, repairing or compensating for a defect
Defect encompasses rework loops (because of defect)
Examples
o Travel approval process: Request sent back to requestor for revision
o University admission process: Application sent back to applicant for
modification; request needs to be re-assessed later due to incomplete
information
o Vehicle inspection process: A vehicle needs to come back to a station due to
an omission
Over-processing
Tasks performed unnecessarily given the outcome of the process
Unnecessary perfectionism
Examples
o Travel approval process: 10% of approvals are trivially rejected at the end of
the process due to lack of budget
o University admission process: Officers spend time verifying the authenticity of
degrees, transcripts and language test results. In 1% of cases, these
verifications uncover issues. Verified applications are sent to the admissions
committee. The admission committee accepts 20% of the applications it
receives
o Vehicle inspection process: Technicians take time to measure vehicle
emissions with higher accuracy than required, only to find that the vehicle
clearly does not fulfill the required emission levels
Over-production
Unnecessary process instances are performed, producing outcomes that do not add
value upon completion. These instances in the process should never have existed and
enough information should be provided at the initial stages of the process to stop
this from happening.
Examples
o Order-to-cash: In 50% of cases, issued quotes do not lead to an order
o Travel approval: In 5% of cases, travel requests are approved but the travel is
cancelled
o University admission: About 3000 applications are submitted, but only 800
are considered eligible after assessment
Example
Equipment
Rental Process: Wastes
Transportation
o Site engineer sends request to clerk
o Clerk forwards to works engineer
o Works engineer send back to clerk
Inventory
o Equipment kept longer than needed
Waiting
o Waiting for availability of works engineer to approve
Defect
o Selected equipment not available, alternative equipment sought
o Incorrect equipment delivered and returned to supplier
Over-processing
o Clerk finds available equipment and rental request is rejected because
equipment not needed
o Rental requests being approved and then cancelled by site engineer
Over-production
o Equipment being rented and not used at all
Stakeholder Analysis
While waste analysis primarily focuses on identifying internal inefficiencies, we will
now shift our attention to external factors through stakeholder analysis.
To conduct a stakeholder analysis, we must first identify all the individuals or groups
involved in the process, including customers, managers, workers, contractors, and
other relevant parties.
Stakeholder analysis is about gathering data from multiple sources by interviewing
stakeholders of different types and reconciling their viewpoints. In BPM, stakeholder
analysis is commonly used to gather information about issues that affect the
performance of the process from different perspectives.
There are typically five categories of stakeholders:
o The customer(s) of the process.
o The process participants (workers).
o The process owner and the operational managers who supervise the process
participants.
o The external parties (e.g., suppliers, sub-contractors) involved in the process.
o The sponsor of the process improvement effort and other executive managers
who have a stake in the performance of the process
Issue Register
The purpose of the Issue Register is to:
o maintain, organize and prioritize perceived weaknesses of the process (issues)
o Assess their impact both qualitatively (e.g. nuisance to the customer) and
quantitatively (e.g. financial loss).
o Sources of issues:
Input to a process modelling project
Collected as part of ongoing process improvement actions
Collected during process discovery (modelling)
Value-added/waste analysis
As a process analyst, you should talk to different people through a series of
interviews, you should take interview notes, and you need to compile them into a
table that we call it Issue Register.
The Issue Register can take the form of a table including the below columns:
o Issue identifier
o Priority
o Short name
o Description
o Assumptions
o Impact: Qualitative and Quantitative
o Possible improvement actions
Issue Register
Example
Pareto chart
Useful to prioritize a collection of issues
Bar chart where the height of the bar denotes the impact of each issue
Bars sorted by impact
Superposed curve of cumulative percentage impact
Have all quantitative impact in the same unit e.g., dollar values
You solve the issues from the one that has the highest impact to the one that has the
lowest impact.
Then a cumulative line is drawn with percentages that say that this issue alone
counts for 78% of the total value of the issues etc.
o If all issues are costing, for example, $10,000, then this issue is costing
$7,800, the second costs $9,700. The third issue is costing 100% of the total
value of what we are losing because of this issue.
o Usually we’ll have more issues e.g., 12-15
Your target is a Pareto target means you address those issues that come between
60-80% of the total value
80/20 rule applies here as well – where 80% of profit comes from 20% of your
customers. In here, 80% of the benefits or values can be reached by fixing 20% of the
issues – if you prioritise well, the first 20% of issues will reflect 80% of values.
Root-Cause Analysis
Our last qualitative analysis is the root-cause analysis
After conducting an analysis of our issue register, we prioritize the issue that has the
most significant impact on our organization. From there, we pinpoint the critical
factors and underlying root cause of the issue using a method called why-why
analysis. Alternatively, we can use a cause-effect diagram to approach the analysis
from a different perspective.
Why-Why Analysis
Why-why analysis can facilitate interviews and workshops with stakeholders. This
approach involves presenting specific instances of issues, such as instances of late
delivery to customers, and working with stakeholders to understand the underlying
reasons for each occurrence. The goal is to gain insight into why the issues that arose
and identify potential solutions.
Your task is to investigate why
the customer received the
product three days later than
expected without any prior
notification.
o To accomplish this, you
collaborate with
stakeholders to compile
a list of potential factors
that contributed to the
issue, also known as
first-level factors.
o Then, you analyze and prioritize these factors to identify the underlying
causes. This is where the why-why analysis comes in. By delving into each
factor and asking "why" it occurred, you can uncover the root causes of the
problem, up to five levels deep (going being the fifth level is not
recommended). Addressing all factors at the fifth level is necessary to resolve
the issue effectively. This approach is based on empirical practice and
experience.
Issue Register
Select the most important issues(s) and document them into an Issue Register. Make
sure that you document any assumptions that you make
o Write an issue register for this process. Including a column for qualitative and
quantitative impact
Main issue for customer:
o Long queue times, order is not ready when the customer comes for pickup
Need to put this in table format with priority, description with what is
happening to the customer
Note: example below missing priority
Why-Why Analysis
Why did the customer have to wait?
o The order is not ready
Why is the order not ready?
o I don’t know where the medicines bag is
Why can’t I locate the bag?
o I don’t have a tracking system
Example 2
Root Cause Analysis
Week 7 - Quantitative Analysis
To drill further into the analysis of process performance we can use probabilistic
simulations to evaluate as-if process performance as well as consider “what if”
scenarios. These simulations use the process model, and statistical information or
assumptions about activity performance. Large volumes of process executions are
simulated, and the results can be analysis to see issues and opportunities.
We will not cover the mathematical queuing and simulation models here, but if you
are interested to learn more, it is covered in Dumas et al (2018) Chapter 7.
How to calculate the Cycle Time (CT) during the execution of the process.
Cycle time has two components:
o The processing time (also called service time), which is the time actually
spent doing work, meaning doing activities of the process. This is when
someone is working on the process, i.e., doing work. It does not matter if it is
one or 10 people, as long as there is work done on this process instance.
o Waiting time, which is the rest of the time (when no one is working on the
process). Waiting time has several sources, most notably the fact that a given
activity may be ready to be executed, but the resource that will execute it is
not available. Waiting time could be in the form of:
Idle time: waiting time in the process
Transit time: time spent in transit between steps, working being
moved but no value is being added
Queue time: The time an item is lined up before a critical or
bottleneck resource: the work item is ready to go on, but is waiting for
the resources for the next step to get to it.
Setup time: time required for a resource to switch from one type of
task to another.
Once we know the processing time and the waiting time for every process instance,
we can calculate the average of the processing time and the average of the waiting
time.
The average processing time + the average waiting time is called the Cycle time.
In summary, Cycle time = Difference between a jobs’ start and end time.
Flow Analysis
Flow analysis is a technique to estimate and understand the
performance of a process, given a process model and performance
measures of the activities in the process.
For a process model, if we have information about the performance
of each activity in this process model, we can calculate the process
performance.
For example, if we are given a process model and the cycle time of each activity in
the process, we can use flow analysis to calculate the performance of the process. In
doing so, we can understand where the performance of the process is coming from,
that is, which activities in the process affect most the performance of the overall
process.
Note: Cycle time analysis: the task of calculating the average cycle time for an entire
process or process fragment
o Assumes that the average activity times for all involved activities are available
(activity time = waiting time + processing time)
o Pay attention to time units (e.g., weeks, days, hours, minutes…)
In this example, we have this process model and the cycle time of each activity in the
process. For the flow analysis, we should be able to calculate how many days on
average this process takes.
As a process analyst, you will take a sample and ask questions to the people in the
sample of “When did it end?”. You can also make some assumptions.
Calculating cycle time: sequential path
Assuming task A takes 10 hours and task B takes 20 hours, where the cycle time
includes waiting time and processing time, the cycle time is 30 hours.
Rework Loop
Back to example
Example 2:
Let us assume that in 90% of the cases the credit is granted. What is the average
cycle time for the process segment in this figure?
Example 3:
Consider the process model given below, calculate the cycle time
under the following assumptions:
o Each task in the process takes 1 h on average.
o In 40% of the cases the order contains only Amsterdam products.
o In 40% of the cases the order contains only Hamburg products.
o In 20% of the cases the order contains products from both warehouses.
Example:
Must find out when someone started working on the task and finished working on
that task (get information from database)
o Here, the system does not include waiting time.
The processing time would be 8.9 hours. The cycle time efficiency is processing time/
cycle time. We first have to change 8.65 days to hours.
o We assume 8 hours a day so 8.65 x 8 hours.
In this case, it is not a good process as most of the time is waiting time.
Example 2:
Calculate the overall cycle time, theoretical cycle time (processing time), and cycle
time efficiency of the ministerial enquiry process introduced in Example 3.7 (page
90). Assume that the rework probability is 0.2 and the waiting times and processing
times are those given in the table below:
In the treasury minister’s office, once a ministerial inquiry has been received, it is first
registered into the system. Then the inquiry is investigated so that a ministerial
response can be prepared. The finalization of a response includes the preparation of
the response itself by the cabinet officer and the review of the response by the
principal registrar. If the registrar does not approve the response, the latter needs to
be prepared again by the cabinet officer for review. The process finishes only once
the response has been approved.
Solution:
CTE of Pharmak O2C as-is process
If cycle time does not include processing time, you add each of the processing time
to each of the CT.
If cycle time includes processing time, you do it separately and leave PT to the end
and calculate.
It is not a good process
as there is a lot of
waiting time.
Quiz
1. What is the average CT with the probability of the upper branch being executed
being 90% and the lowest branch 10%?
a. 29 units = 10 + (0.9x20)+(0.1*10)
2. What is the average cycle time for the process segment in this figure?
a. 10+max(20,10) = 30
3. Let’s assume that in 90% of the cases the credit is granted. What is the average cycle
time for the process segment?
a. 1+3+3+(0.9*1)+(0.1*2) = 8.1
Introduction
Process redesign falls in the middle of the BPM lifecycle, following the analysis
stages.
During the analysis phase, we created an issue register and conducted root-cause
analysis to identify the process issues. In this module, we will apply specific principles
and heuristics to brainstorm and derive ideas on how to solve these issues and
improve the process.
The application of heuristics and principles will lead to improvements of certain
performance measures which ultimately leads to improvement in areas such as Cycle
Time.
This module will demonstrate to us how to enhance our process. To begin, we must
have a comprehensive understanding of our process. This involves creating a well-
designed BPMN model of the as-is process and gaining a thorough understanding of
how it operates.
Our objective is to create the to-be model, which will serve as the blueprint for our
process implementation. This area requires creativity, you need to come up with a
variety of ideas. To accomplish this, we will use redesign heuristics, which are a set of
principles that guide us on how to develop these ideas.
Transformational Methods
As is process
Here, accounts payable will end up
with a pile of receipts, invoices,
and purchase orders.
They will have to do three-way
matching, where they take an
invoice, look at the corresponding
purchase order and check the
corresponding delivery receipt.
A suggestion by the company that Ford hired was for Accounts Payable to
computerise everything given to them (as a lot were done on paper), which would
save 20% of the process.
Outcome
75% reduction in head count
Simpler material control
More accurate financial information
Faster purchase requisition
Less overdue payments
Lessons:
o Why automate something we don’t need to do at all?
o Automate things that need to be done.
1. Capture information in the process once and only once at the source.
a. Shared data store
i. All process workers access the same data
ii. Don’t send around data, share it!
b. Self-service
i. Customers capture data themselves
ii. Customers perform tasks themselves (e.g., collect documents)
2. Subsume information-processing work into the real work that produces the
information
a. Evaluated receipt settlement: when receiving the products, record the
fulfillment of the PO, which triggers payment
b. Do not have tasks that simply “catch” information. Do not have data entry
tasks in the process. Do not have a task where you only enter data. These
tasks are not constructive. Ensure the task is completed as part of their job or
work so the data gathering task is part of the actual work. This will allow the
worker to subsume the information-processing work (data entry) into the real
work that produces the information. Ensure information is part of the actual
work and it is recorded when completed.
3. Have those who use the output of the process drive the process
a. Vendor-managed inventory (VMI)
i. E.g. if Woolworths need products from suppliers, an inventory
manager will be looking at this to see that products are running low.
However, in some cases, by the time they order the products, the
inventory might have already ran out. The problem here is that the
inventory manager may not have the highest incentive to do this.
ii. The vendor would have the highest incentive as they want to sell their
products so they should have access to Woolworth’s stock level and
issue the order themselves when stock is running low.
b. Scan-based trading
i. Vendor-managed inventory may have some issues if it is borderline
unethical, scan-based trading is used.
ii. More common in manufacturing
iii. If a factory wants lots of materials, the supplier brings a track
container and they put it into the manufacturer with a door and
scanning machine at the door. What is inside the container is for the
supplier. When an employee needs to use an item from the container,
they will open the door to the container and scan the item.
Immediately, the supplier knows that the item is scanned and it now
belongs to the factory. The supplier can then invoice the factory.
iv. The items are the supplier’s as long as the items are not scanned.
v. Reduces warehouse cost
c. Push work to the actor that has the incentive to do it
i. We can refer back to the idea of performing PO matching at delivery,
rather than postponing this till the invoice is received.
o Left diagram is not convenient for the client as they have to make the claim
themselves, wait for approval, arrange repair, get invoice etc.
o Therefore, the work is not being done by the one who has incentive to do it
(the glass vendor)
o The process can be redesigned with the glass vendor in mind, where the
claim should be made by the vendor and insurer pays directly to vendor.
4. Put the decision point where the work is performed, and build control into the
process
a. Empower the process workers
i. E.g., letting people rent equipment without approval if under $500,
but only approval when above $500
b. Provide process workers with information needed to make decisions
themselves
c. Replace back-and-forth handovers between workers and managers
(transportation waste) with well-designed controls
i. Instead of controlling every single invoice, you let the invoices go
through, and then control based on statistical basis by checking a
sample at the end of month etc.
When you stretch it in one direction, it will shrink in the other direction.
Flexibility
Ability to react to changes in:
o Workload
o Customer demands and expectations
o Resource and business partner availability and performance
E.g., people are on leave, business partner lowering their performance
Example: Following natural disasters (e.g., storms), the number of home insurance
claims increases by tenfold
o To address this surge, flexibility is required at:
Resource level: Staff redeployment, faster performance
Process level: Performing tasks differently to speed up the front-end
Management: Relaxing business rules and controls where possible
E.g., during COVID when lots of people lost jobs and were queuing up at Centrelink,
the government added flexibility in letting people make claims online
Redesign Heuristics
The Devil’s Quadrangle is one part of the theory. The second part of the theory is the
improvement patterns by looking at how companies have been improving business
processes for many years. Over time, modifications have been implemented to
enhance the process, and we record these alterations as redesign heuristics or
patterns.
We will go through 9 most common of the heuristics, broken down into:
o Task Level
Task elimination
Task composition/decomposition
Triage
o Flow-level
Re-sequencing
Parallelism enhancement
o Process Level
Specialization & standardization
Resource optimization
Communication optimization
Automation
Task Elimination
Eliminate non-value-adding steps wherever these can be isolated
o Forward, send, receive, ...
Consider reducing manual control steps
(checks & approvals) by:
o Skipping them where feasible
o Replacing them with statistical
controls
o Skipping them selectively
Consider trade-off between the cost of the
check and the cost of not doing it
o Examples:
Procure-to-pay process: some types of employees are empowered to
trigger isolated purchases below $500 without supervisor approval
Order-to-cash process: invoices from trusted suppliers under $1000
are not checked on a one-by-one basis
University admission process: authenticity check is very expensive, yet
it leads to only 1% of applications being rejected
Task Composition/Decomposition
Consider composing two tasks to eliminate transportation and reduce
“context switches”,
o Composition example:
Procure-to-pay process: Merging two checks: “Check
necessity of purchase” and “Check budget” done by 2
different people to be done by 1 person – reduces waiting time and
improves flexibility
OR, consider splitting a task into two and assign to separate, specialized
resources
o Make-to-order process: Separate a single thick “prepare quote” task
into “prepare bill of materials”, “prepare production plan” and
“estimate costs and delivery time”
o Useful if they can be split into tasks requiring different expertise
Triage
Specialize a task: divide a general task into two or more
alternative tasks
o Specialization example:
Procure-to-pay process: Separate
approvals of small purchases, medium
purchases and large purchases
Re-sequencing
Re-order tasks according to their cost/effect
ratio to minimize over-processing
o Postpone expensive tasks that may
end up not being necessary until
the end
o Put knock-out checks first in order
to identify problems early
Make-to-order process: If “Prepare production plan” is time-consuming, postpone it
until after the quote price has been tentatively accepted by the customer
Procure-to-pay process: If “Check necessity of purchase” leads to 20% of knock-outs
and “Check budget” leads to 2%, perform “Check necessity of purchase” first
University admission process: authenticity check (very slow) leads to
1% of applications being rejected while committee’s check leads to
80% of applications being rejected. Put committee’s check first
Parallelism Enhancement
Parallelize tasks where possible to reduce cycle time
Example:
o Procure-to-pay process: Parallelize “Approve budget” and
“Approve necessity of purchase”
o Make-to-order process: After “Prepare bill of materials”,
perform “Prepare production plan” and “Estimate costs”
in parallel
Process specialization/standardization
Process specialization
o One process is split into multiple ones: by customer class, by geographic
location, by time period (winter, summer), etc.
o Resources are split accordingly
o Example:
Procure-to-pay process: One process for Direct procurement (e.g. raw
materials) and one for Indirect procurement (MRO - Maintenance,
Repair and Operations)
Claims handling process: One claims handling process for the summer
season (stormy season - peak) and one for the winter season (off-
peak)
Process standardization
o Two processes are integrated
o Resources are pooled together
o Example:
Claims handling process: Integrate claims handling for motor
insurance across different brands of a group
Resource Optimisation
Use resources of a given type as if they were in one room
o Avoid one group of people overloaded and another (similar) group idle
Let people do work that they are good at
o However, avoid inflexibility as a result of specialization
When allocating work to resources, consider the flexibility in the near future
o Allocate work to specialized resources first
o When there is a task, give to junior person first because the senior might be
needed for a harder task
Avoid setups as much as possible
o Chain multiple instances of the same task [sequential]
o Batch multiple instances of the same task [parallel]
Communication Optimisation
Automate handling, recording and organization of messages
Monitor customer interactions, record exceptions
Optimize
o Number of interactions with customers and business partners
Gather sufficient information to get to the next milestone (reduce
external interactions)
o Type of interaction (synchronous vs. asynchronous)
Synchronous interactions effective to resolve minor defects
Asynchronous to notify, inform, resolve major defects, request
additional information to reach next milestone
o Timing of interactions
Front-loaded process: bulk of information exchange and processing
happens upfront
Complete-Kit Concept: “Work should not begin until all pieces
necessary to complete the job are available” - Boaz Ronen
Principles for complete-kit process design
o Provide complete and easy-to-follow instructions for
those who will initiate the process.
o If a process cannot start, the client should be notified
of all defects that could be reasonably identified at the
onset of the process
o Consider the trade-off between “incomplete-kit”
process initiation vs. roundtrip to revise and resubmit a
request
Back-loaded process: bulk of information exchange and processing
happens downstream
Example: CVS Pharmacy in early 2000s
Automation
Use data sharing (Intranets, packaged enterprise systems) to:
o Increase availability of information to improve visibility and decision-making
(subject to security/privacy requirements)
o Avoid duplicate data entry and transportation
Use network technology to:
o Replace physical flow (e.g., paper documents) with information flow
o Enable self-service via e.g., online forms and Web data services
Use tracking technology to identify and locate materials and resources
o Identification: Bar code, RFID
o Location: GPS, indoor positioning
Use business rules technology to automate information processing tasks (including
decisions)
Automate end-to-end processes with a dedicated BPM system or system with
process automation functionality
Analytics Investigations
Customer Journeys
The Approach
Context – The Break to Fix Process
Recommended Actions
Qualified Benefits
Insights
We analysed 1.9 million Sales document items and 445K Sales orders created from
1st Mar – 5th May 2022 for 2 Sales Organisations (0500,0601) in Signavio Process
Intelligence.
Key Insights
o 33 % of Sales Order being changed of which 18% are related to Master data-
Improvement here will increase the cycle time and efficiency
o Return order and delivery changes, attributes to about 82% manual work-
Scope for automation
o Setting Delivery Block in Sales order seems to be manual here- Automation
can be done at config level or custom exit functionality
o 17% return orders are created of which Trade return constitute the major
part- Business analysis required on why there is a high rate
o Cycle time for Sales Organisation 0500 appears to be thrice than Sales
Organisation 0601- Business analysis required
o Credit release cycle time takes an average 3 days – Scope of improvement by
automating /improved turn around in approvals
o About 26 % of order items are getting into back orders- ‘material x’ being one
material group to be focused on- Stock requirements/MRP settings can be
fine tuned to improve the cycle time here.
o 160K sales order items appear as rejected items- Scope to improve rework
and efficiency
Focus Areas
Quiz
1. Which of the following processes is the least suitable for being redesigned?
a. The production of a wafer stepping machine
i. If you redesign it, you probably won’t do much since it would just be
manufacturing
b. Providing financial advice on composing a portfolio
c. Dealing with a customer complaint
d. Obtaining your student ID card at UNSW
2. For a purchase order (PO), when do you match received items?
a. Upon receiving the delivery/items
3. Put the decision point where the work is performed, and build control in the process
a. Empower the process workers
b. Replace back and forth handovers between workers and managers
4. Self-service is to give work back to the customer and the customer thanks you
because you make them more powerful. This is:
a. Automation
Solution:
1) Notification. Customer has to wait when picking up orders. The description of the
issue can be found in the issue register. There could be seen that the Pharmak loses
approximately $100 000 per year because of issue 1 and $60 000 because of issue 2.
The two issues are closely connected, so they should be fixed together.
o Send a notification to the customer via the mobile app or via SMS once the
order is ready, so that they don’t have to come and wait for nothing. Also,
give them an estimated timeframe when the order would be ready when they
put their order, so they are happier because they can plan their day better.
2) Order tracking. To prevent the “where’s the bag” events, which further contribute
to issues 1 and 2, the pharmacy should introduce an order tracking system using bar
codes or QR codes. When a bag is assigned to a customer (by the technician), a bar
code is attached to it and this bar code is associated to the customer record. Every
time the bag moves from one station to the next one in the process, the
corresponding actor (technician, pharmacist, customer service rep) will scan the bar
code and mark the order as “under collection”, “collected”, “ready for pickup” and
“delivered” respectively.
3) Automated medicine check. Push the manual check performed by pharmacist to
an automated check, performed when the customer places the order. This redesign
suggestion is based on the issue number 3 in the issue register. The meaning is that
the pharmacist does not have to check manually the orders to decide if there is a
need for additional interview. This could be easily done by computer programs.
o If the customer places an order, have him / her fill in the form of their health
indicators, special conditions, and any other relevant information. Then have
the system to fetch the information from the national health databases and
perform automatic checks if the customer needs to be aware of some drugs
or anything else. If the system detects something, the order should get the
marking that there is a need for pharmacist interview. The same information
is then told to the customer as well, so he/she knows, that the order must be
collected by the pharmacist, which is a bit more time consuming.
o The cost to pharmacist salaries are approximately ($60 000) per year, because
of this issue. The IT system development would cost about $120 000 as a
purchased development service. The whole project would benefit in two
years’ time. It would also increase customer satisfaction, because they know
if they need to talk to the pharmacist or can just get the package.
Reference from week 5 issue register on Pharmak
The result of the changes let us get to the to-be process model which we document
in BPMN.
The next step is to take this to-be process model and turn it into an executable
process model. We can automate this model with the use of an Information System
(IS).
o There are a few ways of doing this:
One way is to take the to-be process model as the basis, derive
requirements and build an IS using whatever software development
methods we want to apply.
The second approach is suitable for scenarios where there are
numerous process changes, and it involves transforming the complete
process model into an executable format. The outcome is to deploy
the artefact in a specialised application server – this is called a
Business Process Management System (BPMS) and then generate the
forms, messages, inputs, outputs, and all requirements to execute this
process. This is how to take a to-be process model and turn it into
something that has all the information needed to execute the
process.
The Business-Engineering Gap
The challenge is you have to deal with 2 different communities – the business
analysts who are producing the “to-be” process models and the process owners, as
well as the IT experts.
These people need to turn the process into an artifact.
In BPMN, we have 3 types of tasks (manual which has hand icon, user which has
person icon, and automated task).
In BPMN, there are 4 types of automated tasks.
o Send Task: When we want to notify the customer that an order has been
delivered.
o Receive Task: If we are in the process and expecting to receive a message
from the customer e.g., customer has to save confirmation email for order
o Service Task: comes from web service e.g., making query to system to report
remote procedure
o Script Task: do not require communication with external system. Needs some
computation to take place e.g., calculate entitlement for insurance claim.
Example:
For example, if we have this manual task to retrieve a product from the warehouse,
where someone has to go and pick up the product and gets it ready for shipment. We
want to turn it into a user task where someone will enter data into the system, so
how do we do that? What should we do to turn a physical manual task involving
movement into a user task – not an automated task but a user task where someone
enters it into a system?
o We use what is called tracking technology – like barcodes, QR codes, RFID,
etc…
o So we turn it into something where someone will scan the barcode when the
product is ready for shipment, which means somewhere else the code of that
product should be recorded in the system, therefore we might need to split
the task into 2:
1. Initially, we allocate an exclusive code to each order. Subsequently, we
generate a label featuring a QR code or barcode, which is then affixed
to the product that is about to be retrieved from the warehouse. After
printing the barcode label, a manual process follows where the label is
handed over to a worker who attaches it to the product.
2. Following that, the label is scanned at the shipment bay and may be
placed on a pallet, as an instance, in preparation for shipping.
o So what do we do? We can replace this manual task with two parts:
1. We notify the warehouse workers to retrieve the product from the
warehouse and attach the corresponding label to it.
2. Subsequently, the next step involves tracking the moment when the
label is scanned at the shipment bay, indicating the completion of this
task.
As you can see, we replaced the manual task with a send task and a receive task.
How do we do this? How do we notify the people at the warehouse? Do we send an
email or push a list to the people at the warehouse with the copy of the labels they
have to stick to the product, or we might do it in bulk (all orders at the same time)?
Here, we are not automating the manual tasks, we are tracking the tasks. When we
track these tasks, we have to isolate the manual tasks and automate the rest by
leaving the manual tasks out.
Alternative: Isolate Manual Tasks
Process for university admission, where there is a lot of checks done on the
application.
At one stage, there is a batch of applications sent for assessment to the committee.
This committee is not supported by the systems, they only do papers
As a systems analyst, we assume you have tried to see if you can modify the process
to automate but it did not work due to politics and costs.
We have split the task using the green line, where we try to automate what is on the
left and right before and after the manual task.
Example 2:
Prescription fulfillment process:
o Once the prescription passes the insurance check, it is
assigned to a technician who collects the drugs from
the shelves and puts them in a bag with the
prescription stapled to it.
o After that, the bag is passed to the pharmacist who
double-checks that the prescription has been filled
correctly.
o After this quality check, the pharmacist seals the bag
and puts it in the pick-up area.
o When a customer arrives to pick up their prescription,
a technician retrieves the prescription and asks the
customer for their payment.
Assume the pharmacy system automates this process.
Identify the type of each task and link manual tasks to the
system.
o Check insurance automated task as we can connect this to the insurance
provider using some sort of APIs and find out if the customer is insured and
how much they are eligible to pay
o Collect drugs from shelves manual task at the moment.
Can be turned into user task, where when the order arrives, we can
create a barcode that we attach to bag. Technician can scan bar code
and mark the order has been collected. System now knows the
product has been collected.
o Check quality manual task at the moment.
This can be turned into a user task, where the pharmacist can take the
bag, scan it and check that it’s completed using the barcode.
o Seal bag and add to pickup area manual task at the moment
This cannot be turned into user task, so we are isolating or eliminating
it.
o Retrieve prescription bag manual task at the moment
In the picture, it has been eliminated but we can turn it into a user task
(PREFERRED)
o Collect payment manual task at the moment
This could be automated by having eftpos machine linked to system
where the customer can just scan their card and pay.
Step 3: Complete the Process Model
The third step is to complete the process model. When we are doing the conceptual
process model, we might have exceptions that are very infrequent. When automating
something, we have to consider these exceptions and the path they take.
Usually, we provide a workaround to deal with them if they are extremely
infrequent.
If they are not too infrequent, we might need to integrate these exceptions into the
process itself.
We have to find out the exceptions in this to-be process model and determine how
we are going to handle these exceptions.
How do we find exceptions in a process, here are some rules to consider:
1. Every time we send something, something can go wrong. Whether we are
sending a package to a vendor, a supplier, or a customer, we must bear in mind
that there may be a lack of response or delay in receiving a response. We need
to weigh our choices carefully, considering alternative routes, and evaluating
the performance of the entire process before making a decision on how long
we are willing to wait.
2. Second is to determine what messages they might send. We use what is called
CRUD analysis. This is to determine which objects are involved in the process.
For example, if we are dealing with POs, we send POs, requests for quotations,
or we send invoices, etc. We need to know what CRUD operations we need to
provide to the external parties on those objects. If we send an invoice, do we
need to provide a way for the customers to be able to view the invoice? Do we
provide a way to update the invoice? Or do we need to provide an option to
cancel the invoice? To consider all of these operations, the best approach is to
identify the business objects that drive the process. Then, we must determine
the create, read, update, and delete capabilities of each party that may access
the data. This analysis enables us to identify the additional operations that our
system should provide to external parties to manipulate the objects of the
process. Also, we need to specify which electronic business objects we have in
the process, and we need to determine who can create, read, update or delete
them. For every gateway or decision point, it is necessary to ascertain the
objects that are required to determine the next course of action after passing
through the gateway.
Example
In our example we start with the to-be process model and we ask ourselves what can
go wrong in “check raw materials”.
o E.g., the materials could be unavailable so we cannot obtain them on time to
manufacture the product.
o In this case, we notify the customer and have the purchase order cancelled.
When we check stock availability from the database, we can find out which product
is available and the quantity. We can also check for raw materials in the warehouse.
So generally, every task in the process should have an input and output.
o This way, it gives you the scope of implementing the task.
Step 4: Adjust Task Granularity
Principle: BPMSs add value if they coordinate handovers of work between resources.
o Aggregate any two consecutive tasks assigned to the same performer
Sometimes, we may find that these tasks are the same. Therefore, we
should put them together if they are done by the same person to save
time.
The more tasks you have, the more you will have to automate.
o Split tasks if they require different
performers
For “enter and approve money
transfer”, these might be done by 2
different people so we split the task.
When there are two tasks performed by the same performer but it is necessary to
split them because there is empty time between those 2 tasks.
E.g., post documents and after a few days, you receive them. Therefore, they cannot
be performed by the same person in one single go.
Our Example
Quiz
1. When automating a process, all parts of that process that cannot be automated
should be removed
a. False
2. When automating a process, it is NOT recommended to breach each task to as many
steps as possible
a. True, it may create more problems for you
3. The to-be-executed model is:
a. A bridge between the conceptual and executable process model
b. The responsibility of the system analysts
4. What is the first step to reach the executable process model?
a. Identify the automation boundaries
5. When we automate a process, we must automate all tasks in it, otherwise the
process is considered not automated.
a. False, we cannot automate everything – we must put a boundary and the
human role must still be there.
b. For UNSW Admissions, you must have someone look at the application. For
example, if someone was doing medicine, would you trust someone getting in
without looking at their profile?
Pharmak Example
Collect and pack medicines in bag can split into “collect medicines” and “check
and pack medicines”.
o Collect medicines can be a service task (automated)
o Check and pack medicines will be a manual task
Put bag in QA area manual task
Pick up bag manual task
Verify customer details user task
o This is a user task as something has to do something using a computer
Verify medicine correctness user task
Fix issue manual task
Check for special instruction manual task
Mark the bag with Red X manual task
Pack bag in pickup area manual task
When the customer comes and the medicine is not ready, you can automate by
placing lockers where you send a PIN code to the customers’ app where they can
scan using smartphone or enter code manually to open the locker and get the
medicine.
By doing this, a lot of this second part of the process can be automated. The
customer will only have to come to the service desk when there is a red X on the
back of the medicine.
NOTE: THIS IS TO-BE EXECUTED MODEL ONLY THAT USES THE ICONS ON BPMN. FOR
ASSIGNMENT, IT IS TO-BE ONLY