Module 4:

TIME STUDENT LEARNING OUTCOMES LEARNING

(SLOs) CONTENT

Week 2  Understand the process discovery Process Discovery

and its setting The Setting of Process
 Discuss the different process modeling Discovery
method  Discovery Methods
 Understand the different process model  Process Modeling
 quality assurance Method
 Process Model
Quality Assurance

Chapter 5 - Process Discovery

All truths are easy to understand once they are discovered;

the point is to discover them.
Galileo Galilei (1564–1642)

The previous chapters showed how to create a BPMN model. This chapter goes further
by showing how to create models that are both correct and complete. To this end, one
needs to thoroughly understand the operation of a business process, and one needs to
possess the technical skills to represent it in an appropriate BPMN model. These two
types of skill are hardly ever unified in the same person. Hence, multiple stakeholders
with different and complementary skills are typically involved in the construction of a
process model.
This chapter presents the challenges faced by the stakeholders involved in the lead-up
to a process model. Then, we discuss methods to facilitate effective communication and
information gathering in this setting. Given the information gathered in this way, we
show step by step how to construct a process and what criteria should be verified
before a process model is accepted as an authoritative representation of a business
process.
5.1 The Setting Of Process Discovery
 Process Discovery = The act of gathering information about an existing process
and organizing it in terms of an as-is process model.
 Modeling can only start when enough information is gathered.

 Phases of process discovery:

1. Defining the setting = Assembling a team in a company responsible for
working on the process.
2. Gathering information = Building an understanding of the process using
different discovery methods.
3. Conducting the modeling task = Organizing the creation of the process
model in a systematic way.
4. Assuring process model quality = Guarantee that the resulting process
models meet different quality criteria to ensure trust in the model.

Aspect Process Analyst Domain Expert

Modeling Skills strong Limited

Process Knowledge limited Strong

 
Process Analyst
o Responsible for driving the modeling an analysis of a business process.
o Not familiar with all the details of the business process.
o Secure commitment of various domain experts for providing information
on the process.
 Domain Expert
o Any individual who has intimate knowledge about how a process or
activity is performed.
o Multiple domain experts can be involved, each has a different perspective
on the process.
o All the domain experts should jointly have insight into the whole process.
o Ex. Process participants, process owner/manager, suppliers, customers.

Exercise 5.1 Consider the following two tasks, and explain their difference:
 The task of modeling the process of signing a rental contract in your city.
 The task of modeling the process of getting a license plate for your car in Liecht-
enstein as a foreign resident.
  Process Discovery Challenges
1. Fragmented process knowledge
 Business process consist of a set of logically related activities.
 These activities are assigned to specialized participants.
 Domain experts know little about the whole process and a lot
about their own task.
 Different ideas of the expected output.
 Rules of the process are not explicitly defined in detail.
 Difficult to puzzle the different views together.
 A process analyst needs to talk with several domain
experts.
 Multiple iterations:
i. Receive input from domain experts.
ii. Proposal for resolving inconsistencies.
iii. Receive approval/feedback/input from domain experts.
iv. etc.
2. Thinking in cases
 Domain experts can easily describe activities performed for a
specific case, but have problems responding to general questions
about how the process works in a general way.
 " You cannot really generalize, every case is different. "
 Process analysts need to keep asking questions to reverse
engineer the conditions that govern the routing decisions in
the business process
3. Lack of familiarity with process modeling languages
i. Domain experts are often not trained to create and READ
process models themselves.
ii. The process analyst often needs to explain the model in a
natural-language to get feedback from the domain experts .
Exercise 5.2 An online book retailer faces a problem with its order process in terms of
processing time. In order to identify the root cause of the problem, the company decides
that every team involved with the order process should model its part of the process.
Why could this approach be problematic?
 Profile of a process analyst
o Five Factor Model (certain personal dispositions that help becoming an
expert in process analysis)
1. Openness (appreciating art, emotion, and adventure)
2. Conscientiousness (tendency to self-discipline, achievement
and planning)
3. Extraversion (being positive, energetic, and seeking company)
 4. Agree-ableness (being compassionate and cooperative)
5. Neuroticism (being anxious, depressed and vulnerable).
o Process analists apprear to be strong in Conscientiousness and
Extraversion.
o Creation of positive atmosphere despite enterprise-internal politics.
o How process analists navigate trough a project is mainly influenced by
former experiences.
o Expert process analists pay specific attention to the following aspects:
1. Getting the right people on board.
 Make sure the hierarchy backs the process participant
involvement in the process discovery
2. Having a set of working hypotheses on how the process is structured
at different levels of details.
 Prepare a extensive set of questions and assumptions to be
discussed in workshops or interviews.
3. Identifying patterns in the information provided by domain experts.
 Pieces of information typically refer to specific control
structure. (gateways etc.)
4. Paying attention to model aesthetics.
 Models have to look nice to be engaging to a wide audience.
 Us the right level of abstraction
 You should not show a super-detailed model to an
executive-level manager.
Exercise 5.3 You are the manager of a consulting company, and you need to hire a
person for the newly signed process analysis project with an online book retailer.
Consider the following two profiles, who would you hire as a process analyst?

5.2 Discovery Methods

 Discovery methods are techniques for gathering information.

 Evidence based discovery

o Consists of:
 Document analysis
 There is documentation material available that can be related
to an existing process
 Good for getting familiar with the process, the environment and
for creating hypotheses.
 Issues:
 Documentation is not available in a process oriented way.
 The level of granularity of the material might not be
appropriate.
 Many documents are only partially trustworthy.
  Outdated, do not show how it is in reality

 Observation
 Directly follow the processing of individual cases.
 The process analyst can either play:
 Active role such as a customer.
 Process analyst triggers the execution of a process.
 Records the steps that are executed and the set of
choices that are offered.
 Provides a good understanding of the boundaries of
the process and its essential milestones.
 Process analyst will only see parts of the process that
require customer interaction.
 Back office processing remains a black box.
 Passive role such as an observer.
 Appropriate for understanding the entire process.
 Requires access to the people and sights where the
process is being worked on.
 Requires approval from the managers and
supervisors of the corresponding teams.
 People can change their behavior under observation.
 Work faster and more precise.
 Displays how a process is conducted today in contrast
to document analysis that usually captures the past.
 Automatic process discovery
 Use of event logs stored by information systems
 Individual case of the process
 Specific activity of the process
 Precise point in time
 Log information can be misleading (ex. after a system
crash)
 Creates rather complex models.
Exercise 5.4 Imagine you would be assigned the task of modeling the process of how a
book order is processed by your favorite online book retailer. How can you
systematically gather the required pieces of information about this process?
 Interview based discovery
o Labor intensive.
o Interviews have to be conducted with various domain experts involved in
the process. (due to everyone not having the same amount of knowledge)
 Domain experts feel more comfortable with using a free-form
interview than a structured one.
  Information can be hold back when not specifically asked.
 The interview usually describes an sunny-day scenario.
o Expertise of the process analyst is required for understanding how
individual cases are executed in order to create an meaningful process
model.
o Multiple interviews to check if the independent process model is correct.
 Important to ask what happens when something goes wrong or
unexpected cases are handled.
o Strategies for scheduling interviews:
 Upstream = Starting backwards from the result of the process.
 Understanding what input needs to be available before a
activity can start.
 Downstream = Starting forwards from the beginning of the
process.
 Understanding which decisions are taken at a given stage.
Exercise 5.5 Consider that the order process of your favorite online book retailer has
ten major activities that are conducted by different persons. How much time do you
need approximately for creating a process model that is validated and approved by the
process owner? Make appropriate assumptions.
 Workshop based discovery
o Offers the opportunity to get a rich set of information on the business
process.
o Contributors get immediately used to the modeled process.
o Involves more participants.
o Uses 5x 0.5 day sessions.
 Restricting speech time.
 Everybody should express their perspective.
o Lighter method: using sticky notes to represent task/event.
 Purpose: Build a map of activities and their temporal ordering.
 Tasks/events need to be the same level of detail.
o Involves additional roles for facilitating the discussion and operating the
modeling tool.
 Facilitator = Takes care of organizing the verbal contributions of
the participants.
 Tool Operator = Directly entering the discussion results into the
modeling tool.
 Domain Experts = Participate in the discussion.
 Process Analyst = Participate in the discussion.
 Process owner = Participate in the discussion.

 Strengths and Limitations

Aspect Evidence Interview Workshop

Objectivity High Medium-High Medium-High

Richness Medium High High

Time Consumption Low-Medium Medium Medium

Immediacy of Feedback Low High High

 Objectivity
o BEST: Evidence based
 Existing documents, logs and observation provide unbiased account of
how the process works.
o Interview en workshop based
 Rely on descriptions and interpretations of domain experts who are
involved.
 Risk that those persons may have perceptions and ideas of how the
process operates, may not be correct.
 Domain experts might opportunistically hide relevant information
about the process because of fear for loss of power, influence or
position in a political environment.
 Richness
o BEST: Interview and workshop based
 Domain experts are a good source to clarify reasons and
objectives for why a process is setup as it is.
 Provides history of the process and the surrounding organization.
 Provides information about the personal agenda of the different
stakeholders.
o Evidence based
 Might show issues or raise questions, but do not provide an
answer.
 Provides insight into strategic considerations when documented in
white papers.
 Does not tell anything about the personal agenda of the different
stakeholders.
 Time consumption
o BEST: Evidence based
 Can be easily made available to a process analyst.
 Automatic Process Discovery costs a lot of time extracting,
reformatting, and filtering of event logs.
  Passive observation requires coordination and approval time.
o Interview and workshop based
 Conducting interviews and workshops is time consuming.
 Interviews can take many iterations
 Difficult to schedule workshops with various domain experts on
short notice.
 Immediacy of feedback
o BEST: Workshop based
 Inconsistent perceptions can be resolved directly
o Interview based
 Offer opportunity to ask questions
 Interviews take many iterations to come to the right model .
o Evidence based

 Questions can only be answered by talking to a domain expert .

 Best to use an mixture of the discovery methods.

Exercise 5.6 Consider the following two companies. Company A is young, founded
three years ago, and has grown rapidly to a current toll of 100 employees. Company B
is owned by the state and operates in a domain with extensive health and security
regulations. How might these different characteristics influence a workshop-based
discovery approach?
Exercise 5.7 In what situations is it simply not possible to use one or more of the
described discovery methods?

5.3 Process Modeling Method

Modeling a process in the discovery phase via an predefined procedure.
STEP 1: Identify the process boundaries
 GOAL: Essential for understanding the scope.
o Has been partially done with the definition of a process architecture.
 TODO: Identify events that trigger the processes.
 TODO: Identify events that identify possible process outcomes.
STEP 2: Identify activities and events

 GOAL: Identify the main activities of the process.

o Advantage: Domain experts will be able to state what they are doing, even
if they do not "know" they are part of a bigger process.
o Documents might explicitly mention activities (ex. A set of work
instructions).
o When the model is getting too complex, focus only on the main activities
and events.
 TODO: Identify activities that take place within the process.
 TODO: Identify intermediate events that take place within the process.
STEP 3: Identify resources and their handovers

 GOAL: Identify resources and their handovers.

o Who handles the activities and events.
 Definition of pools and lanes.
 Assignment of activities and events to a pool or lane.
o Handover points are important so that the next worker knows what has
been done.
 Make assumptions explicit.
 Identify parts of the process which can be studied in isolation from
the rest.
 TODO: Create pools and lanes.
 TODO: Place events and activities in the correct pool/lane.
 TODO: Define handover points (where is the work handed over to another
resource).
STEP 4: Identify the control flow
  GOAL: Create the control flow using gateways.
 TODO: Create decision points ((X)OR-splits/joins) with their relevant conditions,
where possible.
 TODO: Create AND gateways where possible.
 TODO: Create rework and repetitions using loop structures where possible.
 TODO: Create message flows where information is exchanged between multiple
pools/businesses.
Exercise 5.8 What is the relationship between the type of a gateway and the conditions
of the subsequent arcs?
STEP 5: Identify additional elements
 GOAL: Capture the involved artifacts and exception handlers.
o Additions of data elements and exceptions depends on the modeling
purpose.
 TODO: Add data objects, data stores and their relations to activities and events
via data associations.

5.4 Process Model Quality Assurance

 Gathering information and organizing it in a process model is often done in a
sequential way, and not simultaneously, there is a need for various steps of
quality assurance.
Exercise 5.9 Have a look at Fig. 5.5. Explain what exactly is going wrong in the
unsound process model fragments.

Syntactic Quality and Verification

  Syntactic quality = Producing a process model that complies with the rules and
guidelines.
o Following the modeling language rules (ex. BPMN, no sequence flows across
pools)
 Verification = Structural and behavioral correctness.
o Structural correctness = Types of elements used and how they are connected.
 ex. An activity should always have an incoming and outgoing arc.
o Behavioral correctness = Relates to potential sequences of execution
defined by the model.
 ex. A case should never be able to reach a deadlock or a livelock .
 Soundness = Common sound an unsound process fragments: See
the picture above.

Semantic Quality and Validation

 Semantic quality = Goal of producing models that make true statements. about
the considered domain, either for existing as-is processes or future to-be
processes.
o The model has to be compared with the real-world domain of a
particular business process.
o There are no formal rules to check semantic quality, it has to pass by
the people involved in the process.
 Validation = Checking the semantic quality of a model by comparing it with the
real-world business process.
o Validity = All statements included in the model are correct and relevant
to the problem.
 A domain expert should point out any difference between what
the model states and what is possible in reality.
 Can be checked by simulations and interviews.
o Completeness = The model contains all relevant statements on a
process that would be correct.
 Check if all the possible alternative processing options have
been accounted for.

Pragmatic Quality and Certification

 Pragmatic Quality = Building a process model of good usability.
o Predict how the model is going to be used.
o Can be checked by how well a user understands the model.

 Certification = Checking the pragmatic quality of a process model by investigating

its usage.
o Aspects of usability:
1. Understandability = How easy it is to read a specific process
model.
  Size of the model, structural complexity, graphical layout.
2. Maintainability = The ease of applying changes to a process
model.
3. Learning = Relates to the degree of how good a process model
reveals how a business process works in reality.
o Certification can be conducted using user interviews or user
experiments.
o Improve usability by providing usability by applying design rules on the
structure of the model.
o Checks for Understandability, Maintainability and Learning:
i. Consistency between the visual structure and logical structure.
ii. Using meaningful labels.

Modeling Guidelines and Conventions

 Modeling Guidelines and Conventions are an important tool to make sure the
models are consistent and keep their integrity when multiple people are involved.
 Increase readability and compatibility.
 Using naming conventions, recommends the use of verb-object style for labeling
of activities etc.
 Restrictions, for simplifying the set of elements used in BPMN.

 THE SEVEN MODELING GUIDELINES:

1. Use as few elements in the model as possible. (Less elements == easier
to understand)
2. Minimize the routing paths per element. (Less paths == less chance to
get errors)
3. Use one start and one end event. (Less events == easier to understand)
4. Model as structured as possible. (Better structured == easier to
understand, less errors)
5. Avoid OR-gateways. (No OR-gateways = easier to understand)
6. Use verb-object activity labels. ( Less ambiguous == easier to
understand)
7. Decompose a model with more than 30 elements. (Less elements ==
easier to understand)
Exercise 5.10 Consider the process model of Fig. 5.8. Explain which 7PMG guidelines
point to potential for improvement. Remodel the process based on your observations.

Solutions to Exercises
Solution 5.1 In case you are supposed to map the process of signing a rental contract
in your city, it is likely that you have some experience with this process, either from
renting a flat yourself, or from stories from your friends, or from you or your friends
giving a flat for rent. Assuming you have already studied the chapters on process
modeling, you have both domain expertise and process modeling expertise. This is an
uncommon situation. Most often, you face situations like mapping the process of getting
a license plate for your car in Liechtenstein as a foreign resident. This is a process for
which you would unlikely have domain knowledge. Process discovery typically brings
you as a process analyst into an environment that you do not know in detail beforehand.
Process discovery is concerned with understanding the process under consideration
and also the domain surrounding it.

Solution 5.2 An advantage of having teams modeling processes themselves is first that
a lot of process models can be created in a short span of time. It is critical though that
these teams possess the required skills in process modeling. According to the third
challenge of process discovery, domain experts typically do not have process modeling
skills and feel uncomfortable with the modeling task. Furthermore, domain experts often
think in cases (second challenge) and lack the process perspective to generalize.
Finally, there is the risk that the results from such a modeling initiative might be
fragmented and difficult to integrate. It is typically the responsibility of the process
analyst to integrate the fragmented perspectives.
Solution 5.3 Domain knowledge can be very helpful for analyzing processes. It helps to
ask the right questions and to build analogies from prior experience. On the other hand,
the skills of an experienced process analyst should not be underestimated. These skills
are domain-independent and relate to how a process discovery project can be
organized. Experienced process analysts are typically very skilled in scoping and driving
a project into the right direction. They possess problem-solving skills for handling
various critical situations of a process discovery project. There is clearly a trade-off
between the two sets of skills. It should be ensured that a certain level of process
analysis experience is available. If that is not the case for the applying domain expert,
the process analyst might be preferred.
Solution 5.4 As a customer, we would have to rely mostly on evidence-based process
discovery. We can place exemplary orders and study the different processing options
for them. In relation to these placed orders, we could also contact the customer help
desk and inquire details of the process that we cannot directly observe. If we were
assigned to a process discovery project by the process owner, we would get access to
the domain experts within the company. In this case, we could also use interviews and
workshop-based discovery methods.
Solution 5.5 This process contains ten major activities that are executed by different
persons. We can assume that there will be a kickoff meeting with the process owner
and some important domain experts on day one. One day might be required to study
available documentation. An interview with one domain expert can take from two to
three hours, such that we would be able to meet two persons per day, and document
the interview results at night time. Let us assume that we meet some persons only once
while we seek feedback from important domain experts in two additional interviews.
Then, there would be a final approval from the process owner. This adds up to one day
for the kickoff, one for document study, five days for the first iteration interviews, and
further five days if we assume that we meet five experts three times. Then, we need one
day for preparing the meeting for final approval with the process owner, which would be
on the following day. If there are no delays and scheduling problems, this yields 2 + 5 +
5 + 2 = 14 work days as a minimum.

Solution 5.6 Before starting with process discovery, it is important to understand the
culture and the sentiment of an organization. There are companies that preach and
practice an open culture in which all employee are encouraged to utter their ideas and
their criticism. Such organizations can benefit a lot from workshops as participants are
likely to present their ideas freely. In strictly hierarchical organizations, it is necessary to
take special care that every participant gets an equal share of parole in a workshop and
that ideas and critique are not hold back. It might be the case that the young dynamic
company has a more open culture than the company with extensive health and security
regulations. This has to be taken into account when organizing a workshop.
Solution 5.7 There are various circumstances that may restrict the application of
different discovery methods. Direct observation may not be possible if the process
partially runs in a remote or dangerous environment. For instance, the discovery of a
process of an oil-producing company for pumping oil from an oil rig to a ship might
belong to this category. Then, there might be cases where documentation does not
exist, for example when a startup company, which has gone through a period of rapid
growth wants to structure its purchasing process. Lack of input may also be a problem
for automatic process discovery based on event log data. If the process under
consideration is not yet supported by information systems, then there are no data
available for conducting automatic process discovery. In general, interviews are always
possible. It might still be a problem though to gain commitment of domain experts for an
interview. This is typically the case when the process discovery project is subject to
company-internal politics and hidden agendas. Workshop-based discovery can be
critical in companies with strong hierarchy which have a culture of suppressing creative
thinking of their staff.
Solution 5.8 The type of the gateway has to be consistent with the conditions of the
subsequent arcs. If there is an XOR-split, then the conditions on the arcs have to be
mutually exclusive. If there is an OR-split, then the conditions can be non-exclusive. If
an AND-split is used, there should be no conditions on the arcs.
Solution 5.9 Four unsound fragments are shown with the following problems:
• The lack of synchronization relates to an AND-split followed by an XOR-join. In this
case, the two tokens created from the AND-split are not synchronized XOR-join,
potentially leading to the duplicate execution of activities downstream.
• A deadlock occurs, for instance, if an XOR-split is followed by an AND-join. As the
XOR-split creates a token only on one of its outgoing arcs, the AND-join requiring a
token on each of its incoming arcs gets stuck waiting for a second token to arrive.
• In case there is an OR-split followed by an XOR-join, we potentially get a lack of
synchronization. This depends upon the conditions of the OR-split. If only one token is
generated from it, the process can proceed correctly. If multiple tokens are generated,
there is a lack of synchronization. In the same vein, there is a potential deadlock if the
OR-split is followed by an AND-join.
• A livelock can occur in an inappropriate loop structure. Here, there is an XOR- join
used as an entry to a loop, but the loop exit is modeled with an AND-split. This has the
consequence that it is never possible to leave the loop. Each time the AND-split is
reached, it creates one token exiting the loop, but also another token that stays within
the loop.
Solution 5.10 The process model reveals several problems. Several elements with the
same name are shown twice (end event and archiving activity), therefore G1 is violated.
Also the control structure is very complicated, violating G4 asking for a

Fig. 5.9 The complaint handling process reworked

Fig. 5.10 A loan application process structured model.
Finally, several activities do not follow the naming conventions of G6.The model can be
reworked and simplified to the one shown in Fig. 5.9.

DEFINITION OF TERMS: