A Practical Guide to Business Process Management & Process Mining for Saudi Companies

We find that business process management (BPM) and process mining are usually explained too academically without explaining why they’re useful in practice and more importantly how to apply them in practice!

In this blog post, we motivate the needs for business processes and BPM and explain how process mining naturally emerges in this text. Finally, we give you some tool recommendations.

IMPORTANT:

This post was written in five different versions for different industries

Oil & Gas / Manufacturing: Aramco, Ma’aden, Alfanar…

Petrochemicals: SABIC, Petro Rabigh, Tasnee…

Electricity & Utilities: Saudi Electricity Company, ACWA, Marafiq

Telecommunications: stc, Mobily, Zain

Banking: SNB, Al Rajhi, Riyad Bank

All Industries: for general process excellence representatives

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making a profit. Why do you then need business processes to make a profit? A business maximizes its profit by providing value to its customers. Once an oil and gas company refines crude oil into fuel or other products multiple times, it learns which practices and activities led to the best quality and highest yield. If an oil and gas company always refines crude oil in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always refining crude oil with that same excellence. To ensure the company delivers every barrel of refined oil with the highest quality, it has to document the practices that led to the excellent product. The documentation of these best practices that led to the best-refined products is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure refinery operations are always conducted in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods and specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always refine crude oil in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Conduct refining operations enough times.
  2. Monitor the past refining operations.
  3. Pick out the past refining runs that were the best.
  4. Examine which steps occurred before the best refining runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best refining runs.
  6. Conduct all future refining runs as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep refining oil with perfect quality? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep refining oil as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring the product is always refined in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that all crude oil with a sulfur content > 0.5% needs additional processing. But the circumstances changed in that the company now has new desulfurization technology that makes additional processing unnecessary. The more optimal approach for the company would be to update the process to skip that step for crude oil with sulfur content < 1%.

The second problem is that the business might have not conducted enough refining runs to actually get a good picture of what the best-refined products look like. Consider if we only refined 100 batches of crude oil before writing the step-for-step guide. And in the 101st batch, we discover that pre-heating the crude oil in a certain way results in higher yields. We’d discover a better step-for-step guide for future refining operations.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we conducted enough refining runs. So after defining the should-be process model, we start over again and go through all steps again.

  1. Conduct refining operations enough times.
  2. Monitor the past refining operations.
  3. Pick out the past refining runs that were the best.
  4. Examine which steps/activities occurred before the best refining runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best refining runs.
  6. Conduct refining operations enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Conduct refining operations enough times

    For actually conducting the refining runs, we don’t need any additional requirements or tools.

  2. Monitor the past refining operations

    To monitor the past refining runs, we need to have the outcomes and the steps/activities of previous runs documented. For example, we can have every step of all previous batches documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like a Distributed Control System (DCS) or an ERP system.

    ⇒ Requirement 1: IT system

  3. Pick out the past refining runs that were the best

    To pick out which refining runs were the best, we need to firstly have some definition of what the best looks like. For refining runs, the best could be those with the highest yield and lowest energy consumption.

    ⇒ Requirement 2: Attributes that define best-refined product

    Secondly, we need to have some measurement of the attributes that make the best. So for refining runs, we need to have some systems that capture yield and energy usage. Again, we need an IT system.

  4. Examine which steps/activities occurred before the best refining runs

    After perhaps choosing the top 1% of refining runs, we want to find out which steps/activities led them to being the top 1% of runs. Again, we need an IT system that captures the individual steps/activities of the refining process. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best refining runs

    If we are examining the top 1% of refining runs and are trying to derive a step-for-step guide (should-be process model) that led to the best runs, we might have to examine maybe 1000 runs or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top runs and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the best-refined products. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the top refining runs and their steps and discover a should-be process model that encapsulates the steps that led to the best runs.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Conduct refining operations enough times

    Now once we have a should-be process model, the cycle starts again. Starting from this point, we want to conduct every refining run exactly as dictated by the should-be process model. But how can we know that we are actually following the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific refining run conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

DCS: Honeywell, Siemens, ABB,…

ERP systems: SAP, Oracle,…

Requirement 2: Attributes that define the best-refined product

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making a profit. Why do you then need business processes to make a profit? A business maximizes its profit by providing value to its customers. Once a petrochemical company produces polymers or chemical intermediates multiple times, it learns which practices and activities led to the highest yield, lowest waste, and best energy efficiency. If a petrochemical company always produces chemicals in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always producing its chemicals with that same excellence. To ensure the company consistently achieves the highest production quality and efficiency, it has to document the practices that led to the excellent production. The documentation of these best practices that led to the best production is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure production operations are always conducted in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods and specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always produce chemicals in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Conduct production runs enough times.
  2. Monitor the past production runs.
  3. Pick out the past production runs that were the best.
  4. Examine which steps occurred before the best production runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best production runs.
  6. Conduct all future production runs as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep producing chemicals with perfect quality? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep producing chemicals as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring the chemicals are always produced in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that all reactors must run at a certain temperature to avoid yield loss. But circumstances changed, and the feedstock now allows for a slightly higher temperature that reduces energy consumption without affecting the yield. The more optimal approach for the company would be to update the process to use this new temperature setting.

The second problem is that the business might have not conducted enough production runs to actually get a good picture of what the best runs look like. Consider if we only produced 100 batches before writing the step-for-step guide. And in the 101st batch, we discover that skipping a certain intermediate drying step improves throughput. We’d discover a better step-for-step guide for future production runs.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we conducted enough production runs. So after defining the should-be process model, we start over again and go through all steps again.

  1. Conduct production runs enough times.
  2. Monitor the past production runs.
  3. Pick out the past production runs that were the best.
  4. Examine which steps/activities occurred before the best production runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best production runs.
  6. Conduct production runs enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Conduct production runs enough times

    For actually conducting the production runs, we don’t need any additional requirements or tools.

  2. Monitor the past production runs

    To monitor the past production runs, we need to have the outcomes and the steps/activities of previous runs documented. For example, we can have every step of all previous batches documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like an ERP system or a Manufacturing Execution System (MES).

    ⇒ Requirement 1: IT system

  3. Pick out the past production runs that were the best

    To pick out which production runs were the best, we need to firstly have some definition of what the best looks like. For production runs, the best could be those with the highest yield and lowest energy consumption.

    ⇒ Requirement 2: Attributes that define best production run

    Secondly, we need to have some measurement of the attributes that make the best. So for production runs, we need to have some systems that capture yield, energy usage, and throughput rates. Again, we need an IT system.

  4. Examine which steps/activities occurred before the best production runs

    After perhaps choosing the top 1% of production runs, we want to find out which steps/activities led them to being the top 1% of runs. Again, we need an IT system that captures the individual steps/activities of the production process. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best production runs

    If we are examining the top 1% of production runs and are trying to derive a step-for-step guide (should-be process model) that led to the best runs, we might have to examine maybe 1000 runs or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top runs and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the best production runs. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the top production runs and their steps and discover a should-be process model that encapsulates the steps that led to the best runs.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Conduct production runs enough times

    Now once we have a should-be process model, the cycle starts again. Starting from this point, we want to conduct every production run exactly as dictated by the should-be process model. But how can we know that we are actually following the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific production run conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

MES systems: Honeywell, Siemens, Emerson,…

ERP systems: SAP, Oracle,…

Requirement 2: Attributes that define the best production run

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making a profit. Why do you then need business processes to make a profit? A business maximizes its profit by providing value to its customers. Once a utility company distributes electricity to customers multiple times, it learns which practices and activities led to the most reliable delivery and the lowest number of outages. If a utility company always delivers electricity in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always delivering electricity with that same excellence. To ensure the company delivers electricity with high reliability and minimal disruptions, it has to document the practices that led to the excellent service. The documentation of these best practices that led to the best electricity distribution is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure electricity distribution operations are always conducted in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods and specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always distribute electricity in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Deliver electricity enough times.
  2. Monitor the past electricity distribution runs.
  3. Pick out the past distribution runs that were the most reliable.
  4. Examine which steps occurred before the most reliable distribution runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the most reliable distribution runs.
  6. Deliver all future electricity distributions as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep distributing electricity with perfect reliability? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep distributing electricity as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring electricity is always distributed in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that all transformer inspections were carried out every 6 months. But the circumstances changed, and now sensors provide real-time monitoring of transformer conditions, allowing for condition-based maintenance rather than a fixed schedule. The more optimal approach for the company would be to update the process to focus on real-time data instead of periodic inspections.

The second problem is that the business might have not conducted enough distribution cycles to actually get a good picture of what the best-distributed electricity runs look like. Consider if we only distributed electricity in a particular grid for 100 days before writing the step-for-step guide. And on the 101st day, we discover that a certain route optimization reduces power losses during peak hours. We’d discover a better step-for-step guide for future electricity distribution runs.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we distributed electricity enough times. So after defining the should-be process model, we start over again and go through all steps again.

  1. Deliver electricity enough times.
  2. Monitor the past electricity distribution runs.
  3. Pick out the past distribution runs that were the most reliable.
  4. Examine which steps/activities occurred before the most reliable distribution runs.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the most reliable distribution runs.
  6. Deliver electricity enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Deliver electricity enough times

    For actually delivering electricity, we don’t need any additional requirements or tools.

  2. Monitor the past electricity distribution runs

    To monitor past electricity distributions, we need to have the outcomes and the steps/activities of previous distributions documented. For example, we can have every step of all previous distributions documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like a Supervisory Control and Data Acquisition (SCADA) system or a Geographic Information System (GIS).

    ⇒ Requirement 1: IT system

  3. Pick out the past electricity distribution runs that were the most reliable

    To pick out which electricity distributions were the most reliable, we need to firstly have some definition of what reliability looks like. For electricity distribution, reliability could mean the lowest downtime and the fewest unplanned outages.

    ⇒ Requirement 2: Attributes that define reliable electricity distribution

    Secondly, we need to have some measurement of the attributes that make the distribution reliable. So for electricity distribution, we need to have systems that capture downtime, outages, and power loss. Again, we need an IT system.

  4. Examine which steps/activities occurred before the most reliable distribution runs

    After perhaps choosing the top 1% of distribution runs, we want to find out which steps/activities led them to being the top 1% of runs. Again, we need an IT system that captures the individual steps/activities of the electricity distribution process. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the most reliable distribution runs

    If we are examining the top 1% of electricity distribution runs and are trying to derive a step-for-step guide (should-be process model) that led to the most reliable distributions, we might have to examine maybe 1000 runs or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top runs and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the most reliable distributions. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the most reliable distribution runs and their steps and discover a should-be process model that encapsulates the steps that led to the best distributions.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Deliver electricity enough times

    Now once we have a should-be process model, the cycle starts again. Starting from this point, we want to deliver every distribution run exactly as dictated by the should-be process model. But how can we know that we are actually following the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific electricity distribution run conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

SCADA systems: GE, ABB, Siemens,…

GIS systems: Esri, Smallworld…

Requirement 2: Attributes that define reliable electricity distribution

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making a profit. Why do you then need business processes to make a profit? A business maximizes its profit by providing value to its customers. Once a telecommunications company sets up and manages internet and mobile services for customers multiple times, it learns which practices and activities led to the fastest installations, lowest downtime, and highest customer satisfaction. If a telecommunications company always provides services in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always delivering these services with that same excellence. To ensure the company delivers its services with consistent reliability and efficiency, it has to document the practices that led to the excellent service. The documentation of these best practices that led to the best installations and operations is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure service delivery operations, such as internet installations and network maintenance, are always conducted in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods and specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always provide and maintain telecommunication services in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Provide telecommunication services enough times.
  2. Monitor the past service installations and operations.
  3. Pick out the past service deliveries that were the most efficient and reliable.
  4. Examine which steps occurred before the best service deliveries.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best service deliveries.
  6. Provide all future services as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep providing services with perfect reliability? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep delivering services as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring telecommunication services are always delivered in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that all new fiber installations required on-site verification from a technician. But the circumstances changed, and new remote monitoring technology allows the verification to be done remotely, saving time and costs. The more optimal approach for the company would be to update the process to skip the on-site step when applicable.

The second problem is that the business might have not delivered enough services to actually get a good picture of what the best service deliveries look like. Consider if we only delivered 100 new installations before writing the step-for-step guide. And in the 101st installation, we discover that automating the customer activation step through a new API integration reduces setup time significantly. We’d discover a better step-for-step guide for future service deliveries.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we delivered services enough times. So after defining the should-be process model, we start over again and go through all steps again.

  1. Provide telecommunication services enough times.
  2. Monitor the past service installations and operations.
  3. Pick out the past service deliveries that were the most efficient and reliable.
  4. Examine which steps/activities occurred before the best service deliveries.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best service deliveries.
  6. Provide telecommunication services enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Provide telecommunication services enough times

    For actually delivering telecommunication services, we don’t need any additional requirements or tools.

  2. Monitor the past service installations and operations

    To monitor the past service deliveries, we need to have the outcomes and the steps/activities of previous deliveries documented. For example, we can have every step of all previous installations documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like an Operational Support System (OSS) or a Customer Relationship Management (CRM) system.

    ⇒ Requirement 1: IT system

  3. Pick out the past service deliveries that were the most efficient and reliable

    To pick out which service deliveries were the most efficient and reliable, we need to firstly have some definition of what the best looks like. For service deliveries, the best could mean the lowest installation time, fewest customer complaints, and highest uptime.

    ⇒ Requirement 2: Attributes that define best service delivery

    Secondly, we need to have some measurement of the attributes that make the service delivery efficient and reliable. So for service deliveries, we need to have systems that capture installation times, customer satisfaction scores, and downtime metrics. Again, we need an IT system.

  4. Examine which steps/activities occurred before the best service deliveries

    After perhaps choosing the top 1% of service deliveries, we want to find out which steps/activities led them to being the top 1% of deliveries. Again, we need an IT system that captures the individual steps/activities of the service delivery process. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best service deliveries

    If we are examining the top 1% of service deliveries and are trying to derive a step-for-step guide (should-be process model) that led to the most efficient and reliable services, we might have to examine maybe 1000 deliveries or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top deliveries and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the best service deliveries. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the top service deliveries and their steps and discover a should-be process model that encapsulates the steps that led to the best deliveries.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Provide telecommunication services enough times

    Now once we have a should-be process model, the cycle starts again. Starting from this point, we want to deliver every service exactly as dictated by the should-be process model. But how can we know that we are actually following the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific service delivery conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

OSS systems: Amdocs, Netcracker,…

CRM systems: Salesforce, HubSpot,…

Requirement 2: Attributes that define best service delivery

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making a profit. Why do you then need business processes to make a profit? A business maximizes its profit by providing value to its customers. Once a bank processes loan applications multiple times, it learns which practices and activities led to the fastest approvals, lowest risk, and highest customer satisfaction. If a bank always processes loan applications in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always processing loans with that same excellence. To ensure the bank consistently processes loan applications with optimal speed and quality, it has to document the practices that led to the excellent outcomes. The documentation of these best practices that led to the best loan processing is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure loan processing is always conducted in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods and specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always process loan applications in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Process loan applications enough times.
  2. Monitor the past loan processing activities.
  3. Pick out the past loan applications that were processed the most efficiently and accurately.
  4. Examine which steps occurred before the best loan processing cases.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best loan processing cases.
  6. Process all future loan applications as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep processing loan applications with perfect efficiency and accuracy? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep processing loan applications as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring loans are always processed in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that every loan application above $50,000 required manual underwriting review. But the circumstances changed, and the bank has now implemented a new credit scoring system that allows applications below $100,000 to be processed automatically. The more optimal approach for the company would be to update the process to reflect this new automation capability.

The second problem is that the business might have not processed enough loan applications to actually get a good picture of what the best loan processing cases look like. Consider if we only processed 100 loan applications before writing the step-for-step guide. And in the 101st case, we discover that automating document collection through a customer portal reduces turnaround time significantly. We’d discover a better step-for-step guide for future loan processing.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we processed enough loans. So after defining the should-be process model, we start over again and go through all steps again.

  1. Process loan applications enough times.
  2. Monitor the past loan processing activities.
  3. Pick out the past loan applications that were processed the most efficiently and accurately.
  4. Examine which steps/activities occurred before the best loan processing cases.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best loan processing cases.
  6. Process loan applications enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Process loan applications enough times

    For actually processing loan applications, we don’t need any additional requirements or tools.

  2. Monitor the past loan processing activities

    To monitor the past loan applications, we need to have the outcomes and the steps/activities of previous applications documented. For example, we can have every step of all previous applications documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like a Loan Origination System (LOS) or a Customer Relationship Management (CRM) system.

    ⇒ Requirement 1: IT system

  3. Pick out the past loan applications that were processed the most efficiently and accurately

    To pick out which loan applications were processed the most efficiently and accurately, we need to firstly have some definition of what the best looks like. For loan processing, the best could mean the shortest processing time, fewest errors, and highest approval accuracy.

    ⇒ Requirement 2: Attributes that define best loan processing

    Secondly, we need to have some measurement of the attributes that make the loan processing efficient and accurate. So for loan processing, we need to have systems that capture turnaround time, error rates, and approval accuracy. Again, we need an IT system.

  4. Examine which steps/activities occurred before the best loan processing cases

    After perhaps choosing the top 1% of loan processing cases, we want to find out which steps/activities led them to being the top 1% of cases. Again, we need an IT system that captures the individual steps/activities of the loan processing process. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best loan processing cases

    If we are examining the top 1% of loan processing cases and are trying to derive a step-for-step guide (should-be process model) that led to the most efficient and accurate processing, we might have to examine maybe 1000 cases or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top cases and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the best loan processing cases. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the top loan processing cases and their steps and discover a should-be process model that encapsulates the steps that led to the best cases.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Process loan applications enough times

    Now once we have a should-be process model, the cycle starts again. Starting from this point, we want to process every loan application exactly as dictated by the should-be process model. But how can we know that we are actually following the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific loan application conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

LOS systems: FIS, Ellie Mae, Blend,…

CRM systems: Salesforce, HubSpot,…

Requirement 2: Attributes that define best loan processing

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Edit Content

Why Even Complicate Our Lives with Business Processes?

What is a business? Any enterprise or organization with the purpose of making profit. Why do you then need business processes to make profit? A business maximizes its profit by providing value to its customers. Once a business has delivered some service or product multiple times, it learns which practices and activities led to the best service and best products. The best services and products are the ones that deliver the most customer value. If a business always delivers every product/service in the best way, it will generate more value for the customer, i.e., more profit. The company can increase its profit by always delivering the product/service with that same excellence. To ensure the company delivers its products or services with excellence every time, it has to document the practices that led to the excellent product or service. The documentation of these best practices that led to the best product or service is exactly what a business process is.

A business process is a sequence of steps/activities that aims to fulfill a certain business objective. We now motivated the need for business processes.

Business Process Management

Business Process Management (BPM) is a set of methods to ensure products/services are always delivered in a way that adheres to best practices. Based on this definition, we’ll motivate the need for the methods & specific tools to fulfill that goal of business process management.

Starting with the goal. The goal is to always deliver our product/service in the best way as documented by the business process model. But how do we do that?

We do the following:

  1. Deliver product/service enough times.
  2. Monitor the past deliveries of product/service.
  3. Pick out the past deliveries that were the best.
  4. Examine which steps occurred before the best deliveries.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best deliveries.
  6. Deliver all future product/service as dictated by the step-for-step guide (should-be process model).

But wait. Are we done? Do we till the end of time keep delivering the product/service with perfect quality? Well sadly, it does not work that way.

Iterative Business Process Management

In an ideal world, we would just keep delivering the product/service as dictated by the step-for-step guide till the end of time. We’d follow the so-called “happy-path” every time. We have two problems with the current way of ensuring the product/service is always delivered in the best way.

The first problem is that circumstances change. Perhaps the step-for-step guide (should-be process model) dictated before that all purchase orders > 50,000 $ have to be approved by the procurement team. But the circumstances changed in that the company has too many purchase orders between 50,000 $ and 100,000 $ that it takes up too much time to approve them manually. The more optimal approach for the company would be to change the manual approval to all purchase orders > 100,000 $.

The second problem is that the business might have not delivered product/service enough times to actually get a good picture of what the best delivered products/services look like. Consider if we only delivered 100 purchase orders, before writing the step-for-step guide. And in the 101st order we discover that matching the invoices is not necessary, because it is done automatically. We’d discover a better step-for-step guide for future purchase orders.

This motivates the need for the iterative approach. In this approach, we never settle on one should-be process model till the end of time. We only keep a should-be process model for a specific period of time and then we re-evaluate it once we delivered the product/service enough times. So after defining the should-be process model, we start over again and go through all steps again.

  1. Deliver product/service enough times.
  2. Monitor the past deliveries of product/service.
  3. Pick out the past deliveries that were the best.
  4. Examine which steps/activities occurred before the best deliveries.
  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best deliveries.
  6. Deliver product/service enough times.

In this iterative business process management, we keep improving upon our should-be process model to always ensure we are doing the things the best way we could. But how do we actually do this in practice?

Note: In classical BPM literature, these steps are referred to as Design ⇒ Model ⇒ Execute ⇒ Monitor ⇒ Optimize ⇒ Design. But the idea is essentially the same.

Requirements & Methods

Let’s examine our lifecycle again to find what systems & methods are required to do this in practice.

  1. Deliver product/service enough times

    For actually delivering the product/service, we don’t need any additional requirements or tools.

  2. Monitor the past deliveries of product/service

    To monitor the past deliveries of product/service, we need to have the outcomes and the steps/activities of previous deliveries documented. For example, we can have every step of all previous orders documented on paper, but that would be impractical. The way this is done nowadays is using an IT system like an ERP system.

    ⇒ Requirement 1: IT system

  3. Pick out the past deliveries that were the best

    To pick out which product/service deliveries were the best, we need to firstly have some definition of what the best looks like. For purchase orders the best could be the purchase orders with the least processing time & full automation.

    ⇒ Requirement 2: Attributes that define best product/service

    Secondly, we need to have some measurement of the attributes that make the best. So for purchase orders, we need to have some systems that capture the processing time and the automation rate. So again we need an IT system.

  4. Examine which steps/activities occurred before the best deliveries

    After perhaps choosing the top 1% of deliveries, we want to find out which steps/activities led them to being the top 1% of deliveries. Again, we need an IT system that captures the individual delivery steps/activities of the product/service. We have that requirement already fulfilled through step 2.

  5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best deliveries

    If we are examining the top 1% of deliveries and are trying to derive a step-for-step guide (should-be process model) that led to the deliveries. We might have to examine maybe 1000 deliveries or usually way more. It is not feasible or accurate for a human to examine each of these 1000 top deliveries and the steps that preceded them. As such, we need an automatic way of deriving the steps/activities that led to the best deliveries. This is exactly where process discovery comes in. Prof. Wil van der Aalst invented process discovery as one of the 3 key disciplines of process mining to solve precisely that problem. Process discovery algorithms take in, for instance, 1000 of the top deliveries and their steps and discover a should-be process model that encapsulates the step that led to the best deliveries.

    ⇒ Requirement 3: Process Discovery algorithm

    Now, of course, we can store this should-be process model on paper, but it is simply impractical. Nowadays, we use process modeling software, which allows us to make changes to the should-be process model and allows for easier distribution.

    ⇒ Requirement 4: Process modeling software

  6. Deliver product/service enough times

    Now once, we have a should-be process model. The cycle starts again. Starting from this point, we want to deliver every product/service exactly as dictated by the should-be process model. But how can we know that we are actually delivering as dictated by the should-be process model? This is where conformance checking comes into play. Conformance checking is the second pillar of process mining. It checks if a specific delivery of a product/service conforms to the should-be process model. So conformance checking is used to check if we are actually following the step-for-step guide given by the should-be process model.

    ⇒ Requirement 5: Conformance Checking algorithm

    And once, we have a should-be process model, automation becomes a very natural thing to think about. Having a should-be process model makes automation much easier to apply because you know exactly which steps to execute, namely the steps of the should-be process model. It is much more likely to follow the should-be process model if the steps are automated. Simply, because humans make errors 🙂

    ⇒ Requirement 6: Business process automation software

Tools

Requirement 1: IT system

ERP systems: SAP, Oracle, Microsoft Dynamics, Sage,…

Databases: Oracle, MySQL, Microsoft SQL Server, PostgreSQL,…

Requirement 2: Attributes that define best product/service

No tools 🙂

Requirement 3: Process Discovery

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 4: Process Modeling

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath, Interfacing

Requirement 5: Conformance Checking

Free tools: ProM, PM4Py

Commercial tools: Celonis, ARIS, SAP Signavio, UiPath

Requirement 6: Business Process Automation

Commercial tools: Make (now acquired by Celonis), Zapier, Celonis, ARIS, SAP Signavio, UiPath, Appian…

Summary

  1. Businesses exist to make profit
  2. Profit is maximized when delivery of product / service is perfect every time
  3. Business processes are required to ensure perfect delivery of product / service every time
  4. Business Process Management is a discipline that ensures perfect delivery of product / service every time
  5. The Business Process Management Lifecycle goes as follows
    1. Deliver product / service enough times
    2. Monitor past deliveries of product / service
    3. Pick out the past deliveries that were the best
    4. Examine which steps / activities occurred before the best deliveries
    5. Derive a should-be step-for-step guide (should-be process model) that encapsulates the steps that led to the best deliveries
    6. Start over with a. again
  6. Process mining is a subset of business process management that is used in steps d. and e. of the BPM lifecycle
  7. Kind of tools needed
    1. IT System
    2. Process Discovery tools
    3. Process Modeling tools
    4. Conformance Checking tools
    5. Business Process Automation tools