By Gurdeep Singh, Automation CoE Lead at Tryg
Process Mining is growing to become an established discipline in the IT landscape of large and complex organisations.
IT analysts are also increasingly paying attention to both the innovation happening in the space and the vendors. Still, it can be a confusing market to look at and as usual, if you want to reap the benefits, you need to do your homework.
In brief: Process mining is a scientific tool with capabilities of process blueprinting, process optimization and data visualization with the help of machine learning engines. Typically, a process mining tool visualizes data in much more details then conventional visualization tools. The information captured is used to build a storyline highlighting the journey of a transaction on a process and time flow.
In this article, I’ll share my take on process mining in 2023 based on my experience, which started with a decade of RPA work and now with 5+ years focusing on processes at Danish insurance firm Tryg.
Let’s start at the beginning.
Getting to the real story on process mining
While adopting Process Mining in an organisation one of the biggest questions for procurement, process excellence, enterprise architecture, and CoE teams is which vendor to select and what are the parameters on which you can measure these vendors.
As you probably know, if you talk to most of the sales teams from the vendors in the market they will claim to be the best or in the top right quadrant of one or the other research or analyst teams. These lists can be quite a good starting point to narrow down your search list since it is good to have a smaller and well competitive group to have discussion and analysis with.
Having said that, much depends on the maturity, current architecture and requirements from the organisation which should drive the assessment and the assessment factors. The pointers which I have highlighted below are just driving principles and readers can fine tune or adjust them according to their organisational needs.
Setting the stage for process mining vendor selection
The parameters can be broadly categorized in Technical and Functional parameters. Here’s a hopefully helpful list to get you started:
a. Event Collection
a. Integrated Connectors: The ability of a process mining system to connect to other systems, applications and databases based on events or predefined rule. Out of the box connectors with major systems available by developers and business resources. Incase out of the box connectors are not available an ability to create custom build connectors is possible. For eg: the system has open connectors which can be leveraged for event based data collection or can be customized for the same. Eg: of out of box connector can be connectors for Guidewire, SNOW, Microsoft applications, Salesforce SAP etc.
b. Multisystem Integration: Availability of independent API, Services which can be leveraged for connectivity to multiple systems for data gathering. These services should be integrated but independent of each other and can be leveraged based on a scheduled, trigger points, event, scenario etc. Ability to monitor the same with a user friendly dashboard to manage and maintain connectivity is always an added advantage.
c. Real Time Connection: Ability to leverage the services, API and connectors to consume the information real-time. Since process mining is highly event based it should be a possibility to not only connect but consume real time data and batch processing being interconnected but independent at the same time.
b. Data Hosting:
a. Cloud / OnPrem / Hybrid: Depending upon the organizations cloud, application & data management strategy the application servers and data servers / data hosting flexibility. There is no one is better than the other it is highly governed by the organization however I personally feel the hybrid setup flexibility is preferred considering data handling and governance restrictions in most of the organizations.
b. Data handling – Retention, deletion & Utility: Its important that the data handling policy by the vendor are clearly defined incase we are using a SaaS solution on cloud. Who has the access to the data, how is it stored, how long is it stored what is the data deletion policy, what is the data during the period is used for, how the analysis created by organization leveraged by the software organization, how open / flexible is it to the regular changing / upgrading compliance requirements. Who has the ownership and how open an flexible is it for business / data owners to change or govern it.
c. Data Anonymization: Ability to anonymize data based on rules predefined in the system like PII.
c. Process Discovery and Analysis
a. Data Modeling Capabilities: Since the data connectors can provide data from various sources and the datasets can be interconnected, interdependent, exclusive, mutually exclusive, unique, discrete etc. its important that we have the capability to model various data sources available in tables and columns in create data models. Ability to model data together to create a new data set is an important factor since non availability of the same can resulting in dependence on other data bases for models and conjunction of the same on a mid-server.
b. Out of the box process visualization: Pre defined templates for major business processes or visualizations helping in development as guiding post for business users.
c. Flexibility to connect to external visualization templates: Flexibility to connect to visualization templates predefined by business in Power BI / Tableau / Excels etc.
d. Custom Dashboards: Ability to create custom dashboard apart from out of the box templates available
e. Conformance Dashboard: Ability to create standalone conformance dashboards and process, timelines, relationship and automation dashboards
f. Industry Benchmarking: Ability to plugin or out of the box industry benchmarking for major processes and systems like Claims, sales, U/W, P2P etc. on SAP, Guidewire, ServiceNow etc.
g. Predictive RCA: Ability to trigger RCA based on predefined rules or send alerts based on predefined rules.
h. Save & Share RCA snapshots: Ability to save and share RCA snapshots with business in PDF, image or analysis in tool
i. Simulation Analysis: Ability to create simulation based on synthetic data and compare with actual data to create and save simulation results
d. Improvements and enhancements:
a. Triggered based action: Ability to trigger actions based on triggers defined by business or predefined templates
b. API / Microservices / Connectors for action: Ability to connect with various systems for actions including chatbots, robots, workflows, emails, process services and APIs
c. Next Best Action: Ability to suggest next best action based on predefined rules in business for trigger points like notifications, trigger emails, set reminders etc.
d. Mobile readiness: Ability to view dashboards, trigger points, queue for manual handling etc. be mobile ready along with desktop version
e. Additional considerations:
a. Security and compliance certifications: ISO, GDPR COPC etc. certifications by the vendors
b. Licensing model: Utility based licensing, user based licensing, server based licensing, processing based licensing are different models a lot depends on the consumption criteria’s of an organization
c. Training and Community support : Ability of a vendor to help in the training and development of an adopting organization with online or in person training provided.
Who are the major process mining vendors to consider?
The major players in the market currently are Celonis, UiPath, Paf and Minit.
All of them are at par when it comes to user-friendliness, visualization and resourcefulness however there are few key differences on grounds of:
a. On prem vs. Cloud solution: Celonis is 100% cloud where as UiPath provides both cloud and On-Prem solution – to be considered when hosting data on-prem is a priority
b. User friendly when establishing connectors: Celonis is more user-friendly for business users with limited or no knowledge of SQL and connectors
c. Power BI – If you don’t want to move away from the look and feel of Power BI and connect with Power BI for process visualization Paf is a good look out too
d. Action Engine and Action Center are equally powerful for Celonis and UiPath however if you are using UiPath as a vendor for RPA then it adds great value in connecting Process Mining with RPA leveraging UiPath
e. Close to Microsoft – Minit was acquired by Microsoft last year hence it will be a good option in case we are looking for combining all in an enterprise solution in the future but its not been there as of yet.
Key benefits of process mining post adoption
Okay, so you’ve selected the right tool, implemented it and now what? Here’s the key benefits you should have realised:
a. Automation: Process mining help in identifying automation opportunities, automating them with building in automation and measure the impact of automation with actual or simulated data
b. Understanding: Process blueprinting gives you a deep understanding about the business and processes. It helps you to identify the variations in the process and the root causes. Moreover further analysis can define the impact and potential benefit in case a process improvement is to be incorporated
c. Networking: Process mining help in building a stronger network of information from the up stream and down stream processes. Keep in mind that prior to implementing process mining, your data management and process management is often fragmented and you lacked a one view approach to the entire process or customer journey in most of your organisation
d. Transformation: The two biggest levers of transformation journey is Transaction Journey Visualization and Quantitative and qualitative data visualization. Both of these are addressed with process mining as it helps in driving data driven transformation which gives higher confidence to the solution design and delivery teams
e. Knowledge management: Process demographics and process / transaction flow clubbed with process variation is currently not managed in most of the organisations and process mining will help in managing them to the newest version since the information in the knowledge management is not updated manually but is updated by the system based on the digital footprint left on the system, application and database
Since all you need to know about the business, process, customer, and application comes with a visual representation of data, it empowers you to manage your operations and transformation initiatives more effectively not only for business but also for IT. This can be the first stepping stone towards enhanced collaboration within Business and IT for effective transformation journeys.
Final words: Process mining is not just about technology
There are many vendors in the market helping us in the process mining journey and each one of them have their unique preposition, having said that the base principles and functionalities are similar hence organization should validate the vendors based on the requirements they have for their organizations, their current engagements and priorities.
At the end the tool will take you just far enough to showcase a pilot and a successful pilot if you don’t have the data and the will of the business to invest in streamlining the data, process and customer journey it will be difficult to move forward.
Learn more about process mining
You can also read about process mining in this 2020 post: Understanding supply chain thinking.
The conversation continues in our automation & robotics community.