7 Steps for Process Mining POC's

Process mining is very hot technology these days.?The industry vendors are consolidating quickly at extremely high valuations.?This is creating a lot of buzz and leading companies to scramble to keep from missing out and getting left behind by their competitors.?Since the technology is very new, many companies are looking for proof before committing to purchasing the technology. Many times this proof takes the form of a proof of concept.?Proof of concepts can be very effective or fail miserably.

First, what is process mining??Gartner defines process mining as:

“Process mining is designed to discover, monitor and improve real processes (i.e., not assumed processes) by extracting knowledge from event logs readily available in today’s information systems. Process mining includes automated process discovery (i.e., extracting process models from an event log); conformance checking (i.e., monitoring deviations by comparing model and log); social network/organizational mining; automated construction of simulation models; model extension; model repair; case prediction; and history-based recommendations.”

Successful process mining Proof of Concepts lead to tremendous business value for companies.?

1.??????Identify stakeholders and business driver – Most POC’s are initiated by teams in IT, analytics, or process excellence.?These groups tend to drive change in an organization, evaluate new technology, and look for innovative solutions.?All of these objectives should be focused on driving real, tangible benefit to the organizations bottom-line.?Most executives in procure-to-pay, order-to-cash, supply-chain, or internal audit don’t fully grasp how process mining can benefit their business.?It’s important for the technologists to work with the business executives to choose a specific business driver to evaluate.?

It is not good enough to use process mining to “find something interesting” in the process audit logs.?Proof of concepts with this objective lead to drawn out evaluations that waste time for the business and technologists.?The choice of business driver also determines what data needs to be selected for case and event attributes.?Define a specific objective like:

-?????????Identify which process steps are causing shipment delays to high priority customers.

-?????????Identify the root cause of the uptick in delinquent payments from July.

-?????????What process differences cause plant A to have higher on-time delivery than plan B.

-?????????Are any employees, plants, or regions highly correlated to segregation of duties breaches in P2P?

2.??????Define success criteria – It’s important to know if your project is successful.?Without defined success metrics that might not be as easy as it seems.?It’s highly likely that you will be able to visualize your process from the event log data.?Automatic process discovery is only one component of an enterprise process mining product.?I’ve listed 11 components in this article.?In your POC you might want to test data extraction, cloud security, action connectors, ease of use, or integration with your BI platform.?Make sure that define how you will measure success and build your evaluation around these parameters.

3.??????Build the project plan and timeline – Process mining POC’s have many moving parts.?There is data extraction, audit log transformation, report creation, and business value review.?Each of these steps requires different people from the business, IT, and the process excellence team.?Coordinating everyone involved and ensuring that each step is completed on time requires detailed project management.?It is much easier to when the timeline, steps, pre-requisites, and responsibilities are defined up front.

4.??????Determine data needs – All successful process mining projects require data.?The first data requirement is the audit log.?This consists of transaction ID, event name and timestamp.?These three pieces of data provide the process mining algorithms with the data needed to recreate process map.?Beyond the audit, the rest of the data requirements stem from the questions that need to be answered.?For example, if you want to find maverick buying, you will need to data like PO number, PO amount, and Buyer.?If you want to identify how to capture more vendor discounts, you will need data like Vendor Name, Invoice Number, Discount Offered, and Payment Date.?

5.??????Evaluate data extraction & transformation and storage – Another area to evaluate is the long-term implications of data ETL and storage.?No company is going to mine production data directly inside the ERP system.?The data must be extracted, transformed from multiple tables, and stored for analysis.?Many companies have already extracted most of this ERP data into a data warehouse for reporting purposes.?You should review the existing data model what, if any gaps exist in your existing data.?If additional ETL is required, you should consider where this process happens.?Do you want to extract the data and do the transformation inside your internal infrastructure or inside a vendor’s cloud??Considerations include cost, speed, and security.?

6.??????Review analysis – Once the data is extracted, the next step is to build and/or review the analysis.?There are two main considerations; first – out of the box analysis, second – ease of customization.?It is important to start with pre-built analysis.?Beginning with a template that answers at least generic questions about your process provides a much faster ROI.?When you can simply connect to your transformed data and immediately get insight, the project is perceived as being more successful.?It’s not enough to just review standard dashboards and analysis.?At some point every company has questions that go beyond the standard analytics.?During the POC, you should evaluate the ease of customizing the process mining reports and dashboards.?It might also be a good time to evaluate how much of your teams report configuration skills are transferrable from your internal analytics platform to the process mining tool.

7.??????Report back to stakeholders – It’s important to report your findings and get feedback from the stakeholders you identified in the first phase.?Remember they had specific questions that they wanted to answer.?I like to do this in an iterative approach ending in a formal presentation. ?This presentation will include all of the analysis and a projected return based on the generated insights.??