Process discovery is the structured investigation of how a business process actually operates — capturing the current-state process steps, the systems involved, the data inputs and outputs, the exception handling, the decision points, and the time and effort consumed at each step — as the analytical foundation for an automation programme. Process discovery answers the question “what exactly does this process do?” before automation development begins, preventing the most common automation failure: automating a process that was poorly understood, and discovering mid-implementation that the actual process is significantly more complex than the version documented at the start. In the intelligent automation context, process discovery has two methodologies: manual discovery (interviews, process observation, document analysis, and process mapping workshops with process owners) and automated discovery (process mining tools that analyse system event logs to reconstruct actual process flows from digital evidence).
Process Discovery Methods
| Method | How It Works | Best For | Limitation |
|---|---|---|---|
| Stakeholder interviews | Structured interviews with process owners, practitioners, and supervisors | Understanding intent, exceptions, and workarounds not visible in systems | Subject to recall bias; may reflect intended process, not actual process |
| Process observation | Analyst observes practitioners performing the process in real time | Capturing screen interactions, manual steps, workarounds, and informal knowledge | Observer effect; time-intensive; limited observation period may miss infrequent cases |
| Process mining | Analyses event log data from ERP and other systems to reconstruct actual process flows statistically | High-volume transactional processes with system event logs (Oracle EBS, SAP); identifying deviation from intended flow | Requires accessible event log data; cannot capture steps outside the logged system |
| Task mining | Desktop recording tools capture user interactions (keystrokes, clicks, application switches) across a sample of practitioners | Understanding desktop-level steps that don’t appear in system logs; capturing workaround steps | Privacy considerations; representative sample required; may capture sensitive data |
Process Discovery for Oracle EBS in GCC Finance
Process discovery for Oracle EBS-based finance processes in GCC enterprises has specific considerations beyond standard process mapping. Oracle EBS processes frequently have significant gaps between the intended process design (as documented in the original EBS implementation) and the actual operating process (as practiced by the finance team after years of workarounds, exceptions, and customisations). Process mining against Oracle EBS event logs — using the EBS transaction audit tables as the event log source — reveals the actual sequence of transactions, the frequency of out-of-sequence steps, and the extent of manual journal entry corrections that supplement the automated EBS process. This evidence-based process picture is more reliable as an automation specification than interview-based process maps that capture the intended process rather than the actual process.
What Goes Wrong in Practice
The most common process discovery failure is a discovery that is too shallow — producing a high-level process flow that misses the exception handling, the informal workarounds, and the edge cases that constitute 20% of the process volume but 80% of the automation complexity. A five-step AP invoice process that appears straightforward in the process map may have 15 documented exception paths (PO not found, quantity mismatch, duplicate invoice, supplier blocked, currency mismatch, period closed) that each require different handling. Automation built from the simplified process map fails in production when the exception paths are encountered. Process discovery must specifically elicit and document all exception paths — not just the happy path — before automation development begins.
How Loop Wise Solutions Conducts Process Discovery
We conduct process discovery using a combination of practitioner interviews, process observation sessions, and Oracle EBS event log analysis where available — producing a process specification document that captures the happy path, all exception paths, decision logic, system touchpoints, and current error rates. The process specification is reviewed and signed off by the process owner before automation development begins, establishing the agreed scope and logic baseline for the automation build.
