When offices closed in March 2020, many companies discovered for the first time just how fragile their processes actually were. Order approval steps were waiting days for a response. Purchase requests were stuck in limbo. Customer complaints were piling up in queues that no one fully owned. Managers could feel the disruption, but they lacked the data to show where it started, which step was genuinely blocked, and why. This is the moment when process mining moved from being an abstract analytical method to an operational necessity.
Process mining automatically reconstructs real process flows from the event logs generated by enterprise systems. Every step in an ERP, CRM, or service management platform is recorded with a timestamp; process mining combines these records to visualize how a process actually runs. It quantifies the gap between the designed process and the process as it is lived — capturing deviations, loops, and unexpected paths in numerical terms. The method gained meaningful enterprise traction from the mid-2010s onward. In Turkey, large-scale manufacturing and financial services companies began running pilot projects with these tools around 2017 and 2018, though widespread adoption remained limited to organizations with mature data infrastructure.
During the pandemic, the value of process mining showed itself most clearly in three areas. The first was the wave of new process deviations triggered by the shift to remote work. Steps that had been resolved informally in the office — getting a signature from a manager down the hall, reaching a quick verbal agreement with the finance team — now had to pass through digital channels. This transition happened without formal process redesign. Process mining detected these informal workarounds as soon as they appeared in the event logs. An analysis at a logistics company based in Istanbul revealed that the shipment approval process had completed in an average of four steps before the pandemic; during the remote work period, the same process had fragmented into seven distinct variations. None of those variations had been documented. All of them emerged from employees making real-time decisions under pressure.
The second area was measuring the actual process impact of supply chain disruptions. Mid-sized manufacturers in Turkey could only see clearly how supplier delays were reshaping their procurement workflows once they had process mining data in front of them. Purchase requests were waiting three times longer than usual at the approval stage. Some requests were cancelled before completion and reopened, creating duplicate loops that inflated cycle times without anyone intending it. This picture quantified what managers had been sensing but could not prove with standard reports. Crucially, the data also revealed which supplier categories were causing the longest delays, feeding directly into decisions about alternative sourcing.
The third area was the sudden surge in customer service and complaint management volumes. As e-commerce channels expanded rapidly under pandemic conditions, customer request volumes grew disproportionately fast. When event logs from service management systems were analysed, resolution times for certain complaint categories had stretched by a factor of two to four. More importantly, the analysis showed that this slowdown did not originate from a single step. The underlying problem was that multiple teams were drawing from the same queue with no clear prioritisation rules. Process mining combined this multi-layered bottleneck into a single visual flow map, showing precisely which team was blocked at which step — something no dashboard built on aggregate metrics could have revealed.
The limitations of process mining deserve equal attention. The method works in direct proportion to data quality. If event logs are incomplete, inconsistent, or scattered across disconnected systems, the analysis can mislead rather than clarify. Many small and mid-sized businesses in Turkey run fragmented ERP environments where some process steps live in spreadsheets, others in email chains, and some exist only as verbal agreements. In these conditions, process mining alone is not sufficient. Standardising event records and consolidating data sources must come first. Beyond data quality, process mining is a diagnostic tool, not a corrective one. It shows the bottleneck but does not resolve it. Resolution still depends on human judgement and deliberate process redesign. In a crisis environment, losing sight of this distinction leads to unrealistic expectations from the tool.
The teams that extracted genuine value from process mining during the pandemic treated it as active decision support rather than a reporting layer. Process views refreshed weekly or fortnightly allowed managers to track which steps had stabilised and which remained in deviation. This approach offered a level of operational clarity that was particularly valuable in a period when currency pressure, supply uncertainty, and workforce disruption were all present simultaneously. If your organisation already captures process data in an ERP or service management system, the raw material for process mining is already available. The starting point is not a large transformation programme. It is extracting the event log for a single critical process and comparing the actual flow against the intended design. That gap, made visible, is where the real work begins.
This article was originally written in Turkish by Gökhan MERCANOĞLU on February 10, 2020 and has been automatically translated into English and other languages using machine translation.