Consider the purchase order process at a mid-sized manufacturing company: a request is created, approval steps are completed, and a work order is sent to the supplier. On paper, this flow finishes in five steps and takes roughly two business days. But when event logs are pulled from the ERP system, a different picture emerges. A significant share of cases skip the approval step entirely, others enter unplanned manual review loops, and some cycle back to the starting point. To management, this looks like ‘the process sometimes runs slow.’ In reality, each deviation type carries a measurable cost item. Process mining provides the systematic method to make that cost visible.
Process mining analyzes event logs generated by ERP, CRM, or accounting systems to reveal how a process actually operates. The primary output is a discovered process model — the real flow map. When this map is compared against the ideal model defined during process design, deviations become visible. Deviation analysis runs across three core categories: skipped steps, rerouted flows where the process takes an undesigned path, and looping repetitions. Because each category carries different cost dynamics, pricing must be structured at the category level.
The first step in calculating deviation cost is measuring the time delta for each deviation type. Event logs contain start and end timestamps for every case, so the difference between the average completion time of the ideal flow and the completion time of deviated cases can be computed directly. Multiplying this time delta by the hourly labor cost of the staff involved in that process yields the labor cost per deviated case. If a purchase approval loop adds an average of four hours and two specialists are involved, the labor cost per deviated case is concrete and calculable. Multiplying by deviation frequency produces the monthly or annual total.
The second cost layer is rework and error expenditure. Looping deviations typically originate from data entry mistakes, missing documents, or unclear approval authority. Each loop means the same task is performed at least once more, consuming both labor and system resources. In some processes, error costs are even more tangible: reversed invoices, late payment penalties, or documents cancelled due to e-Invoice compliance failures leave direct financial records. Process mining links these records to event logs, revealing which deviation type generates which error cost.
The third and most frequently overlooked cost item is opportunity cost. When a process deviates from its ideal path, the impact extends beyond that single case — other cases sharing the same resources are delayed as well. Capacity analysis becomes essential here: process mining tools visualize resource concentration and waiting times at individual process steps. The nodes with the highest concentration are where opportunity cost accumulates. Identifying these bottlenecks forms the foundation of the ROI analysis that determines where improvement investment should be directed.
Once deviation pricing is complete, an objective framework for improvement prioritization emerges. The annualized cost calculated for each deviation type is compared against the implementation cost of the solution that would eliminate it. This comparison shows which process improvement delivers a return in the shortest time. In practice, managers often rely on intuitive prioritization: ‘let’s fix the process that generates the most complaints.’ But complaint volume and cost volume do not always align. A small, high-frequency deviation may generate a higher total cost than a large but rare one. Data-driven prioritization eliminates this blind spot.
For SME managers evaluating a process mining investment, the critical decision criterion is straightforward: does the existing ERP or accounting system produce event logs, and are those logs accessible? If the data infrastructure is in place, the time and investment required to calculate deviation costs is considerably lower than traditional consulting engagements. Cloud-based process mining tools have made this access more practical, but data quality and process ownership questions must be resolved at the outset of the project. No matter how strong the analytical output, an improvement recommendation without a named owner does not reach implementation. Organizational readiness determines the return on this investment just as much as the technical infrastructure does.
This article was originally written in Turkish by Gökhan MERCANOĞLU on June 25, 2018 and has been automatically translated into English and other languages using machine translation.