Evaluating a process through one KPI is easy. It is rarely enough.
Average completion time may have fallen while errors and rework increased over the same period. Has the process really improved?
Likewise, SLA achievement may be high, but if achieving it requires significant manual effort, sustainability becomes a separate question.
Process performance should therefore be evaluated across several dimensions together.
Time
End-to-end completion time is one of the most basic measures.
But looking only at total time is not enough.
It is more useful to separate:
- Active processing time
- Waiting time
- Queue time
- Handover time between steps
- SLA achievement
- Cycle time of selected activities
A process taking a long time does not necessarily mean people are working on it for a long time; in many processes, a large share of time is spent waiting between activities.
This distinction can directly affect the decision between adding resources and redesigning the process.
Quality
Speed alone is not enough.
Useful measures of process quality include:
- First Time Right
- Error rate
- Rework
- Reopen rate
- Incomplete processing
- Repeat complaints or reapplication
If controls are reduced to complete a process faster but the error rate rises, overall value may actually fall.
Effort and cost
How much manual effort does a process require?
How many different people handle the same case?
How much of the work is repeated?
Cost per transaction is a useful KPI, but understanding where that cost is created across activities and variants provides a more actionable insight.
Flow
Process performance should not be measured at the activity level alone.
Also consider:
- How many handoffs occur?
- How many variants occur?
- Which flows take longer?
- Which paths produce more rework?
- Where does work return to earlier steps?
Process mining can make it easier to analyze these measures directly on the actual flow.
Customer outcome
Customers do not experience departmental KPIs. They experience the end-to-end result.
Operational KPIs should therefore be connected, wherever possible, to customer outcomes such as response time, resolution time, delivery time, repeat contact or satisfaction.
Good process performance is not about maximizing one KPI; it is about balancing time, quality, cost, conformance and customer outcome.
Conformance and control
Improvements in speed or cost should not come at the expense of required controls.
Mandatory activities, policy violations, SLA, role and authorization rules, and conformance can therefore also be part of performance measurement.
Why are averages not enough?
Average values can hide important differences.
A process may look healthy on average while a specific customer segment, product or variant performs significantly worse.
Median, distribution, percentiles and segment-based comparison can therefore provide a more accurate picture.
What should measurement be for?
The objective is not to produce more dashboards.
Good measurement should answer the question:
Where can we intervene to create a meaningful change in the process outcome?
A KPI creates value only when it is connected to a decision and an action.
← All articles