To be able to say whether a process is improving, it has to be measured. Yet measurement itself needs care: an indicator whose definition changes gives an impression of improvement while in fact saying nothing. This page describes which indicators are measured and how.
Rationale for measurement
A process that is not measured cannot be improved, because the effect of a change cannot be seen. Intuition misleads. Data produces decisions.
Stability of indicator definitions
If the definition of an indicator changes between periods, the comparison loses its meaning. Definitions are kept in writing. Any change is recorded.
On-time delivery rate
The share of jobs delivered on the promised date is measured. This is the most basic indicator of trust. The calculation method is defined.
Distribution of causes of delay
Delays are classified by cause. Waiting for a reply and a capacity problem call for different measures. The distinction is made.
Feedback rate
The share of jobs that receive a correction request is monitored. A rise in the rate gives a quality signal. The root cause is sought.
Distribution of error types
Incoming corrections are classified by type. A recurring type shows a process problem. The list is updated.
Separating differences of preference
Differences of preference that are not errors are kept out of the measurement. Mixing them in produces a false picture. Classification is done meticulously.
Errors caught by the review steps
The number of errors caught at revision and proofreading is monitored. This shows that review is working. The data gains value.
Number of questions
The number of questions that come up in a job shows the clarity of the source text. A high number may also point to a lack of preparation. Both possibilities are assessed.
Time spent waiting for replies
The time spent waiting for replies from the client is measured. This time affects the plan directly. It is shared transparently.
Planned and actual time
The estimate is compared with the actual. The deviation shows the quality of the planning. Data improves the estimates.
Capacity utilisation rate
How loaded the team is working is monitored. Constant maximum utilisation produces fragility. The balance is observed.
Memory and repetition gain
The gain that comes from the memory is measured and shared with the client. Transparency builds trust. The measurement is made regularly.
Limit on the number of indicators
A large number of indicators leads to none of them being monitored. A few meaningful indicators are chosen. The list is kept short.
Honesty of measurement
Figures are not shown as better than they are. Embellished measurement hides the problem and makes it grow. Honesty is fundamental.
Purpose of measurement
Measurement is carried out not to assess individuals but to improve the process. Using it to blame corrupts the records. The purpose is explained openly.
Periodic assessment
The indicators are read together at set intervals. A single indicator misleads when taken alone. The overall view pays off.
Sharing with the client
Some indicators can be shared with the client, and this keeps the relationship transparent. The on-time delivery rate and the time spent waiting for replies show the contribution of both parties. Sharing makes mutual responsibility visible.
Cost of measurement
Collecting data is effort too, and the benefit of the measurement must exceed that effort. An indicator nobody looks at is not collected. The list is pared down regularly.
The next step
The measurement results become input to process improvement. Without data, improvement is guesswork. The sequence is kept.