The metric improved and the customers got angrier
A utilities company set a target to cut average complaint handling time from 6.2 days to 4.0 days within nine months. After three months the average is 4.8 days, a 22 percent improvement against baseline, and the operations director is citing it as a success. At the same time, escalated complaints are up 15 percent, repeat contacts are up 9 percent, and internal audit found that a quarter of complaints are closed between 4:30pm and 5:00pm and reopened the next morning under a new reference. The proportion of complaints recorded as resolved at first contact has risen sharply. Frontline staff say the target is the only thing they are measured on.
1. What is happening to the measurement, and what does it mean for the reported 22 percent improvement?
Read the worked answer
The reported improvement is not measuring what it claims to measure. Three distortions are visible at once:
- Closing at end of day and reopening under a new reference resets the clock on the original record. A quarter of complaints being closed in a half-hour window is not a plausible work pattern; it is the shape of a metric being worked around.
- Repeat contacts up 9 percent means some of those closures are not resolutions. The customer comes back, often as a new complaint with a fresh clock.
- Rising first-contact resolution alongside rising escalations is a contradictory pair. One of the two is being recorded in a way that flatters performance - most likely tickets are being marked resolved before the outcome is known.
Each of these is a displacement effect: the measured variable improved while the underlying process did not, and some of the improvement is being paid for downstream. The honest reading is that true handling time is unknown and probably close to baseline, with additional rework now embedded in the system. That has a second consequence: any target set against this baseline for next year is now built on a distorted number. Report the metric as unreliable rather than as an achievement, and say why.
2. Design a replacement measurement set and explain how you would sequence its introduction so it does not read as blame.
Read the worked answer
The design principle is to pair every speed measure with a quality or recurrence measure, so the fastest way to improve the score is to actually resolve the complaint.
Suggested set:
- Time to genuine resolution, measured to final confirmation with the customer rather than to ticket closure, and frozen once a complaint is reopened - a reopened complaint keeps its original clock.
- Recurrence rate - proportion of closed complaints generating a contact within 14 days. This directly prices the 4:30pm closures.
- Escalation rate and repeat contact rate, reported alongside handling time, never separately.
- Backlog age profile, so improvements cannot be produced by closing easy cases first.
Sequencing matters more than the metrics:
- Start with a two-week parallel run. Publish both the current figure and the new figure, and show where they diverge, so the divergence is presented as a measurement finding rather than an accusation about individuals.
- Remove the old target before publishing the new one. Leaving the old target live while introducing a stricter measure guarantees the workarounds persist.
- Present the change to frontline staff as removing a target they could only meet by cutting corners, and invite them to name the workarounds you have not found. They will know about more than audit did.
Retire the 4.0-day target explicitly and replace it with a paired target, for example resolution time plus recurrence below a stated threshold.