metric gaming
Also called: gaming a metric, gaming the metric
What happens when one number is tied to rewards: people improve the number in ways that hurt the goal behind it, such as closing tickets fast but unresolved. Pairing the target metric with a quality check reduces it.
Metric gaming is what happens when a number starts deciding pay, rank or praise: people find ways to improve the number that don't improve, or even hurt, the thing it was meant to show. As soon as pay or praise depends on a number, people start working on the number instead of on what it was meant to show.
It is rarely cheating. Usually the rule itself makes the shortcut rational: if a closed ticket pays and a reopened ticket closed again pays twice, closing early is the sensible move. If a hard case takes an hour and pays the same as a two-minute one, hard cases wait. People respond to what the system rewards, so the fix belongs to the system, not to a lecture.
Three habits reduce it. Pair every target with a guardrail, a quality number that must not get worse (tickets closed with the reopen rate, reply speed with satisfaction). Read the numbers at team level and talk before paying, so a guardrail is a question rather than an automatic penalty. And ask the team how the rule could be gamed before it starts: the people doing the work spot the loopholes first.
Example
Klasna's Support team runs a bonus of UAH 20 per ticket closed above 60 a week, for four April weeks. The bonus total is UAH 7,200. Two coordinators, before (four March weeks) and under the bonus:
Taras's closes rise by 98 (98 / 381 = 25.72%) and his reopens roughly double (about 36 → 73). He receives 4,780 / 7,200 = 66.39% of the bonus. Maryna keeps her pace and her quality and earns UAH 60. Across the team, 74 of the 277 extra closes (26.71%) are tickets that were reopened and closed again, counted twice.
Nobody broke a rule. The rule paid for closing, not for solving. The guardrail that would have caught it, the team's reopen rate, went from 5.61% to 8.91%.
Common mistakes
- Blaming the people. If the rule pays for the shortcut, the rule is the problem; change it before judging anyone.
- One number tied to money. Every single-metric incentive needs a guardrail read before anyone is paid.
- Counting events that can repeat. A reopened ticket closed again, a split conversation, a re-sent message: check whether the count can be inflated.
- Turning the guardrail into a new league table. Read quality at team level and talk first, or people game the guardrail too.
- Skipping the loophole question. Ask the team how the rule could be gamed before launch; they will know.
Learn it in the course
- Goals people accept · Specific, hard but reachable, agreed goals with progress feedback, and guardrails against what one metric does to behavior.
- Case: the "tickets closed" bonus · Four weeks of bonus data: closes are up, quality is down. Read the numbers and propose to Iryna what to change.