『Five-minute Deming: Tampering』のカバーアート

Five-minute Deming: Tampering

Five-minute Deming: Tampering

無料で聴く

ポッドキャストの詳細を見る
A leader can make a process worse while trying very hard to improve it. That is the danger of tampering. When every disappointing result triggers a new rule, a new explanation, or a new adjustment, management may be reacting to routine variation as if something unusual happened.The result is more noise, more frustration, and less learning. The better question is not, What changed yesterday? The better question is, Do we know what kind of variation we are seeing?The temptation to chase the last numberW. Edwards Deming used a simple funnel experiment to make this problem visible. Imagine dropping a marble through a funnel toward a target. Even if the funnel stays in the same position, the marble will not land in exactly the same place each time.The natural impulse is to move the funnel after each miss, trying to compensate for the last result. That feels sensible. It feels active. It feels like control. But if the process is already stable, repeated adjustment can spread the results farther from the target.The act that feels like control becomes a source of instability. Management can fall into the same pattern whenever it treats the latest result as a command.A dashboard turns red. A customer complains. A weekly number dips. Someone asks for an explanation, and the organization rushes to change the work. Sometimes that response is necessary. A real special cause deserves attention. But when the process is stable, the better work is to improve the system, not chase each point.That is the problem ClearStep, a mid-sized B2B software company, faced when its support leaders tried to improve response time by changing the process after every bad day.The support dashboard that would not settle downClearStep sold project management software to manufacturers. Its support team handled setup questions, bug reports, billing issues, and urgent support calls. The team was capable, but its work arrived unevenly.Rina, ClearStep’s head of customer support, watched one number more than any other: median first response time. When it rose, customers complained. When it fell, the executive team relaxed.Monday morning, the dashboard looked bad. Response time had jumped from twenty-three minutes to thirty-seven. Rina opened the team meeting with a decision already forming.“We need a new rule. For the rest of the week, no one works on follow-up tickets until the new queue is under control.”Marcus, the operations analyst who helped the support team study workflow data, hesitated. He had been plotting daily response time for the past six months.“I know thirty-seven minutes looks bad,” Marcus said. “But it is still inside the range we have seen before.”“Customers do not care about ranges,” Rina said. “They care that we were slow.”“Agreed,” Marcus said. “But if we change the rule every time the number moves, we may be adding variation ourselves.”That was not what Rina wanted to hear. She was trying to be responsive, not careless. The team had already changed the escalation rule twice that month. One week, senior agents took every urgent ticket first. The next week, new tickets came first.By Thursday, response time improved, but reopenings were up. Customers got quick replies that did not resolve the issue. The team was moving faster and learning less.Deming named the trap plainly: “Mistake 1. To react to an outcome as if it came from a special cause, when actually it came from common causes of variation.”Mistake 1. To react to an outcome as if it came from a special cause, when actually it came from common causes of variation.— W. Edwards DemingRina asked Marcus to show the chart again. The bad Monday was unpleasant, but it was not outside the usual pattern. The system had been predictable for months. Response time bounced within a wide band because of uneven ticket routing, inconsistent urgency definitions, and too few agents trained on integration issues.“So doing nothing is the answer?” Rina asked.“No,” Marcus said. “Studying the system before we change the rules is the answer.”“Then what do we change?”“Not the queue every morning. We change the conditions that keep creating these wide swings.”That distinction changed the conversation. ClearStep still investigated real signals: outages, product releases, unusual customer spikes. But it stopped rewriting queue rules after ordinary variation. Rina’s team clarified urgency definitions, cross-trained agents on integration questions, and reviewed blocked tickets each day to remove causes of delay.The solution was not inaction. It was action aimed at the system.Why we keep treating noise like a signalWe drift into tampering because the pressure to respond is real. A leader sees a bad number and feels responsible for it. A customer is waiting. A team is anxious. An executive wants an explanation. In that moment, studying variation can sound like delay.But the demand for an explanation can create its own distortion. If every up ...
adbl_web_anon_alc_button_suppression_t1
まだレビューはありません