Latest / The Operator's Manual with Fexingo: Daily Decisions, Tactical Wins, and Business Execution / How Toyota Cut Line Stops 70 Percent With One Andon Tweak
Transcript
- Lucas: There is a moment in every Toyota assembly plant that looks like chaos to an outsider. A cord drops from the ceiling, an alarm sounds, and the entire line stops — every robot, every conveyor, every person. It is not a breakdown. It is the andon system in action. Any worker who sees a defect can pull that cord and stop production. And for decades, that was the whole point: stop the line, fix the problem, never let a defect pass. Luna: Right. And that philosophy has been taught in every operations class for thirty years. Empower the worker. Quality over throughput. But I remember reading that Toyota itself started to question whether too many stops were actually hurting quality in the long run. Lucas: Exactly. And that is what today's episode is about. In 2019, Toyota's Georgetown, Kentucky plant — their largest in North America, employing about 8,000 people — made one change to how the andon cord worked. Not removing it. Not limiting who can pull it. Just one tweak to the protocol. And the result was a 70 percent reduction in line stops — unplanned downtime — without any increase in defects. In fact, first-time-through quality actually improved by a few percentage points. Luna: Seventy percent. So what was the change? Did they add a delay? A manager override? Lucas: Both, sort of, but in a really specific way. The classic andon system works like this: a worker sees something wrong — a bolt that doesn't look right, a trim piece that's misaligned — and they pull the cord. The line stops immediately. A team leader runs over, assesses the problem, and either fixes it or escalates. The line stays stopped until the issue is resolved. That's the theory. In practice, what Georgetown found was that the line was stopping an average of 14 times per shift. And the average stop lasted about 45 seconds. Doesn't sound like much — but 14 stops times 45 seconds is about ten and a half minutes of lost production per shift, per line. Across multiple lines and shifts, that adds up to millions of dollars in lost output per year. Luna: And I'm guessing many of those stops were for minor things that could have been handled without a full stop. Lucas: That's exactly what the data showed. Roughly 80 percent of stops were for issues that a team leader could fix in under 30 seconds — a loose bolt, a part that just needed a tap, a label that was peeling. But because the line stopped instantly, you wasted 45 seconds each time. The fix was a two-step process. First, they added a 30-second 'problem timer' that starts when the cord is pulled. The line doesn't stop right away — it decelerates and stops only after 30 seconds if the problem isn't resolved. Second, they gave team leaders a handheld device that shows a live feed of which station pulled the cord and what the issue category is — so they can start walking toward the problem before the line even stops. Luna: So the worker still pulls the cord. The alarm still sounds. But now you have a buffer — a grace period — where the team leader can get there and fix it before the line stops at all. That's clever. Lucas: It is. And the results were dramatic. In the pilot area — one line assembling the Camry — line stops dropped from 14 per shift to just 4 per shift. First-time-through quality went from 97.5 percent to 99.1 percent. And productivity on that line went up about 12 percent because they weren't losing those ten-minute chunks every shift. Toyota has since rolled this out to most of its North American plants. Luna: I love this example because it's so counterintuitive. The instinct for most companies when they see a problem like this is to take the tool away. 'Workers are abusing the andon cord, so let's limit who can pull it, or require manager approval.' But Toyota didn't blame the workers. They changed the system around the cord. Lucas: That's the key insight. The andon cord is not the problem. The binary stop/no-stop logic was the problem. By adding a state between 'running' and 'stopped' — a 30-second grace period — they preserved the worker's authority to flag a problem while giving the team leader a chance to respond without a full shutdown. This is a broader principle: before you remove a tool or restrict access, ask whether you can add a buffer or a triage step. Luna: I've seen this same idea in software deployments. Instead of a full rollback when a bug is detected, you do a gradual ramp — deploy to 1 percent of users, then 10 percent, then 50. If the alert fires, you pause the ramp, not the entire service. Lucas: Great analogy. And in customer service, instead of immediately escalating a complaint to a supervisor — which takes the frontline agent out of the loop — you give the agent a 30-second window to check a knowledge base or consult a colleague before the escalation triggers. Same principle: buffer before full stop. Luna: You know, I want to pause on that principle for a second, because it ties into something bigger. A handful of listeners chip in monthly through buymeacoffee dot com slash fexingo, and that support is what keeps this show ad-free and lets us dig into cases like this. If these conversations have moved your work forward in some small way, it's a simple way to keep them coming. No pressure. We're grateful either way. Lucas: Yeah, it genuinely helps. And we're back to the andon cord. The thing I find most interesting is that Toyota didn't just change the protocol — they also changed the physical layout of the team leader's workspace. The handheld device they introduced shows a map of the line with colored dots: green for stations that are running, yellow for a cord pulled but not yet stopped, red for a full stop. That real-time visibility let team leaders prioritize which station to go to first when multiple cords were pulled. Luna: So it's not just a timer — it's a triage system. And that raises a question: did workers feel like their authority was undermined? I've read that some union representatives were initially worried that the 30-second buffer would make workers hesitate to pull the cord. Lucas: That was a real concern. Toyota spent about three months on training and communication before the pilot. They explained that the worker's role hasn't changed — you still pull the cord when you see a problem. The only difference is that the line might not stop if the team leader fixes it in time. In fact, workers quickly realized that their voice was still heard, and the alarm still went off. The plant manager at Georgetown said in an interview that pull rates actually increased slightly after the change because workers felt more comfortable pulling the cord knowing they weren't going to stop the whole line for a minor issue. Luna: That's a really important psychological point. When you remove the penalty — even an indirect one like 'my cord stop costs the company money' — people are more willing to flag problems. It's the opposite of what you'd expect. Lucas: Exactly. And this gets to the heart of why most process improvement efforts fail. They focus on the procedure without considering the behavioral incentives. Toyota didn't just change a process; they changed the cost-benefit calculation for the worker. Pulling the cord now has a lower downside — the line might not stop — so workers do it more. And because team leaders have better information, they resolve issues faster. The system becomes more resilient. Luna: So what's the one takeaway for someone listening who doesn't run an assembly line? How do you apply this to, say, a marketing team or a finance department? Lucas: I think the framework is: identify any process that has an all-or-nothing outcome. A full stop, a full rollback, a full escalation. Then ask: can we add a short buffer — a 30-second grace period, a 1-percent rollout, a 'pause and check' step — that preserves the ability to stop but reduces the cost of stopping? The andon cord tweak is a case study in adding a small amount of slack to a rigid system. And in most organizations, that slack is exactly what's missing. Luna: It also requires trust. You have to trust the team leaders to use that 30-second window correctly. Toyota invested in training and tools. They didn't just say 'go fix it faster.' Lucas: Right. The handheld device, the real-time map — those are enablers. Without them, the 30-second buffer would just be a delay. The combination of a procedural change and a tool change is what made it work. And that's a good lesson for any process redesign: don't change just the rule. Change the information flow that supports the rule. Luna: So for our listener who wants to try this — map out the all-or-nothing moments in their team's workflow, pick one, design a buffer and a triage tool, and pilot it with a small group. Lucas: Exactly. And measure before and after. Toyota measured line stops per shift, first-time-through quality, and productivity. You need at least one leading indicator — like stop frequency — and one lagging indicator — like quality or output. If you see improvement in both, you scale it. If you don't, you iterate. Luna: I think the Georgetown plant is still running that system today. It's become part of their standard work. And it's a reminder that even the most famous best practice — the andon cord — can be improved. The question is never 'is this process good?' It's 'how can we make it better right now?' Lucas: That's the kaizen mindset in one sentence. Toyota's own system gave them the permission to challenge the system. And that's the real meta-lesson: the best processes have built-in mechanisms for their own improvement. The andon cord was already a feedback loop. They just made the feedback loop faster and more granular.