Andon
= Signal Lantern
<1 Cycle
Response Time Target
3
Signal Levels
Escalate
Not Solve at Wrong Level

What Is Andon?

Andon (Japanese for "lantern") is a visual and audible signaling system that allows any operator to call for help immediately when they detect an abnormality. It is the operational mechanism behind jidoka — the principle that defects must never pass to the next process.

At its simplest, andon is a button or cord at each station connected to a light board visible to the entire area. When activated, it signals which station needs help and triggers a rapid response from the team leader. The system ensures that problems are surfaced in real time, not discovered hours or days later.

Signal Levels

LevelSignalMeaningResponse
🟢 GreenSteady green lightNormal operation — no issuesNone required
🟡 YellowFlashing yellow + chimeHelp needed but production continuesTeam leader responds before current cycle ends
🔴 RedSteady red + alarmLine stopped — cannot continue safely or with qualityTeam leader + supervisor respond immediately

Yellow Is the Most Important Signal

Most people think red (line stop) is the critical signal. But yellow — "I need help before this becomes a stop" — is where the real value lives. Effective yellow response prevents most red stops. Track yellow-to-red conversion rate: if operators go straight to red, it means yellow responses are too slow and people have given up on getting help before it is too late.

Escalation Structure

When a problem is not resolved at one level, it must escalate to the next — automatically, with defined time limits:

TimeResponderAction
0 – 1 minTeam LeaderArrive at station. Assess. Fix or call for support.
5 min unresolvedSupervisorJoin the response. Authorize resources. Contain quality risk.
15 min unresolvedDepartment ManagerDeploy additional resources. Evaluate production impact.
30 min unresolvedPlant ManagerStrategic decisions: run at reduced rate, shift to other line, notify customers.

These time thresholds connect directly to the tier meeting escalation structure. What cannot be resolved at T1 (team leader) escalates to T2 (supervisor), and so on.

What an open andon costs Nobody hands off and leaves — the supervisor joins, then the department manager, then the plant manager — so the price of an unresolved andon curves upward instead of running flat: 5 management-minutes by the 5-minute rung, 25 by 15, 70 by 30. The same problem closed at rung one costs 1.

Every time and responder is the page's own Escalation Structure table: team leader at 0 – 1 min, supervisor at 5 min unresolved, department manager at 15, plant manager at 30. The table's verb in row two is "join the response", so the earlier responders stay and the count on station runs 1, 2, 3, 4 rather than being passed along. Management-minutes are that count multiplied by the minutes it is on station: 1 × 5 = 5 by the 5-minute rung; + 2 × 10 = 25 by 15 minutes; + 3 × 15 = 70 by 30 minutes; and 4 more for every further minute, so 110 by 40. The rung-one close is the same arithmetic stopped at the page's own response target — one team leader at the station within the cycle: 1 × 1 = 1 management-minute, drawn to the same scale. That is the 70 : 1 the two marks compare. No cycle time is assumed and no quantity is added that the page does not give.

Implementing Andon

Install the physical systemButton or cord at every station. Central display board visible from the entire area. Audible signal. Can be high-tech (networked system with data logging) or low-tech (lights and buttons). Start simple.
Define response rolesWho responds to yellow? Who responds to red? What is the escalation path? Post it visibly. Every person must know their role.
Set response time targetsTeam leader at the station within 1 cycle time of a yellow call. Red stop response: immediate. Track and post actual response times daily.
Train operators on when to pullBe specific: quality abnormality, safety concern, missing material, equipment acting unusual, behind on cycle. When in doubt, pull. Over-pulling is always better than under-pulling.
Celebrate pulls, not just savesEvery andon pull that catches a problem early is a win. Post daily andon data: number of pulls, response times, problems caught. Thank operators who pull — they are protecting the customer.

Andon Data: A Goldmine

Every andon activation is a data point. Over time, andon data reveals:

AnalysisInsightAction
Pulls by stationWhich stations have the most problems?Pareto and focus improvement there
Pulls by reason codeWhat types of problems dominate?RCCA on top categories
Pulls by time of dayProblems cluster at shift start, after break, end of shift?Adjust startup procedures, handoff improvements
Response time trendsIs response getting faster or slower?Team leader workload, staffing adequacy
Yellow-to-red ratioAre yellows being resolved before they become stops?Response effectiveness metric
✅ Healthy Andon Culture
  • Operators pull without hesitation
  • Team leaders respond within 1 cycle time
  • Andon data reviewed daily, trended weekly
  • "Thank you for pulling" is the standard response
  • Top andon reasons drive improvement projects
❌ Broken Andon Culture
  • Operators avoid pulling ("I will get in trouble")
  • No one responds — lights stay on for hours
  • Andon system installed but not connected or monitored
  • Data not tracked or analyzed
  • "Just keep running" overrides the signal

🎯 Key Takeaway

Andon is the nervous system of your production floor. It makes every problem visible in real time and triggers an immediate response. Install a simple system, define response roles and time targets, train operators to pull without fear, and mine the data for improvement opportunities. The goal is not zero andon pulls — it is fast response to every pull and steady reduction of the root causes that trigger them.

Interactive Demo

Simulate an andon system on a 4-station production line. Trigger alerts and watch response times.

⚡
Try It Yourself
Andon System Simulator
▼
Click a station to trigger an andon call (yellow = help needed). If not acknowledged within 8s, it escalates to red (line stop). Click "Acknowledge" to respond. Track response times and escalations.
Stamping
Calls: 0
Welding
Calls: 0
Paint
Calls: 0
Assembly
Calls: 0
Andon Board — All stations running normally
0
Total Andon Calls
0.0s
Avg Response Time
0
Active Andons
0
Escalations
Ready for the full knowledge check? Test your understanding with guided scenarios and data export.
PROTake the Pro Knowledge Check →
Free forever · Every feature included

Stop reading, start modeling

Model your process flow, run simulations, optimize staffing with TOC math, and test your knowledge with 107 interactive checks — all in one platform.

Open Workbench → Pro Knowledge Check

Take this to a room

The running order

For production leadership designing or fixing an andon. They should leave with a response-time target per level and a named responder for each.

7 beats · 9 min
  1. 1

    Andon means signal lantern

    It makes a problem visible the moment it happens, to people who are not standing there.

    • Without it, a problem is known only by the person it is happening to.
    • Its job is not to solve. Its job is to summon.
    • The whole system is judged on one number: how long until someone arrives.

    Ask the room How does a problem at a station reach anyone here today?

  2. 2

    Three levels

    Green, yellow, red - and the responses are genuinely different, not just louder.

    • Green: normal. No action.
    • Yellow: help needed, production continues. Team leader before the current cycle ends.
    • Red: stopped. Team leader and supervisor immediately.
  3. 3

    Yellow is the important one

    Everyone assumes red matters most. Yellow is where the value is.

    • Yellow means I need help before this becomes a stop.
    • Effective yellow response prevents most red stops.
    • Track the yellow-to-red conversion rate as your real measure.

    Ask the room Do our operators use yellow, or go straight to red?

  4. 4

    Escalation is a clock

    Four levels, each with a time. Escalation is not about seniority - it is about unlocking resources.

    • 0 to 1 minute: team leader arrives, assesses, fixes or calls.
    • 5 minutes unresolved: supervisor joins, authorises resources, contains quality risk.
    • 15 minutes: department manager, additional resources.
    • 30 minutes: plant manager - reduced rate, other line, notify customers.
  5. 5

    Train when to pull, and mean it

    Be specific about the triggers, then make over-pulling explicitly safe.

    • Quality abnormality, safety concern, missing material, equipment acting unusual, behind on cycle.
    • When in doubt, pull. Over-pulling is always better than under-pulling.
    • If someone hesitates because they might get in trouble, the system is already dead.
  6. 6

    Celebrate pulls, not saves

    What you thank people for is what you will get more of.

    • Post daily: number of pulls, response times, problems caught.
    • Thank you for pulling is the standard response.
    • The goal is not zero pulls. It is fast response and fewer root causes.
  7. 7

    The data is a goldmine

    Every pull is a logged problem with a station, a reason and a time. Almost nobody mines it.

    • Pulls by station: Pareto and focus improvement.
    • Pulls by time of day: shift start, after break, end of shift - fix handoffs and startups.
    • Response time trend: team leader workload and staffing adequacy.

    Ask the room Are we recording pulls anywhere we could analyse?