Lean teams use andon to surface problems while the evidence is still fresh. It makes stopping to fix a defect normal instead of risky. Over time, andon data shows which stations and causes need permanent countermeasures.
Where it comes from
Andon became part of the Toyota Production System as the visible side of jidoka. Toyota assembly lines used pull cords so any worker could flag a problem or stop the line.
At a glance
How an andon call works
Swipe sideways to see the whole diagram.
How to apply Andon
- 1
Define the trigger conditions
Agree on what counts as an abnormality: a defect, a missing part, a safety concern or falling behind takt. Make the list short and specific so operators never hesitate.
- 2
Design the signal
Choose signals that can be seen and heard from where responders work. Use consistent colors, such as yellow for help needed and red for line stopped.
- 3
Set the response rules
Name who responds to each signal and how fast. Set a response target the team can meet, so the team leader arrives in time to decide whether to fix or stop.
- 4
Train and protect the people who pull it
Explain that pulling the andon is expected, not a mark against anyone. Leaders should thank the person who raised the signal, every time.
- 5
Log every call
Record the station, time, cause and response time for each signal. Paper logs work, but digital andon makes trends easier to spot.
- 6
Review patterns and fix root causes
Review the log in daily meetings. Pick the most frequent causes and run problem solving so the same signal stops coming back.
Worked example (illustrative)
A packing cell adds a button at each station tied to a light above the line. When a carton jams, the operator presses it and the team leader arrives in under a minute on average. A week of logs shows 46 calls, and 31 of them come from one carton size, so the team adjusts the erector settings for that size. The next week brings 17 calls, with 4 from that carton.
Common mistakes
- Treating andon calls as failures, so operators stop using the system and hide problems.
- Installing lights and buttons without a defined responder or response time.
- Clearing each call without logging it, which throws away the data needed for root cause work.
When Andon is not the right tool
Andon only works if someone responds. Without a clear responder and a response target, calls go unanswered and people stop pulling the cord.
It also needs a culture where calling for help is welcomed, not blamed.
Andon cord
An andon cord is a rope or button at the workstation that any operator can pull when they see a problem. At Toyota, a pull lights a signal and calls the team leader, and the line stops at the end of the cycle if the problem is not fixed in time.
The cord only works if pulling it is safe and expected. Leaders thank people for pulling it, because each pull is a problem found early.
Andon board
An andon board is a display, often above the line, that shows the status of each station. Green means running, yellow means help called, and red means stopped. Many boards also show the shift target, actual output and downtime.
Digital andon systems send the call to a phone or watch and log each call, so the team can see which problems repeat. The andon software page lists tools for this.
Andon and jidoka
Jidoka is the principle of stopping to fix problems at the source. Andon is the signal that makes the stop visible and brings help.
Toyota describes workers being able to stop the line with a cord when they see an abnormality.
What to measure
Track calls by type, time to respond and time to resolve. A Pareto chart of call reasons shows which problems to solve first.
A falling number of calls can be good news or a sign that people have stopped calling. Check which one it is before reading it as progress.
A sample andon log and Pareto
These rows are illustrative, from the packing cell's first week. Each call records the station, time, cause, minutes to respond and minutes to resolve.
Station 3, 09:12, carton jam, 0.7, 4. Station 3, 10:40, carton jam, 0.9, 6. Station 1, 11:05, label missing, 1.2, 3. Station 4, 13:30, scale error, 0.8, 9.
Over the week, the 46 calls sort into carton jams 31, labels 8, scale errors 5 and other 2. The first bar holds two thirds of the calls, so that is where problem solving starts.
Signal colors and what they trigger
There is no universal color code, so each site writes its own and posts it where people can see it. A common starting point:
Green
Normal running
Response: Continue standard work
Yellow
Help needed, work can continue for now
Response: Team leader comes to the station and helps correct it
Red
Stopped, or a serious abnormality
Response: Stop, contain, escalate and fix before restarting
Blue
Material or part shortage
Response: A material handler restocks the station
White
Quality check needed
Response: A quality technician or trained leader confirms the condition
Each color needs a written meaning and a required response. 'Yellow means the team leader arrives within two minutes' leaves no room for different shifts to read it differently.
How a call moves from signal to fix
A typical andon call follows the same path every time:
- 1
Detect. An operator, a sensor or a machine control spots a problem with safety, quality, material, equipment or pace.
- 2
Signal. A cord, button, app or machine alert lights the andon and notifies the right responder.
- 3
Acknowledge. The responder confirms the call, so the operator knows help is coming.
- 4
Assess. Together they decide whether work can continue safely or has to stop, and whether product needs isolating.
- 5
Fix or escalate. If the first responder can't solve it within the agreed time, the next level is called.
- 6
Record and review. The call is logged with its cause and times, and recurring causes go to problem solving.
Manual and automatic triggers
Most plants use both:
Manual triggers
Cords, buttons and app requests. Operators often notice subtle problems, such as a part that doesn't feel right, before any sensor does.
Automatic triggers
Machine faults, slow cycles and out-of-spec test results sent straight from the equipment. They remove delay and capture repeatable conditions without anyone pressing a button.
Types of andon system
Choose by layout, response structure and how much data you need:
| System | Fits | Watch out for |
|---|---|---|
| Cords and buttons | Fixed lines with nearby team leaders | Easy to start, but calls are lost unless someone logs them |
| Stack lights | Machines and cells visible from a distance | Clear status at a glance, little detail on the cause |
| Wireless lights and buttons | Layouts that change, mobile cells, warehouses | Cheaper to install, but battery life and coverage need planning |
| Digital andon boards | Several lines and responders out of sight | Timers, reason codes and history, if screens are placed where people gather |
| Andon apps and mobile alerts | Support staff who move across a large site | Fast acknowledgment, but sending every alert to everyone gets them ignored |
Cords and buttons
Fixed lines with nearby team leaders
Watch out for: Easy to start, but calls are lost unless someone logs them
Stack lights
Machines and cells visible from a distance
Watch out for: Clear status at a glance, little detail on the cause
Wireless lights and buttons
Layouts that change, mobile cells, warehouses
Watch out for: Cheaper to install, but battery life and coverage need planning
Digital andon boards
Several lines and responders out of sight
Watch out for: Timers, reason codes and history, if screens are placed where people gather
Andon apps and mobile alerts
Support staff who move across a large site
Watch out for: Fast acknowledgment, but sending every alert to everyone gets them ignored
Andon software covers the digital options. A simple system with disciplined response often outperforms a complex one nobody trusts.
Andon examples on the floor
The same principles apply in assembly, machining, packing, food and logistics:
Quality concern
A component doesn't fit as expected. Instead of forcing it, the operator calls the team leader, who checks the standard and decides whether earlier units need containing.
Material shortage
A station is about to run out of fasteners or labels. A blue signal reaches the material handler before the line stops. Repeat calls point to kanban sizing or replenishment timing.
Equipment fault
A machine stops or slows. A red signal brings maintenance, and the downtime reason is captured for TPM work.
Safety concern
A spill, a damaged guard or a blocked aisle. Safety calls get their own urgency and are never buried among routine calls.
Questions for the andon review
Use these at the daily or weekly review to move from 'how many reds?' to 'what is the process telling us?':
- Most frequent calls: Which causes came up most often this week?
- Slowest responses: Which calls took longest to answer or to resolve?
- Repeats: Did any problem come back after being marked resolved?
- Reason codes: Did people pick the right signal type and reason?
- Escalation: Did responders follow the agreed path and times?
- Next problem to solve: Which recurring cause deserves root cause analysis or a kaizen?
- Avoided signals: Are there calls people no longer make because the response is poor?
A review rhythm
Spread the review across three time frames:
Daily
Open issues, major stops, missed responses and any safety or quality concern.
Weekly
Top recurring reasons, stations with repeat calls and countermeasures that are running late.
Monthly
Trends in downtime and response times, improvement actions completed and how widely the system is used.
Watch the gap between calls acknowledged and calls resolved. A responder who arrives quickly but lacks the parts or authority to fix the problem shows up there.
Choosing an andon setup
Start from the problem you want to solve, then check:
- Layout: Are stations fixed, spread out or often rearranged?
- Alert types: Quality, material, maintenance, safety, pace, or all of them?
- Responders: Who receives each alert, and what do they do first?
- Escalation timing: When does an unresolved call move up a level?
- Data needs: History, reason codes, response times, photos or comments?
- Integration: Should alerts come from machines or feed maintenance and production systems?
- Ease of use: Can operators raise a call in seconds without leaving standard work?
- Upkeep: Who maintains devices, updates rules, trains new starters and reviews performance?
Pilot in one area, then scale
Try andon in one area with a committed leader and visible problems. Watch what happens: are lights visible, alerts routed correctly, responders arriving, reason codes accurate? If calls are rare, find out whether the process is stable or people are hesitant. If calls flood in, check whether they are real problems or noise.
When you expand, keep the colors, escalation rules and daily review the same across the site, and let local details vary. Name an owner for the andon system as a whole, or it drifts into inconsistent use.
Mistakes that make andon go quiet
Most andon failures are about behavior, not hardware:
- 1
Lights with no written meaning. Yellow means one thing on days and another on nights.
- 2
Calls treated as operator errors. People stop calling and hide problems.
- 3
Data collected and never used. The same alert appears daily and nobody removes the cause.
- 4
Every alert sent to everyone. People learn to ignore them.
- 5
Reason codes too detailed. Operators pick the first one to save time.
- 6
Only the count measured. Response quality and causes removed matter more.
- 7
Supervisors overriding signals. To protect short-term output.
- 8
New and temporary staff not trained. They don't know the system exists.
- 9
Devices left unmaintained. Dead buttons and dim lights teach people the system doesn't work.
Guides on this topic
- How to measure OEE: start tracking without a big IT projectHow to measure OEE with a free paper tracking sheet before buying software: what to record, how to calculate it and when to automate OEE tracking.Read the guide
- One-piece flow: how single-unit flow cuts waiting and WIPOne-piece flow explained: a worked comparison with batches, takt and cycle time, the benefits to check, production and transfer batches, flow and pull, where it fits, the conditions it needs, a pilot plan and how to measure it.Read the guide
- Muri: how to spot and reduce overburden in leanMuri explained: what overburden means, how it differs from muda and mura, the common causes, an observation checklist, six ways to reduce it, examples from factories to software, and how to measure progress without hiding overload.Read the guide
- Lean manufacturing examples from shop floor to officePractical examples of lean in action: what the problem looked like, which lean tool was used and what changed, in factories and beyond.Read the guide
Sources
Books and papers show their publication year. Web pages show the month we last checked them.
- 1.Jeffrey K. Liker, The Toyota Way: 14 Management Principles from the World's Greatest Manufacturer (McGraw-Hill, 2004)
- 2.Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production (Productivity Press, 1988)
- 3.Toyota Motor Corporation, Toyota Production System (global.toyota, checked October 2026)