Problem Solving & Quality · Contain
Andon and Escalation Rules
Give every operator the right to signal an abnormality, and fix who responds, within how many minutes, before it escalates one level up.
- Time30 min
- FormatTeam
- StageContain
Andon and Escalation Rules: what it is and why it works
Andon and escalation rules give every operator the right and the means to signal an abnormality, and fix in advance who responds, how fast, and what happens if the problem is not resolved in time. Each station defines what counts as abnormal; a simple signal (cord, button, light or radio call) is always within reach; responders have a set arrival time and the authority to decide; and unresolved calls escalate automatically to the next level. Signals are counted and reviewed every week.
The method comes from jidoka: build the ability to stop on abnormality into the process, so problems surface while they are small and the evidence is fresh. It works only when signaling is safe and valued; people judge the system by the response, not by the signal. Compared with end-of-shift reports, andon cuts the time from detection to response from hours to minutes and makes the load on support functions visible. Andon does not replace emergency stops or safety alarms, which follow their own rules. It relies on standard work, which defines normal so that abnormal can be recognized, and it can be triggered by control charts, which provide statistical signals. Repeat calls become problem-solving topics.
What you need
- A definition of normal for each station: standard work, specifications, control limits
- A list of abnormality types that should trigger a call
- A signal system suited to the area: cord, button, light tower, radio
- Named responders and escalation levels for each shift
- Management commitment to respond promptly and without blame
What you get
- A one-page andon rule per area: triggers, responders, response times, escalation
- A working signal at every station
- A log of calls with times, causes and actions
- A weekly review that turns the top repeat causes into problem-solving topics
When to use it
When problems are hidden until the end of the shift because raising them costs too much.
How to do it, step by step
- Define what counts as an abnormality at each station: out-of-tolerance, missing material, unusual noise, safety concern.
- Provide a simple signal — cord, button, light, radio call — available to every operator.
- Fix the response: who comes, within how many minutes, and what they decide on arrival.
- Set escalation steps: if not resolved within the time limit, the next level is called automatically.
- Review signals weekly: count them, thank people for pulling, and turn repeats into problem-solving topics.
Worked example: Andon on an aerospace sheet-metal line
Illustrative scenario — figures are realistic but not from a real company.
A supplier of aircraft sheet-metal brackets finds that most nonconformances are discovered at final inspection, a day after the parts were made. Operators say calling a supervisor 'takes forever' and that stopping a press counts against them on the shift report.
- Each of eight press stations listed its abnormalities: first-piece check out of tolerance, missing or wrong material certification, die damage, unusual press noise, and any safety concern.
- A light tower and push button were installed at each press, and the line lead carried a pager showing the station number.
- Response rules: the line lead at the station within 3 minutes, with authority to stop, adjust or hold; after 15 minutes unresolved, the area supervisor and a quality engineer paged automatically; after 45 minutes, the operations manager.
- Time stopped for an andon call was removed from the downtime-against-operator metric and reported separately.
- The weekly review counted calls by type. Die damage and material-certification mix-ups led the list and became two A3 topics.
Result. Calls rose from almost none to about 25 a week in the first month, then settled near 15. Over the following quarter, nonconformances found at final inspection fell by more than half because problems were stopped at the press. The area supervisor's lesson was that the early surge in calls showed the system was trusted, not that quality had worsened.
Common pitfalls and how to avoid them
- Responding slowly or with blame.Measure response times, arrive as promised and thank the caller; one bad reaction can silence a team for months.
- Vague triggers.Define station-specific abnormalities against standard work, with concrete examples.
- No escalation path.Set time limits at each level and make escalation automatic rather than a judgment call.
- Treating fewer calls as the goal.Watch call counts alongside defects found downstream; a drop in both is good news, a drop in calls alone may mean silence.
Frequently asked questions
What is an andon system?
An andon system is a visual or audible signal that lets any operator call for help when something is abnormal: a defect, missing material, an equipment issue or a safety concern. It comes from the Toyota Production System, as part of jidoka. The signal is only half of it; the other half is a defined response by named people within set times, with escalation if the issue is not resolved.
Does pulling the andon always stop the line?
Not always. In many systems the first call alerts a team leader, who has a short window to resolve the issue before the line or station stops, often at the end of the cycle. The rules should be designed for the process. What must never be compromised is the operator's right to stop for a safety issue or to keep a defect from moving on.
How do you measure whether andon is working?
Track the number of calls, response time at each level, time to resolve and the main call causes, and compare them with defects found downstream and at the customer. A healthy system shows fast responses, recurring causes being eliminated and fewer escapes. Falling calls with steady downstream defects suggest that people have stopped signaling.
Origin
Andon — Toyota Production System, part of the jidoka principle (Taiichi Ohno, Toyota Production System, 1978).
Related methods
- First-Hour Incident ProtocolA fixed sequence for the first hour: make safe, stop the spread, preserve evidence, inform, appoint a leader.…
- Standard WorkWrite the current best way to do the task — sequence, key points, reasons, timing — so the fix becomes the…
- Control Chart (SPC)Plot the process over time with limits calculated from its own variation, to tell common-cause noise from…
More in “Contain”
Protect people, the environment and the customer while the real fix is found.