Andon: 7 Rules for Actionable Shop Floor Alerts

Andon gives managers a practical way to improve decisions, control costs, or remove avoidable delays. An assembler notices the wrong fastener kit. An alert names the station and lets the team correct the supply before restarting. To get a useful result, start with accurate inputs, clear ownership, and a method for testing decisions before changing the business process.

This article explains 7 topic-specific steps, worked examples, common mistakes, and ways software can support the process. It also shows what information teams should record and which results are worth tracking. The goal is a usable framework, not another report nobody acts on.

What andon actually means

An assembler notices the wrong fastener kit. An alert names the station and lets the team correct the supply before restarting. This illustrates why a reliable definition matters. Different teams may interpret the same number or status differently if its basis is unclear. For instance, an estimate is not an actual result, and a planned date is not proof that a task is complete.

First, identify the decision your organization needs to make. Then describe the data and assumptions supporting it. Keep a record of the timeframe, person responsible, and intended outcome. When conditions change, the team should update the forecast or workflow rather than hiding the difference.

1. Alert types

Focus: warning and line stop state is central to alert types. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated warning value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with warning and line stop state. Then ask how alert types should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place warning and line stop state near the action a user must take. A useful integration should carry the same verified value to the next step, signal trigger, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of alert types can be reconciled with the preceding stage, repeat analysis. If it cannot, correct the workflow before rollout.

2. Signal trigger

Focus: manual pull or sensor is central to signal trigger. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated manual pull or sensor value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with manual pull or sensor. Then ask how signal trigger should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place manual pull or sensor near the action a user must take. A useful integration should carry the same verified value to the next step, owner routing, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of signal trigger can be reconciled with the preceding stage, alert types. If it cannot, correct the workflow before rollout.

3. Owner routing

Focus: supervisor and backup is central to owner routing. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated supervisor value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with supervisor and backup. Then ask how owner routing should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place supervisor and backup near the action a user must take. A useful integration should carry the same verified value to the next step, machine location, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of owner routing can be reconciled with the preceding stage, signal trigger. If it cannot, correct the workflow before rollout.

4. Machine location

Focus: station and work order is central to machine location. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated station value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with station and work order. Then ask how machine location should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place station and work order near the action a user must take. A useful integration should carry the same verified value to the next step, response time, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of machine location can be reconciled with the preceding stage, owner routing. If it cannot, correct the workflow before rollout.

5. Response time

Focus: acknowledged timestamp is central to response time. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated acknowledged timestamp value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with acknowledged timestamp. Then ask how response time should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place acknowledged timestamp near the action a user must take. A useful integration should carry the same verified value to the next step, clearance check, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of response time can be reconciled with the preceding stage, machine location. If it cannot, correct the workflow before rollout.

6. Clearance check

Focus: verified remedy and release is central to clearance check. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated verified remedy value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with verified remedy and release. Then ask how clearance check should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place verified remedy and release near the action a user must take. A useful integration should carry the same verified value to the next step, repeat analysis, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of clearance check can be reconciled with the preceding stage, response time. If it cannot, correct the workflow before rollout.

7. Repeat analysis

Focus: repeat incident cause is central to repeat analysis. Identify the original information and the person who can validate it. In addition, set a clear boundary for this stage so that another employee knows when the work is ready to proceed.

Why it matters: Andon often depends on several people or systems sharing the same understanding. For example, an outdated repeat incident cause value could cause a false forecast or an incorrect operational decision. When that occurs, staff need to see the source, its effective date, and the reason for any revision.

How to apply it: Begin with one recent real transaction or job. Compare the available evidence with repeat incident cause. Then ask how repeat analysis should work if the inputs arrive late, change unexpectedly, or contradict another record. Make sure the person who owns the decision can document an exception and request review without overwriting the earlier result.

What software should show: Provide an ID, current status, responsible role, and relevant supporting fields. Do not fill the dashboard with every available statistic. Instead, place repeat incident cause near the action a user must take. A useful integration should carry the same verified value to the next step, alert types, without asking someone to type it again.

Practical check: Test the normal case, a missing input, and a duplicate or revised record. Confirm that the process preserves the history and that unauthorized staff cannot approve exceptions. Also examine whether the outcome of repeat analysis can be reconciled with the preceding stage, clearance check. If it cannot, correct the workflow before rollout.

Putting the method to work

Return to the starting example: An assembler notices the wrong fastener kit. An alert names the station and lets the team correct the supply before restarting. Walk through the inputs in the sequence above. Note where a value is estimated rather than observed, and where approval is required. This is especially important when a process affects cash, employees, customers, or safety.

Software can reduce repeated entry and make handoffs visible, but it cannot resolve uncertain business rules on its own. A ready-made product may be suitable when your requirements follow standard workflows. Meanwhile, a tailored system may make sense when different teams need unusual approvals or connected reports. Anomoz’s guide to related business software decisions can help assess how this work fits the broader operation.

Good data, governance, and external guidance

Use documented definitions, version dates, and access rules. For further information, see the relevant professional or official guidance and this additional authoritative resource. Apply the parts that match the business and ask qualified specialists to review legal, tax, accounting, or safety obligations when they are involved.

For example, a value can be mathematically correct but still misleading if the team uses the wrong period or counts a transaction twice. Therefore, keep the underlying records accessible to approved reviewers. Also separate original data from adjustments so managers can explain each variance.

Rollout checklist

1. Identify one important decision to improve.
2. Write down the calculation or process rule in plain language.
3. List the evidence and source of truth for each input.
4. Assign responsibility for verification and approval.
5. Test a normal case plus an exception or revision.
6. Train employees with their own real examples.
7. Compare results with a documented baseline.
8. Improve unclear steps before scaling the workflow.

For the pilot, choose a small scope that is easy to measure. Gather comments from the people who perform the work, not only from project sponsors. Next, review recurring mistakes and decide whether a change to the rule, data, or user interface will solve the problem.

Common mistakes to avoid

One mistake is using an impressive-looking result without explaining its assumptions. Another is relying on an old spreadsheet after introducing a new application. When two sources give different answers, employees often pick whichever figure seems familiar.

Instead, give each metric or workflow a named owner and an approved definition. Document exceptional changes and review whether the chosen method still fits the business. As a result, the same report can support useful decisions rather than arguments about whose data is correct.

Conclusion

Andon is most useful when teams can explain the reasoning behind each result. Use the 7 steps as a practical starting point, check them against real data, and adjust your workflow based on measured outcomes.