# From a production stop to a supported explanation A practical way to investigate production downtime: establish the event, examine competing explanations and keep the limits of the evidence visible. Canonical: https://vieox.com/insights/investigating-production-downtime/ Publisher: VIEOX Published: 2026-10-20 Updated: 2026-10-20 ## How can a team investigate production downtime without jumping to a cause? Define the event, assemble the relevant records and keep observations separate from explanations. Check the sequence, examine reasonable alternatives and ask what evidence would support or challenge each one. Record both the conclusion and any uncertainty that remains. A production stop can leave several records and several plausible explanations. A useful investigation gives each one a fair examination, connects it with the actual event and shows the team why a conclusion is supported. ## Give the event a precise description Begin with a sentence the team can agree on: which process was interrupted, during which period and what work was expected. Record how the interruption's beginning and end were determined. If those boundaries are uncertain, keep the uncertainty in the description. An alarm message, an equipment state and an operator's account may describe different parts of the same event. A label in a record is useful information to examine. Its meaning depends on the equipment and circumstances in which it was recorded. Decide which question this review should resolve. Understanding why a line paused, why recovery took time and why similar interruptions recur may require different evidence. Give each question a clear place in the investigation so its answer can be assessed. ## Build a sequence the records can support Place the available observations before, during and after the interruption. For each entry, retain the equipment reference, source of the information and what its time represents. Include relevant operator accounts and identify where a time is approximate. HSE's investigation workbook recommends an objective examination of the chain of events. Although written for health and safety investigations, that discipline is also useful when reviewing production records. The practical aim here is a sequence the team can examine and question. Do not force uncertain entries into a precise order. If two records use different time references, establish what comparison is possible. If the evidence covers only a broad interval, show that interval. Gaps are useful questions for the review, rather than spaces to fill with an assumed event. Sources: [1](https://www.hse.gov.uk/pubns/hsg245.pdf) ## Give each explanation a fair test Write a proposed explanation as a statement that can be examined. Ask what would need to be true for it to account for the interruption. Then identify which records support it, which challenge it and which necessary information is missing. NIST's statistical guidance makes a relevant distinction: a relationship between observations does not establish cause and effect. A signal appearing near a stop can direct attention to a period worth examining. Further process knowledge and evidence are needed to explain its role. For each plausible explanation, choose a question that could change the team's view. Could the signal also appear during uninterrupted production? Is its timing consistent with the proposed sequence? Was it recorded at the equipment thought to be involved? Use available evidence and the site's authorised investigation methods to examine those questions. Sources: [2](https://www.itl.nist.gov/div898/handbook/eda/section3/scatterp.htm) ## Check whether recurring events belong together Several records carrying the same reason may still need closer reading. NIST's maintenance-log research discusses inconsistent terminology in machine records. For a downtime review, establish what each description meant before using it to group events. Compare a manageable set of events with a clear common basis. Check the equipment, product or activity, operating circumstances and the way the event was recorded. Note differences that could matter to the explanation. A shared word such as jam or trip can be a useful way to find records. Read the detail behind it before treating every entry as the same cause. Keep the original account available if a later review changes the classification. The group should remain explainable to someone who did not create it. Sources: [3](https://www.nist.gov/publications/using-text-visualization-aid-analysis-machine-maintenance-logs) ## An example: one interruption, several clues Hypothetical example: a packaging line stops and the team finds a pressure-related message in the event history. An operator also recalls a wait at the upstream process. The first suggestion is that the pressure condition caused the interruption. The team places both observations against the production period. It finds that the operator's note gives an approximate time, while the event record uses a different time reference. The current evidence does not establish which observation came first. The review checks what the message represents and asks whether comparable messages appear in periods without a stop. It also looks for information about the upstream wait. Those enquiries could strengthen, weaken or change the initial explanation. The interim finding states what is known, what remains unresolved and who will examine the missing information. It gives the next review a concrete starting point without turning the earliest suggestion into a confirmed cause. ## Close the question with an explanation people can follow Record the conclusion alongside its supporting observations and the alternatives considered. State the scope: this event, this group of events or a broader recurring concern. If the records are insufficient, describe the unresolved question and what evidence would help. Keep the conclusion about cause distinct from the record of work completed. A task can be finished while the explanation still needs examination. Agree how subsequent information will be reviewed and where the team will record a revised conclusion. These are review practices for the people responsible for the process. VIEOX's focus is making operational information easier to understand; ConverEDGE is designed to help teams examine current conditions and recorded activity. A recurring interruption and the question behind it provide a useful starting conversation. ## Practical starting points - Define the equipment, production period and interruption boundaries. - Keep each observation linked to its record and time reference. - Write down the explanations being considered. - Look for evidence that could challenge the preferred explanation. - Check the meaning of repeated reason labels before grouping events. - Record the conclusion, its scope and the questions still open. ## Questions, answered ### Is the first alarm in the record necessarily the cause? Its position in the record is a reason to examine it. Check what the message means, how its time was recorded and how it relates to the process. Establishing its role requires supporting evidence about the event. ### When is it reasonable to leave the cause unresolved? When the available evidence cannot support a reliable explanation. Record what has been established, the limits of the record and the next useful enquiry. Keep operational responses within the site's established procedures while the investigation continues. ## Sources and further reading - [HSE — Investigating accidents and incidents (HSG245)](https://www.hse.gov.uk/pubns/hsg245.pdf) - [NIST — Scatter Plot](https://www.itl.nist.gov/div898/handbook/eda/section3/scatterp.htm) - [NIST — Using Text Visualization to aid Analysis of Machine Maintenance Logs](https://www.nist.gov/publications/using-text-visualization-aid-analysis-machine-maintenance-logs) ## Related reading - [Put production activity in context](https://vieox.com/solutions/production/#bring-in-context) - [Understand what the timestamp represents](https://vieox.com/insights/operational-data-context/#understand-the-time) - [Keep an open question clear at shift handover](https://vieox.com/insights/industrial-shift-handover/#carry-open-questions) - [Interpret variation in a quality review](https://vieox.com/solutions/quality/#interpret-variation)