When machines report good performance but the factory struggles to deliver finished goods.
Can technology solve a factory’s production problems when the real causes remain hidden on the shop floor?
Mr. Seth owned a manufacturing unit. His machines were running for long hours, overtime had become routine, and production teams were under constant pressure. Yet the factory continued to miss production targets.
Customer delivery commitments were becoming difficult to meet. Work-in-progress (WIP) was increasing, and management meetings were increasingly focused on explaining production shortfalls.
His production manager suggested installing an AI-based production monitoring system.
“We have several machines, multiple production stages and a lot of manual reporting. By the time we collect the figures, prepare reports and identify the problem, the production shift is already over. We need real-time information. An AI-based system could help us monitor production and identify losses faster.”
Mr. Seth wanted an independent opinion. He invited Sharad Chandra, Founder & Lead Consultant | Operations at Sampurna Operations Consultancy, to understand the situation before making an investment.
Sharad listened carefully. He respected the production manager’s experience, but wanted to understand the factory’s existing information and operating conditions first.
“Technology can certainly help us capture and analyse information faster. But before deciding what to install, I would like to understand three things: what information are we collecting, how reliable is it, and does it explain the performance of the complete production process?”
The production manager responded confidently.
“We already prepare daily production reports and machine OEE reports. We know the performance of every machine.”
Sharad asked:
“Do we also know how effectively the complete factory converts raw material into quality finished goods?”
There was a moment of silence.
The production manager showed Sharad the machine-performance reports. Several machines had satisfactory OEE figures. Their availability and performance appeared reasonable.
But finished goods output was still below plan.
Sharad examined the reports.
“Are we measuring how effectively individual machines operate, or are we also measuring how effectively the complete production system delivers finished goods?”
The manager explained that production quantities were collected from each department.
“That is useful,” Sharad replied. “But if one machine produces more components than the next operation can process, where does that additional production go? If material is unavailable at another operation, what happens to the production flow? And if WIP increases every day, are we sure that the additional machine output is helping us meet customer demand?”
The discussion had moved beyond machine performance. Sharad wanted to understand the complete journey of material—from the first operation to finished goods and dispatch.
Sharad requested the production plan, machine-wise production reports, shift records, material availability reports, WIP records, rejection reports and finished goods data.
The factory collected information about:
Planned production and actual production.
Machine running time and downtime.
Production quantities at different operations.
Raw material receipts and consumption.
WIP at different stages.
Rejection and scrap quantities.
Manpower deployment.
Finished goods and dispatch.
However, collecting information was not the same as verifying it.
Sharad asked:
Does reported production match the actual accepted quantity?
Does recorded WIP match the physical quantity on the shop floor?
Are material shortages recorded when they occur?
Are rejection quantities and reasons correctly classified?
Does machine downtime reflect the actual reason for stoppage?
Are production targets and revisions traceable?
Are significant production deviations investigated?
He explained:
“A report can contain hundreds of figures and still give us the wrong picture if the information is incomplete, incorrectly classified, delayed or manipulated. Before relying on a report, we must establish whether it represents what is actually happening in the factory.”
Sensors and machine controls can capture operating signals. Production-monitoring systems and ERP or MES applications can organise operational information. Analytical tools, including AI, can help identify patterns and highlight deviations.
But none of these automatically guarantees that the underlying information is correct.Production quantities must be reconciled with actual accepted output. WIP records must be checked against physical stock. Material consumption, rejection and downtime must have clear recording rules. Important figures should be traceable to their source.
The objective is not simply to generate more reports. It is to establish whether management can trust the information used to make decisions.Sharad requested the previous thirty days of production records.
He compared machine reports, handwritten shift logs, production summaries and material movement records.
One example attracted his attention.
An operator had recorded a 45-minute tool change in the handwritten shift log. However, the digital production summary had classified the associated lost production time under speed loss and minor stoppages.
Sharad asked the production manager:
“Why has a 45-minute tool change been recorded as a minor stoppage?”
The manager checked the reporting system. The available dropdown categories were limited. Operators and supervisors were selecting the closest available options, even when those options did not accurately describe the event.
Sharad explained:
“How can we analyse production losses correctly if our reporting categories do not represent what actually happened?”
Under conventional OEE practice, a full stop for setup or tool change during planned production time is generally an Availability Loss, classified under setup and adjustments. Reduced speed and brief minor stops while the process is running are generally Performance Losses.
The factory must define planned production time, stop thresholds and loss categories consistently. If a particular planned activity is excluded from the OEE calculation under the factory’s agreed methodology, it should still be recorded and reviewed separately. Excluding it from one calculation should not make the operational loss invisible.
As Sharad continued the review, he found similar gaps in other areas.
A material shortage could be recorded as machine downtime without identifying the material issue.
A machine’s operating status could fail to show that production was waiting for a component.
WIP could be recorded at shift-end without physical reconciliation.
Rejection could be reported as a percentage without identifying whether the cause was material, tooling, machine condition, process settings or working method.
A production delay could be explained as a general operational issue without structured root-cause analysis.
The factory was collecting information, but its reporting system was not consistently connecting each loss to its actual cause.
Sharad explained that a useful production-reporting system must connect planned performance, actual performance and the reasons for deviations.
What was the planned production quantity?
Was the plan based on realistic process capacity?
Were production targets and revisions authorised?
Was the production sequence followed?
What was the actual running time?
How much time was lost?
What caused the stoppages?
Was output consistent with the machine’s validated operating capability?
Was the required material available at the right time?
Were there shortages or inventory inaccuracies?
Was material consumption recorded correctly?
Were shortages caused by procurement, stores or internal movement?
How much unfinished material was present at each stage?
Did physical quantities match system records?
Where was material accumulating?
How long was it waiting for the next operation?
How much was rejected, reworked or scrapped?
What were the actual defect reasons?
Were recurring defects investigated?
Were corrective actions implemented and verified?
Were operators available as planned?
Was manpower allocated according to workload?
Were operators waiting for material or work?
Were skill gaps or workload imbalances affecting output?
What was delayed and for how long?
What was the actual reason?
Was root-cause analysis conducted?
Was corrective action implemented and checked for effectiveness?
Overall Equipment Effectiveness (OEE) measures equipment performance through three components:
The proportion of planned production time during which the equipment was running.
How effectively the equipment operated compared with its ideal operating rate, including the effect of reduced speed and minor stops.
The proportion of total output that met the required quality standard without being rejected or requiring rework.
OEE is valuable for understanding equipment-related losses. However, an individual machine’s OEE does not independently measure total labour productivity, production-line balance, factory throughput or customer delivery performance.Management needs additional measures to understand the complete operation:
Labour productivity: Accepted output in relation to labour input.
Line balance: Whether workload and process capacity are appropriately distributed.
WIP: The amount of unfinished material within the production system.
Production lead time: The time material takes to move through the manufacturing process.
Finished goods output: What the complete system actually delivers.
Delivery performance: Whether customer requirements are being met.
A machine can have excellent OEE while the factory continues to struggle with finished goods production.
That does not make OEE useless. It means management must understand what it measures—and what it does not.After reviewing the reporting system, Sharad visited the production area with the production manager.
He wanted to observe the actual sequence of operations rather than depend only on reports.
They followed the material from CNC machining to the final operation.
At the CNC section, the machine was producing approximately 60 good parts per hour under the operating conditions being studied.
The next operation was manual deburring and inspection.
The production manager initially estimated that this section could process around 45 parts per hour. Sharad did not accept the estimate without verification. He observed the work sequence, checked cycle times, reviewed output records and discussed the working method with the operators.
The sustainable output of the downstream operation was approximately 45 good parts per hour under those conditions.
Sharad then examined the WIP between the two operations.
The difference was clear:
CNC machining: 60 good parts per hour.
Deburring and inspection: 45 good parts per hour.
Assuming both operations run continuously at these rates, producing the same accepted part, and no additional losses occur, the potential WIP accumulation is:
60 − 45 = 15 parts per hour.This was not merely a reporting issue. It was a production-flow problem.
The production manager explained that management encouraged every machine to achieve maximum production and maintain good OEE.
Sharad responded:
“Improving an individual machine’s output may improve its performance figure. But does it necessarily increase the finished goods output of the complete factory?”
When an upstream process consistently produces more than the next operation can handle:
Unfinished components accumulate between operations.
Floor space is occupied by excess WIP.
Material handling and movement increase.
Operators spend time shifting, arranging and searching for material.
The downstream operation may become overloaded.
Other operations may remain starved because components have not reached them.
Production lead time can increase even when individual machines appear busy.
The fastest machine may look productive, but its additional output may simply become unfinished inventory.
“If we produce more components than the next operation can process, we are not necessarily increasing finished goods. We may only be increasing the quantity of material waiting on the factory floor.”
Sharad explained that line balancing is not simply asking every operator to work faster or forcing every machine to run continuously.
It means aligning the capacity and workload of the production system with the required output.Customer demand and the required production rate.
Available production time and the required takt time.
Actual cycle time of each operation.
Sustainable machine capacity and reliability.
Manpower workload and skill requirements.
Material availability and movement.
Quality requirements and process losses.
Product mix and production sequence.
Suppose the required production rate is 45 good parts per hour.
If CNC machining can sustainably produce 60 parts per hour, there may be no reason to run it continuously at maximum output when the downstream process can handle only 45. Upstream release can be controlled to match the required flow, while maintaining appropriate buffers and ensuring the production plan is met.
But if customer demand requires 60 good parts per hour, the factory must examine how to increase the capacity of the downstream operation.
Possible improvement options include:
Adding a parallel workstation or additional equipment where technically and economically justified.
Redistributing tasks, reviewing operator numbers and improving workload distribution.
Improving work sequence, reducing unnecessary movement and simplifying tasks.
Introducing better tools, fixtures or equipment to improve process capability.
Developing operator capability so that tasks can be performed correctly and consistently.
Introducing suitable automation where it improves the overall production system—not simply the output of one machine.
The right solution depends on the process, demand, investment, quality and manpower requirements.
The objective is to design the complete line around the required output, not to maximise every machine independently.A properly managed production system should support the timely movement of material from one operation to the next.
Material should not accumulate unnecessarily at one process while another waits for work. Workloads should be distributed appropriately, and machines should not remain unnecessarily idle because of poor planning, material shortages or unbalanced processes.
This does not mean eliminating every planned break, maintenance stop or controlled buffer. Some planned downtime and limited buffers are necessary to protect the production system.
The objective is to reduce avoidable waiting, blocking, starvation, excess inventory and unproductive movement.
A production line should be balanced around its required output. The number of machines and operators at each process should be determined by actual workload, cycle time, sustainable capacity and customer demand.
The objective is not maximum production at every workstation. It is the effective delivery of required finished goods, with controlled WIP and balanced workloads.Individual machine OEE is important, but it cannot independently measure the performance of the complete manufacturing system.
Production reports need correct definitions, disciplined recording, physical verification and reconciliation.
Excess output at a faster operation can create unnecessary WIP without increasing final production.
Machine capacity, manpower, workload and material movement must be coordinated to achieve the required production rate.
Recording downtime or a production shortfall is only the first step. Root-cause analysis and effective corrective action are necessary.
Cycle times, machine capabilities, manpower requirements and working methods can change. They must be reviewed against actual operating conditions.
AI and automation can improve monitoring and analysis, but they cannot replace reliable data, process understanding and sound operational decisions.
An implemented improvement should be measured, standardised, communicated to employees and monitored. The factory must continue looking for further opportunities.
The problems in Mr. Seth’s factory are connected to several important manufacturing and operations-management concepts.
Understanding these concepts helps manufacturing leaders move beyond daily production reporting and focus on the complete manufacturing system.
OEE measures equipment effectiveness through Availability, Performance and Quality. It helps identify equipment-related losses and opportunities to improve machine performance.
Labour productivity evaluates accepted output in relation to labour input. It helps management understand manpower productivity, workload distribution and opportunities for skill or work-method improvement.
Line balancing determines how machines, operators and workloads should be allocated across operations to achieve the required production rate and coordinated material flow.
A bottleneck is an operation or resource that restricts the throughput of the complete production system. Improving a non-constraining operation does not necessarily increase finished goods output.
WIP represents unfinished material within the manufacturing process. Excessive WIP can increase storage requirements, material handling, waiting time and production lead time.
Lean manufacturing focuses on improving customer value by reducing waste such as unnecessary movement, waiting, excess inventory, overproduction, defects and unnecessary processing.
Root-cause analysis investigates the underlying reasons behind production losses, breakdowns, delays, defects and performance deviations. Methods such as Why-Why analysis and Fishbone Analysis can support the investigation.
Standard work establishes the approved method for carrying out an operation. Cycle times, work sequences and process requirements must be validated and revised when improvements or operating changes justify it.
Continuous improvement involves reviewing performance, identifying opportunities, implementing changes, measuring results and updating standards to sustain better performance.
Reliable data definitions, disciplined recording, system integration and process understanding are essential foundations for effective digital monitoring, automation and AI-based analysis.
Production plan accuracy.
Production sequencing.
Production targets and the basis on which they are set.
Schedule adherence.
Authorised changes in production plans.
Alignment between production plans and customer requirements.
Approved and actual cycle times.
Standard operating procedures.
Working methods and process sequence.
Actual process capability.
Changes in product, process or operating conditions.
Actual production output.
Machine availability and operating performance.
Downtime and its actual reasons.
Equipment condition and recurring breakdowns.
Whether existing capacity meets current production requirements.
Manpower allocation at each process.
Workload distribution.
Operator skills and training.
Labour productivity.
Operator waiting time.
Opportunities for multi-skilling.
Whether manpower requirements have changed with process improvements.
Number of machines required at each process.
Number of operators required at each process.
Workload distribution and cycle-time differences.
Material flow between workstations.
Machine blocking and starvation.
Excess WIP accumulation.
Opportunities to improve the complete production flow.
Raw material availability.
Material shortages and their causes.
Material consumption.
Inventory accuracy.
Material handling methods.
Internal material movement.
Availability of components at the required operation and time.
Physical WIP quantities.
WIP recorded in the system.
Differences between physical and system stock.
Accumulation between processes.
Material waiting time.
Excessive storage and movement.
Whether WIP levels are controlled according to production requirements.
Rejection, rework and scrap quantities.
Defect categories and reasons.
Inspection methods.
Process capability.
Recurring quality problems.
Corrective actions and their effectiveness.
Machine breakdown history.
Preventive maintenance compliance.
Recurring equipment failures.
Maintenance response time.
Equipment reliability.
Maintenance effectiveness.
Critical spares availability.
Accuracy of production records.
Machine data and downtime classification.
Clear reporting definitions.
Physical and system reconciliation.
Accuracy of material and WIP records.
Traceability of production target revisions.
Reporting discipline.
Whether management decisions are based on verified information.
Safe working methods.
Workplace conditions.
Material handling practices.
Operator comfort and ergonomic risks.
Machine guarding and safety arrangements.
Compliance with applicable safety requirements.
The purpose of an operational audit is not simply to check whether procedures are being followed.
Management should also ask:
Are existing standards still correct?
Can cycle time be reduced without compromising quality or safety?
Can workload be distributed better?
Can manpower be trained to improve productivity?
Can machine capacity and reliability be improved?
Can material movement be simplified?
Can WIP and waiting time be reduced?
Can recurring quality losses be prevented?
Can technology or automation deliver a measurable operational benefit?
If the new method performs better, the approved standard should be revised, employees should be trained, and the improved process should be monitored.
An operational audit should not end with an observation report. It should lead to measurable improvement and sustained performance.Mr. Seth initially believed that his factory needed better technology to understand its production problems.
The investigation revealed a more fundamental requirement.
The factory needed reliable operational information, properly balanced production processes, accurate performance measurement and regular reviews of its existing standards.
Individual machine OEE was important, but it could not tell management the complete story.
A machine running at full capacity may increase its own output without increasing the finished goods output of the factory. Excess WIP, unbalanced workloads, material shortages, inaccurate reporting and unresolved production losses can continue to restrict overall performance.
The real objective is to ensure that every process contributes effectively to the complete manufacturing system.Verify what is happening on the shop floor.
Compare actual performance with approved standards.
Investigate deviations and identify root causes.
Balance machines, manpower and process workloads.
Improve material flow and control WIP.
Evaluate improvements through training, process changes and suitable technology.
Validate improvements and revise standards.
Monitor performance to sustain results.
AI and automation can make monitoring faster, improve visibility and support data-driven analysis. They can help identify patterns and alert management to unusual conditions.
But technology must support a well-understood and properly managed production system. It cannot compensate for unreliable information, incorrect process standards or a failure to understand the actual production constraint.
A factory should not be judged only by how efficiently its machines run. It should be judged by how effectively the complete system converts raw materials into quality finished goods, at the required rate, with controlled WIP, minimum avoidable losses and reliable delivery performance.That is where manufacturing excellence begins.Manufacturing improvement begins with understanding what is actually happening on the shop floor.
At Sampurna Operations Consultancy, the approach is built around three stages:
SEE — Understand actual operating conditions, verify performance information and identify operational losses.
SOLVE — Investigate root causes and develop practical improvements in processes, equipment, manpower, material flow and production control.
SUSTAIN — Validate improvements, update standards, develop employee capability and establish systems to monitor performance.
The objective is to move beyond recommendations towards practical, measurable operational improvements and the management capability required to sustain them.
