What is loss analysis and how do you use it in production?

This guide explains how loss analysis works in practice, which questions it should be able to answer, the pitfalls that make the analysis misleading, and how factories have used it to reduce downtime by over 70%.

Why is the OEE figure not enough without a loss analysis?

The OEE figure alone is not enough, as it only shows that you have a fever without indicating which diagnosis is causing it. Similarly, an OEE of 58% shows that 42% of planned time did not become value-adding production, but it says nothing about the cause: breakdowns, changeovers, material shortages, speed losses, or scrap. Without that knowledge, it is impossible to choose the right countermeasure, and improvement work turns into guesswork.

Loss analysis breaks down the gap. It is based on a loss model where every lost hour is assigned an address: a category, a cause, a location, and a timestamp. The foundation lies in the TPM tradition's six big losses: breakdowns, setup and adjustments, idling and minor stops, reduced speed, defects in normal production, and startup defects. Added to this are micro-stops—short interruptions lasting seconds that are insignificant on their own, but collectively can account for a large portion of performance losses.

An important insight alters how the analysis must be structured: data from factories that have worked with OEE over long periods show that around 80% of downtime causes are not machine-related. They relate to materials, planning, waiting for orders, communication, and staffing. This means automated machine data is never sufficient for a comprehensive analysis. Operators' coding of reasons is what provides the losses with their addresses.

Which questions should a loss analysis be able to answer?

A good loss analysis should be able to answer specific prioritisation questions within minutes, without requiring specialist intervention. The key questions are:

What are our three largest losses in hours and money? The fundamental question. The answer directs where improvement resources should be deployed.

Which losses are frequent but short, and which are rare but costly? The two types require different countermeasures. A hundred micro-stops a day are tackled differently than one breakdown a month.

Why does the same part number perform differently across two different lines? The difference between lines often holds a ready-made solution: the better line's way of working can be standardised across the poorer-performing line.

Which downtime reasons have increased over the past month? Trends catch problems that are on the verge of becoming expensive before they actually do.

Which technical issues recur on the same machine component? Recurring faults on the same component represent a maintenance pattern pointing toward the root cause, rather than symptoms.

How does performance vary between shifts? Differences between shifts rarely come down to effort; they typically relate to working methods—and working methods can be standardised.

At Sibbhultsverken, this type of breakdown made all the difference. By systematically analysing and addressing the largest loss categories, they reduced technical stops by 73% and unplanned stops by 63% in just 12 months. OEE rose by 19.4% in one cell and 40% in another, whilst net production increased by 15% without adding shifts.

What is required for loss data to be reliable?

Loss data is trustworthy when three elements are in place: consistent definitions, simple coding for operators, and visibility for poor-quality data.

Consistent definitions. Planned production time, changeovers, planned maintenance, and ideal cycle times must be handled consistently over time and across lines. Otherwise, the figures cannot be compared and the analysis rests on shaky ground. A documented OEE policy, where the team has defined what is measured and how, is the simplest safeguard.

Simple coding. If it takes longer to code a stop than the stop itself lasts, the coding simply will not happen. The shop floor interface must allow coding in seconds, featuring clear reason codes designed in collaboration with the operators themselves. Orkla Nidar eliminated the "Other" category from their reporting once operators were given tools that made it easy to code the correct cause immediately. A growing "Other" category is otherwise the surest sign that the analysis is losing its integrity.

Visible data quality. The system itself must highlight where data falls short: uncoded stops, out-of-schedule activity, and unrealistic speed patterns. This allows shortcomings to be resolved continuously, rather than being discovered after someone has already made decisions based on flawed data.

How do you go from analysis to action?

Moving from analysis to action follows prioritisation, root cause, and follow-up in a repeating cycle. Analysis in itself improves nothing. Value is created when it directs action.

A proven workflow looks like this:

Prioritise by value. Select the two or three losses that cost the most, not the ten that are merely noticeable. Doing a few things with dedicated resources beats tackling many with none.

Identify the root cause. Start preferably with a fishbone diagram (known as Ishikawa in Lean) to brainstorm and identify causes. All suggestions are welcome. Plausible causes are then prioritised, and the most likely ones are examined using the "Five Whys" technique to find underlying issues. This method generates ideas and narrows the focus to the most probable causes to uncover the root cause. 

For complex issues, a structured DMAIC project may be needed, as seen at Bostik, where analysing changeover data led to a 70% reduction in changeover times.

Implement actions with an owner and target date. An action without an owner remains indefinitely on the list.

Follow up in the metrics. The addressed loss should disappear or decrease noticeably in the data. If it does not, the problem was not solved at its root and another cycle is required. Follow-up takes place in the same system as the analysis, making it virtually effortless.

Barilla Wasa in Filipstad illustrates the complete picture. Through systematic work on losses, capacity utilisation rose from 75% to 94%, downtime dropped by 22%, product waste fell by 15%, and net production increased by 15%. At the same time, CO2 consumption decreased by 28%, as lost machine time also equals wasted energy. Loss analysis supported the entire chain, from prioritisation to verified results.

What are the most common pitfalls in loss analysis?

The most common pitfalls are analysing too infrequently, getting hung up on the aggregate figure, comparing the wrong things, and overlooking minor losses.

Quarterly analysis. Problems grow expensive before they are discovered. Loss analysis belongs in the weekly cadence, with top losses serving as a standing agenda item in continuous improvement meetings.

Fixating on overall OEE. The aggregate figure moves slowly and masks conflicting movements underneath when some causes decrease whilst others rise. A line can improve changeovers whilst micro-stops climb, leaving the overall figure stagnant. Analyse the components, not just the total.

Comparing lines and plants directly. Differing definitions and operating conditions make direct comparisons misleading. Compare each unit against its own history, and use discrepancies between units as a prompt for questions rather than a scorecard.

Ignoring micro-stops. Individually they are barely noticeable and are not captured manually, yet combined they can represent one of the largest loss categories. Automated monitoring is the only way to visualise them.

How does Good Solutions work with loss analysis?

Loss analysis is the core of the platform from Good Solutions, built on a proprietary loss model that gives every lost hour an address. Reports break down losses by downtime reason, reason group, line, part number, shift, and trend over time. The timeline visualises the 24-hour cycle, displaying downtime, volume, and OEE in the same view, with stops coded directly within it. The operator interface makes coding fast enough to be performed in the moment—an essential prerequisite for the analysis to cover the roughly 80% of downtime causes that machine data cannot explain.

Working practices support the analysis. Operational implementation ensures definitions and data quality right from the start. The OEE policy workshop documents what is measured and how, ensuring figures remain comparable. Training provides operators and superusers with the shared understanding required to keep coding consistent. And because the platform is designed to drive continuous improvement, not merely measure, the analysis flows directly into prioritisation, action, and follow-up within the same tool.

Customer results are achieved through this exact path. Sibbhultsverken reduced technical downtime by 73%. Kopparbergs cut quality deviations by 68%. In every case, the work started with loss analysis and finished with proven results.

Frequently Asked Questions

What is the difference between OEE and loss analysis?
OEE is the metric; loss analysis is the explanation. OEE reveals what proportion of planned time became value-adding production. Loss analysis breaks down the remainder by cause, location, and time, ensuring improvement initiatives are correctly targeted.

Which loss categories should we use?
Start with the six big losses, treating micro-stops as an independent category, and adapt the reason codes to your operations in collaboration with operators. Codes should be few enough to select quickly and distinct enough to prevent confusion. A growing "Other" category signals that the classification requires review.

How often should we perform a loss analysis?
Continuously. Top losses should be a standing item in weekly improvement meetings, and deviations should be visible in daily operational reviews. Quarterly reviews are too infrequent, as issues can become costly between sessions.

Can we conduct loss analysis in Excel?
On a small scale, as a starting point, yes. However, manual capture misses micro-stops, coding remains incomplete, and analytical work consumes time that should be spent taking action. An OEE system captures data automatically, maintains consistency, and delivers granular breakdowns in minutes.

How do we know the analysis is pointing us in the right direction?
Through follow-up. When a prioritised loss is addressed, it should decrease noticeably in the data. If it does, the analysis was accurate. If the countermeasure addressed a symptom rather than the root cause, repeat the analysis. In that case, the analysis must go deeper, preferably leveraging operator insight alongside machine data.