How AI Can Help Engineers Perform 5 Why and Pareto Analysis Faster

Most experienced maintenance engineers will adhere to a version of a similar process after a significant machine breakdown which is basically to pull the SAP export, find the records, open the spreadsheets, cross-reference the maintenance logs and try to piece together what actually happened and more importantly why it happened.

The tools that engineers use during this process have not changed much in decades and they still use the same methodologies like the 5 Why analysis and the Pareto chart. These are both proven methodologies that have helped manufacturers understand and solve operational problems for generations. The issue is not with the methods themselves but with everything engineers have to do before they can actually use them.

What is 5 Why Analysis?

The 5 Why technique was developed at Toyota as part of the Toyota Production System and has been one of the most widely used problem solving methods in manufacturing worldwide. What it does is instead of just looking at the visible symptom, you ask "why" repeatedly typically five times until you reach the underlying cause that, if addressed, prevents the problem from recurring (Fiix, 2026; FlowFuse, 2025).

Let us look at a simple example. A motor overheats and stops the line. Why did it overheat? The cooling fan failed. Why did the fan fail? Dust blocked the air intake. Why was there a dust blockage? The cleaning schedule was not followed. Why was the schedule not followed? There was no system to track or alert on it. That fifth answer is where the real fix lives. It is not by replacing the motor, but in building a maintenance process that prevents the blockage.

The power of 5 Why is that it forces engineers to think past the broken component. But the challenge is to gather accurate, complete maintenance history in order to perform the five why. It is necessary to know how many times the failure has occurred, what was done each time, whether the fix held, and what conditions were present. Gathering all of this information from disconnected records like SAP exports, paper logs, maintenance notes in different formats is heavily time consuming and inefficient.

What is Pareto Analysis?

The Pareto chart is designed to address a different but equally important question: out of all the failures happening across the factory, which ones deserve the most attention first?

The underlying principle was developed by quality pioneer Joseph Juran, drawing on Vilfredo Pareto's economics work. His concept was that roughly 80% of effects come from 20% of causes (Tractian, 2025; MDCplus, 2025). In manufacturing, a small number of failure categories typically account for the majority of total downtime hours. Addressing those few significant causes will provide far more operational improvement than putting effort evenly across every breakdown on the list.

A Pareto chart makes this visible by ranking failures caused by their cumulative impact of total downtime hours and not just frequency so that the engineers know which problems to investigate first. Simply, by Pareto theory, the machine that breaks down once for eight hours matters more than the one that trips a sensor five times for ten minutes each.

The difficulty here lies in the data preparation because building a Pareto chart manually requires the engineers to extract downtime records, clean and categorize them, calculate cumulative totals, and build the chart in Excel. A thorough analysis consumes most of a long time and under production pressure, it often gets shortened, delayed, or skipped.

The Real Problem of Scattered Data

Both methods depend on data being accessible and organised before the analysis begins which is rare in most factories in reality. Operational data is scattered across multiple systems and formats: SAP work orders in one place, Excel downtime logs in another, paper-based maintenance reports in a binder, machine manuals in a shared drive nobody has reorganised since the system was installed. When an engineer needs to perform a proper 5 Why investigation or build a meaningful Pareto chart, they first need to collect the data across these scattered systems and formats, retrieving, translating, and manually consolidating information from sources that were never synced.

Research from the manufacturing sector shows that manual root-cause analysis for a single significant failure can cost close to USD 50,000 in engineering time and investigation effort alone (Acerta AI, 2025). AI-powered systems, by contrast, can correlate data across hundreds of process parameters simultaneously, compressing investigations that previously took days into minutes (QualityLine, 2026; Datagrid, 2025). That is a significant improvement in efficiency. It could be the difference between an analysis that happens after the next breakdown and one that happens before it.

How AI Handles the Data Layer

The methodology of 5 Why and Pareto analysis are proven. What needs to improve is the time and effort required to assemble the data those methods depend on.

AI-assisted maintenance analysis is the solution for this operational gap. It connects the sources engineers are currently consulting one by one such as maintenance records, sensor data, historical downtime logs, component usage history, SOPs etc. and does the pattern recognition and categorisation work automatically. An AI system can classify downtime causes at over 85% accuracy from existing work order and sensor data, updating Pareto rankings in real time as new events are logged rather than requiring end-of-week manual compilation (Oxmaint, 2026).

For 5 Why investigations, the relevant maintenance history is assembled in one place for the engineer to start the investigation and not spread across three systems that need to be manually cross-referenced. The engineer still applies their judgment and domain knowledge to the analysis but they do not have to spend a long time retrieving and organising the evidence.

McKinsey's research on predictive maintenance consistently shows that this kind of connected data approach reduces unplanned downtime by up to 50% and cuts maintenance costs by 18 to 25% (McKinsey, 2020).

Need for Platforms Like Protonest Connect

Platforms like Protonest Connect are built to bridge this operational gap that Protonest Connect is addressing. The backend connects sensor data from power monitoring, maintenance records, machine manuals, SOPs, KPI information, and inventory history, bringing the sources that currently sit disconnected across Sri Lankan manufacturing operations into a single accessible environment.

Beyond the real-time dashboards and alerts, the intelligent analysis layer is designed to answer the questions engineers are already asking, but currently have to spend hours researching manually such as why is this failure repeating, what components were used in the last repairs, what spare parts should be kept in stock based on history, what actions could improve the current KPI etc. Especially for Pareto analysis, instead of a maintenance engineer spending several hours extracting, cleaning, and charting a month's worth of downtime data, the analysis is generated from connected operational records and immediately shows which failure categories are driving the most production loss. The engineer's job becomes reviewing and acting on the insight, rather than producing the analysis.

Solving the Real Problem

5 Why and Pareto analysis still remain among the most effective problem solving methods available to maintenance teams. What limits their impact in most factories is not the methodology but the conditions under which they are applied: incomplete data, time pressure, and systems that were never designed to share information with each other.

AI-assisted maintenance intelligence does not replace the engineer, it simply removes the manual overhead that makes rigorous analysis impractical under normal production conditions. When the data layer is handled automatically, engineers can apply their expertise to the part of the process that actually requires it which is understanding the cause, making the decision, and stopping the same failure from coming back.

References

1. Fiix Software, 5 Whys root cause analysis in manufacturing: https://fiixsoftware.com/blog/5-whys-simple-root-cause-analysis/
2. FlowFuse, Five Whys methodology guide: https://flowfuse.com/blog/2025/12/five-whys-root-cause-analysis-definition-examples/
3. Tractian, Pareto Principle in industrial maintenance (Juran / 80/20 background): https://tractian.com/en/blog/pareto-chart-how-to-apply-and-eliminate-equipment-failures
4. MDCplus, Pareto Principle in manufacturing downtime: https://mdcplus.fi/blog/downtime-pareto-principle-free-template/
5. Acerta AI, manual RCA cost (~USD 50,000 per failure) and automated RCA time savings: https://acerta.ai/blog/six-reasons-why-you-need-automated-root-cause-analysis/
6. QualityLine, AI-powered RCA correlates 100+ parameters, reducing investigation from days to minutes: https://quality-line.com/ai-root-cause-analysis/
7. Datagrid, AI agents reducing root cause investigation from weeks to minutes: https://datagrid.com/blog/ai-agents-automate-manufacturing-root-cause-analysis-quality-assurance-directors
8. Oxmaint, AI Pareto analysis with 85%+ downtime classification accuracy: https://oxmaint.com/industries/manufacturing-plant/manufacturing-downtime-tracking-software-analysis-root-cause
9. McKinsey & Company, predictive maintenance downtime and cost reductions. Via Com4: https://www.com4.no/en/blog/predictive-maintenance-how-to-use-iot-to-reduce-downtime-and-costs
10. Protonest Connect: https://www.linkedin.com/company/protonest-connect/