Key takeaways
See our roundup of RCA software built around this kind of analysis.
The Pareto principle, most of the impact comes from a small number of causes, is a real and useful observation about how losses distribute in operational systems. The Pareto chart is a clean visual of it. Both have been part of maintenance practice for decades.
The misreads come from over-trusting the visual. A chart with five bars where the top two are clearly tallest does look like 80/20. The mistake is concluding that fixing the top two would recover 80% of the value. That conclusion is only true if the bars are measuring the right thing. Most Pareto charts in maintenance settings are not.
The most common Pareto on the wall is "failure count by cause." The top bar is whatever cause produced the most work orders last month. The team treats it as the top priority. This is often wrong.
The right Pareto is failure cost by cause: the sum of production-minute impact (or labour-hours, or parts cost, or some weighted blend) per cause. A cause that fires 40 times a year and costs 5 minutes each event is a 200-minute problem.
A cause that fires 4 times a year and costs 90 minutes each event is a 360-minute problem. The second is bigger and almost never tops the count chart.
For many asset classes, the cost Pareto and the count Pareto disagree on the top causes. Acting on the count chart sends the team after the loud-but-cheap causes; acting on the cost chart sends them after the quiet-but-expensive ones. The article on manufacturing KPIs covers the cost-per-minute attribution that the cost Pareto depends on.
A Pareto from last quarter, posted on the wall, used all year. This is the same anti-pattern as the unmaintained min/max levels in spare parts: the chart was right when it was built and gets less right every week.
Underlying causes shift. A failure mode that was the top contributor in Q1 gets fixed by April and drops out of the chart. A new failure mode that started in Q2 is not yet on it. The team continues to act on Q1's priorities while Q2's are accumulating cost.
The fix is monthly rebuild with period-over-period comparison; the same chart over four months tells you whether your interventions are moving the bars or not. The piece on root cause analysis covers the trend-vs-snapshot framing in adjacent terms.
The most subtle misread. A Pareto chart shows up in the monthly review, the team looks at the top bar, someone says "we should fix that," the meeting moves on. No specific action gets named, no owner assigned, the chart appears again next month with the same top bar.
A working Pareto arrives with a proposed action against the top one or two bars: "the top cause is bearing failure on the packer; the proposed action is to tighten the PM interval from 90 to 60 days for the next quarter." The chart is the evidence; the action is the conversation.
Reviews that look at Paretos without proposed actions produce diagnoses, not interventions. The article on the work order management system covers how the proposed action becomes a standing rule rather than a one-off.
One useful technique for catching misreads: compare the count Pareto and the cost Pareto side by side. The diagonal is the diagnosis:
Most plants find that the mid-of-count-top-of-cost cell is where the highest-leverage interventions live. It is also the cell the team almost never works on without the cost Pareto being on the wall.
Most charts should be cost, not count. Cost can be production-minute impact, labour hours, parts spend, or a weighted composite, but it should be a measure of consequence, not occurrence.
Monthly is usually right. Weekly is too noisy for trend signal; quarterly is too slow to catch shifts.
More than 8 bars and the chart loses readability. Group the long tail into "other" and revisit the boundary quarterly.
Side-by-side with last month. The story is in the change, not just the levels.
The top one or two bars get a one-line proposed action. The chart is now a decision input, not a status report.
The piece on the preventive maintenance schedule covers how Pareto-driven interventions get codified into PM cadence changes.
The cost Pareto depends on production-minute attribution at the failure-event level, which is hard in plants where OEE events and work orders live in separate systems. Where a unified OEE + CMMS platform helps is that every work order carries the OEE-event link, so cost-per-cause is a query rather than a quarterly reconciliation.
Fabrico is built so the monthly Pareto rebuild takes minutes, not days, and the count-vs-cost side-by-side is a default view. To see what the right Pareto looks like for your top asset classes, book a demo .
If the top bar is less than 1.5x the median bar, the distribution is roughly uniform and a Pareto adds little. Use a ranked table instead and look for the cluster of causes that share characteristics.
For broad situational awareness, yes. For driving action, run it per asset class. Plant-wide aggregates hide the per-class story that the action needs.
The Pareto bars map to failure modes from the catalogue. Without a stable catalogue, the bars drift as people categorise the same failure differently. The catalogue is what makes the period-over-period comparison meaningful.
Useful complement, not replacement. The asset Pareto tells you which assets to look at first; the cause Pareto tells you what to fix when you get there. Most plants benefit from both, used together.
Treating the count Pareto as the cost Pareto because it is easier to compute. The cost calculation requires data the team may not have wired up yet; doing the count chart anyway is fine as a starting point, but it should be labelled as count, not as priority. Mislabelling is what produces the wrong actions.
Login
Blog
