A downtime Pareto chart is a bar chart of stop causes, sorted from the most lost minutes to the least, with a line showing the running total. You sum minutes by reason code, sort, add the cumulative percent and fix the first two or three bars.
What a downtime Pareto chart shows
Every line has many causes of downtime. A few of them cause most of the lost time.
The Pareto chart makes those few visible. Bars go from biggest to smallest. A line on top adds them up, so you can see when you reach 80% of all lost minutes.
It answers one question: where do we start? Without it, the loudest problem of the week gets the work.
Worked example
Example: a packaging line logs 520 minutes of stops in a month, using reason codes. Here they are ranked.
The two overruns are the minutes past the standard changeover and cleaning time. They are still planned stops (STO), so keep them marked as planned.
Total: 150 + 110 + 90 + 70 + 50 + 30 + 20 = 520 minutes.
The first four causes add up to 420 minutes, which is 420 / 520 = 80.8%. Four causes out of seven give four fifths of the loss.
The four teal bars are the "vital few". The grey bars can wait.
The table behind the chart
| Cause | Minutes | Share | Cumulative |
|---|---|---|---|
| Jam at labeller | 150 | 28.8% | 28.8% |
| Waiting for packaging material | 110 | 21.2% | 50.0% |
| Changeover overrun | 90 | 17.3% | 67.3% |
| Filler fault | 70 | 13.5% | 80.8% |
| Waiting for an operator | 50 | 9.6% | 90.4% |
| Metal detector rejects | 30 | 5.8% | 96.2% |
| Cleaning overrun | 20 | 3.8% | 100.0% |
Share = minutes / 520. Cumulative = running total of the shares.
Build this table first. The chart is only a picture of it.
How to build it
- 1Pick one asset and one period. Start with the bottleneck. A month is a good first period, because a single week can be odd.
- 2Use coded data. Every stop needs a reason code. Do not use "other".
- 3Sum the minutes by code. Add all stops of the same reason.
- 4Sort from largest to smallest. Biggest on the left, or on top in a horizontal chart.
- 5Add share and cumulative share. Each row's share of the total, then the running total.
- 6Draw the bars and the cumulative line. In a spreadsheet, a combo chart does it. Excel (2016 and later) has a built-in Pareto chart type.
- 7Choose the first two or three bars. Hansen advises no more than three key projects at a time.
- 8Drill into the top bar. Draw a second Pareto for the causes inside it.
The drill-down is described in Overall Equipment Effectiveness (R. C. Hansen, 2001). Rank downtime minutes by loss category first, then within the top categories by component, frequency, failure type and cause. Fix the top two or three, update the chart and repeat.
Pareto by minutes is not enough
A chart by minutes favors long stops. A chart by count favors short, frequent ones. You need both, because the fix is different.
Example: the same 520 minutes, counted by stops.
| Cause | Stops | Avg min | Total min |
|---|---|---|---|
| Jam at labeller | 60 | 2.5 | 150 |
| Metal detector rejects | 30 | 1.0 | 30 |
| Filler fault | 14 | 5.0 | 70 |
| Waiting for an operator | 10 | 5.0 | 50 |
| Waiting for packaging material | 5 | 22.0 | 110 |
| Changeover overrun | 3 | 30.0 | 90 |
| Cleaning overrun | 2 | 10.0 | 20 |
| Total | 124 | 520 |
Ranked by count, the metal detector is number two with 30 stops, but it is only 30 minutes. Material waiting is number five by count and number two by minutes.
Two different fixes follow:
- Many short stops (labeller jams, rejects) need a root cause fix at the machine. See micro-stops and mean time between stops.
- Few long stops (material, changeover) need planning, staging and a faster method. See SMED changeover reduction.
Count is also the weak spot of hand logging. If the labeller jams are 30-second stops that operators never write down, they will be missing from both charts. See operator downtime logging.
Common mistakes
- Mixing planned and unplanned. A planned changeover and a breakdown do not have the same owner. Chart them apart, or mark them. See planned vs unplanned downtime.
- A huge "other" bar. It means the codes are bad. Fix the list, not the chart.
- Too many bars. Show the top 7 to 10 causes. Group the tail only in the picture, not in the data.
- One day of data. A single shift is anecdote. Use enough data that one odd event does not set the order.
- Treating 80/20 as a law. It is a pattern, not a rule. Your chart may be flatter.
- Never redrawing. After you fix the top bar, the order changes. Redraw every month.
From chart to action
A Pareto chart does not fix anything. It tells you where to put your team's time.
Take the top bar. Ask why until you reach a cause you can change.
The root cause analysis guide shows how. Then give the fix an owner and a date.
To turn minutes into money, use the downtime cost calculator. A short stop that happens 60 times a month may cost more than a long one that happens once.
A Pareto is only as good as the stop data behind it. Fabrico takes stops from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line, so the ranking is ready per line and shift. See how Fabrico builds the ranking for you.
What people ask next
What is the 80/20 rule in downtime?
It is the pattern that a small number of causes give most of the lost time. In the example, four of seven causes gave 80.8% of the minutes. Your chart may be more or less steep, so read it, do not assume it.
Should I rank by minutes or by number of stops?
Use both. Minutes show the size of the loss, and count shows how often it happens and points to a different fix. Overall Equipment Effectiveness (R. C. Hansen, 2001) suggests looking at frequency inside the top loss categories.
How much data do I need for a Pareto chart?
Enough that one unusual event does not change the order. A month on one line is a good start. Use less only if the line has few stops.
Can I build a Pareto chart in Excel?
Yes. Sum minutes by cause, sort them, add a cumulative column and make a combo chart. Newer Excel versions also have a ready Pareto chart type. A downtime log template gives you the data in the right shape.
What do I do after the Pareto?
Pick the top two or three causes, find the root cause of each and fix them. Then update the chart and repeat. The reason code list in downtime reason codes keeps the data clean.
Related guides: Machine Downtime Reasons: 12 Causes to Check