Find a process cause · printable packet

A process cause is a condition in how the work is set up that makes a failure possible for anyone; ask why until it can be changed, draw what led to what, and compare failure rates with and without the condition.

Paper packet. Every task here also exists on screen, where it is checked automatically; answers written on paper are not assessed by Nydus. When you are back at a device, enter your answers there.

1. What you will learn

You will tell a process cause from blame, trace a failure back through 'why' to a condition the business can change, draw the chain of causes, and compare failure rates with and without a suspected condition.

2. What you already have

Complaints counted by kind point to the process step where failures start. This lesson goes one level further: what, in that step, makes the failure possible, and how to tell a real cause from someone to blame.

3. Words for this lesson

4. Keep asking why until the answer is something to change

Neighborhood Kitchen sent five vegetarian boxes with meat in them this month. The first explanation offered was 'one of the packers is careless'. But three different people packed those boxes. When different people make the same mistake, the mistake is being made easy for them.

Ask why, and keep asking, until the answer is something the business can change:

The last answer is a condition. Colored lids and a label check against the order sheet remove it, for every packer, including the ones who have not been hired yet. Blaming the packer would have changed nothing, and the next packer would have made the same mistake.

Term	What it means
Process cause	A condition in how the work is set up that makes a failure possible for anyone.
Condition	Something set up, missing or unclear: a missing check, two look-alike boxes.
Blame	Attributing the failure to a person's character.
Asking why	Tracing a failure back, answer by answer, to a condition.
Cause chain	A drawing of what led to what, ending at the failure.
Comparison	The failure rate with the condition present and absent.

Another way: table

Blame against cause.

Blame says	The process cause
The packer is careless	Boxes look identical; no label check
The mechanic rushed	No torque setting for the cable bolt
Priya is forgetful	No key safe in the van; key not on the checklist

Another way: steps

State the failure exactly.
Ask why, and write the answer.
Ask why of that answer, and keep going.
Stop when the answer is a condition the business can change.
Draw the chain of what led to what.
Test the suspected condition by comparing failure rates.

5. Test the suspected cause with a comparison

A suspected cause is still a guess. Test it by comparing how often the failure happens with the condition present and without it. Neighborhood Kitchen suspected the river bridge behind its late deliveries: of 25 deliveries across the bridge, 10 were late, 40 percent; of 25 that did not cross it, 2 were, 8 percent. The failure goes with the condition, so scheduling bridge deliveries first is worth trying.

If the rates had been about the same, the bridge would not be the cause, and the search would go on. A comparison also protects people: it is much harder to blame a driver when the numbers show the lateness follows the route, whoever drives it.

6. The method, step by step, and how to check it

State the failure exactly. Not 'deliveries are bad' but 'lunch arrived after the promised time'. A vague failure produces vague causes.

Ask why, and write each answer down. Keep going while the answer is about what happened; stop when it is about how the work is set up, and the business could change it.

Draw the chain. Put each condition in a box and draw an arrow from each to what it led to, ending at the failure. Often two arrows meet: two conditions together made the failure possible, and each needs a fix.

Test it. Count the failure rate on jobs with the condition and jobs without it, over the same period.

Check the reasoning. Read the chain backwards from the failure: does each arrow make sense as 'this happened because of that'? Ask whether the cause would still be there with a different person doing the job; if not, the chain has stopped at a person, not a condition. And check that the comparison groups are alike in every way except the condition.

7. Why each step is allowed

Asking why repeatedly is allowed because each answer is itself an event with its own reason. The first answer, 'the wrong box went in the bag', is true but cannot be fixed directly; the deeper answers can.

Stopping at a condition, rather than a person, is needed because a condition affects everyone who does the job. Removing it prevents the failure for every future packer; replacing the packer does not.

Comparing rates with and without the condition is allowed because if the condition really causes failures, they should be more common when it is present. If the rates are the same, the condition is not doing the work the story says it is, however convincing the story sounds.

8. When several conditions work together

Many failures need more than one condition. Maya's cracked cups needed two: cups packed warm, which made hairline cracks likely, and no crack check before packing, which let them through. Either fix alone would help; both together close the gap.

Drawing the chain shows this. When two arrows lead into the same box, ask whether removing either would have prevented the failure. Fix the cheaper one first if it is enough, and keep the other in mind if the failures continue. A check is often the cheapest fix and the one to add first; removing the underlying condition is the better long-term answer.

9. Fair comparisons

A comparison is only fair if the two groups differ in the suspected condition and little else. Comparing bridge deliveries made on rainy Mondays with other deliveries made on dry Wednesdays mixes two conditions, the bridge and the weather, and cannot say which mattered.

Take both groups from the same weeks, the same staff and the same kinds of job. Where possible, change only the condition: pack half of a firing's cups warm and half cooled, and compare. Small counts jump about by chance, so look for a large gap, or repeat the comparison before acting on a small one.

10. Talking about causes with staff

Root-cause work goes badly if staff think it is a hunt for someone to blame. Say at the start that the question is what made the failure easy, not who did it, and then behave that way: when the chain stops at a person, ask what about the job set them up to make that mistake.

Ask the people who do the work. The packer knows the boxes look the same; the driver knows which bridge backs up at noon. They often see the condition straight away, and they will tell you if they trust that the answer will change the process rather than their standing.

11. From cause to fix

Once the condition is found, the fix usually takes one of four forms. Remove it: cups cool overnight before packing. Make the mistake impossible: colored lids for vegetarian boxes. Add a check where the failure would show: a label check before sealing. Write it into standard work: the torque setting for the cable bolt.

Removing the condition or making the mistake impossible are the strongest fixes, because they do not depend on anyone remembering. A check catches what still slips through. A reminder alone, such as a sign saying 'be careful with vegetarian orders', is the weakest, because it changes nothing about the work.

12. Checking that the fix worked

Count the failure again after the change, the same way as before. Neighborhood Kitchen's late rate on bridge deliveries should fall toward the rate on other routes once those deliveries go first. If it does, the cause was right and the fix works.

If it does not, the search goes back to the chain. Perhaps the bridge was only part of it, and the real condition was that bridge deliveries were also the largest orders, which took longest to pack. Treat every fix as a test of the cause, and let the counts decide.

13. Keeping a cause log

Write each root-cause search on one page or one row: the failure, the chain of whys, the condition found, the comparison counts, the fix and the counts afterwards. It takes ten minutes and pays back many times.

Over a year the log shows which kinds of condition keep turning up in this business: look-alike items, steps missing from standard work, handoffs with nothing traveling with the job. Once a pattern is visible, the owner can look for it before it causes a failure, checking a new product for look-alikes, or a new process map for handoffs with nothing attached. That is the point where root-cause work stops being a response to failures and starts preventing them.

14. When the cause is outside the business

Sometimes the chain ends at a condition the business does not control: a supplier's late deliveries, a bridge that backs up, a client who is never home at the booked time. The method still helps, because it shows where to work around the condition. Bridge deliveries can leave first. A second supplier can be lined up, as the supplier lessons later in this course show. A confirmation text the evening before can catch the client who will not be home.

The question changes from 'how do we remove this condition?' to 'how do we stop it causing this failure?', and the answer is still a change to the process.

15. In the world: a Phoenix pharmacy and its look-alike bottles

An independent pharmacy in Phoenix had three dispensing errors in two months, each caught by the pharmacist's final check but alarming all the same. Each time a technician had picked the wrong strength of the same drug. The first reaction was to retrain the technician involved, but the three errors involved three different technicians.

Asking why led quickly to a condition. The two strengths came in bottles of the same size and color, shelved side by side in alphabetical order. Anyone reaching quickly for one could pick up the other.

The pharmacy separated look-alike strengths onto different shelves, added a colored 'check strength' tag to the higher one, and required a barcode scan at filling. Over the next six months, wrong-strength picks caught at the final check fell from about one every three weeks to none. Nobody was blamed, and the technicians suggested two more look-alike pairs to separate.

16. In the world: root-cause analysis in health care

Hospitals reviewed by The Joint Commission must carry out a root-cause analysis after certain serious events, looking for the system conditions that allowed the event rather than an individual to blame. The method scaled down, with 'why' asked until the answer is a condition, works the same way in a small business.

17. Where this goes wrong

Find the person responsible. A condition that trips up anyone will trip up the next person too.

One 'why' is enough. The first answer is usually a symptom; keep asking until it is something to change.

A convincing story is a cause. Compare failure rates with and without the condition.

18. Monica runs out of change

19. Bright Home Cleaning's keys

20. Neighborhood Kitchen's late deliveries

21. Your turn: Northside Repairs' loose cable

22. Guided practice

Neighborhood Kitchen has sent $3$ vegetarian boxes with meat in them this month, packed by three different people. Which explanation points to a process cause?

23. Guided practice

Complete the worked solution: Maya suspects cups crack when they are packed straight from the kiln while still warm. Of $50$ cups packed warm, $9$ cracked; of $50$ cups left to cool overnight first, $2$ cracked. Find the crack rate for each group and the gap between them.

24. Guided practice

A Northside Repairs customer's brake cable came loose after pickup. Dev asks 'why?' at each step. Put the answers in the order they come.

25. Guided practice

Bright Home Cleaning's notes on why keys were left in $5$ client doors this month. Mark every note that describes a condition in the process.

26. Practice

Maya's cups keep arriving at cafés cracked. Draw what led to what, using 'leads to'.

27. Practice

Neighborhood Kitchen found that $32$ percent of its deliveries across the river bridge were late, against $16$ percent of those that did not cross it. It makes $75$ bridge deliveries a month. If serving the bridge first makes those deliveries as punctual as the rest, how many late deliveries a month will it avoid? Fill in each figure on the sheet.

28. Practice

A dental lab in Boston makes crowns and suspects that remakes come from impressions mailed in rather than scanned. Of $20$ crowns made from mailed impressions, $7$ had to be remade; of $20$ made from digital scans, $1$ did. By how many percentage points is the remake rate higher for mailed impressions?

29. Somewhere new

A bakery suspects its cakes sink when the oven door is opened in the first ten minutes. Of $20$ cakes where the door was opened early, $9$ sank; of $20$ where it was not, $0$ sank. Complete the sentence.

30. Lesson test

Lesson test: one question per skill, one attempt each, no hints. Your answers are checked when you submit.

31. Test question

Neighborhood Kitchen checked whether late deliveries depend on the route. Of $25$ deliveries that went across the river bridge, $12$ were late; of $25$ that did not, $1$ were. Fill in the late rate for each, as a percentage.

32. What you can do now

You can find the condition behind a repeated failure instead of the person who was there. Tell someone why three people making the same mistake points to the process. Next: how much a process can deliver in an hour.

Working for the steps left to you

	Amount
Gap in late rates, percentage points
Late deliveries avoided a month

	Late, percent
Across the bridge
Not across the bridge

$\text{torque setting in the standard work; a torque wrench}$