Quality management checklist review card with twelve questions and wrong answers. Working through the quality management checklist in order
Image: Operations Process Control

Costs

Part of Making sense of quality management, 2027 edition

Working through the quality management checklist in order

Quality management review in twelve questions, each with the answer that should worry you, plus how to build a checklist people use instead of sign off.

A checklist is not a list of good intentions. It is a short set of questions whose wrong answers you have decided in advance to act on, used at a moment you have decided in advance.

Two things follow. The questions below are worth asking about your own operation once a quarter, and the way you run them matters as much as the questions.

What to take away

  • A checklist item earns its place only if a wrong answer changes what you do. Everything else is decoration that trains people to tick without reading.
  • Ask at a moment, not in general. A check with no trigger gets done when someone is free, which is when nothing is happening.
  • The wrong answers below are more useful than the questions. Read them first.
  • Checklists rot toward length. The maintenance job is deleting items, and nobody is ever assigned to it.

How to build one people actually use

Two shapes exist and they are not interchangeable.

Three drafting rules

  • Keep it to what fits on one screen
  • Give it a trigger, an explicit moment
  • Name who calls it

Read and do vs do and confirm

Read and do

When used
unfamiliar task
Order matters
yes
Risk
missed step
How run
perform as read
Common error
written as do-confirm

Do and confirm

When used
competent performer
Order matters
no
Risk
forgetting under pressure
How run
run at pause point
Common error
used as read-and-do

Read and do. The reader performs each step as they read it. Used where the task is unfamiliar or the order matters and a missed step is unrecoverable.

Do and confirm. The work is done from memory or from practice, then the list is run at a defined pause point to catch omissions. Used where the performer is competent and the risk is forgetting under pressure.

Most organizational checklists are written as the first and used as the second, which is why they feel patronizing to experienced people and get ignored. Decide which one you are making.

Three drafting rules hold up. Keep it to what fits on one screen, because a longer list gets skimmed and skimming defeats the purpose. Give it a trigger, an explicit moment when it is run. And name who calls it, because in practice a check that anyone may call is called by nobody.

Twelve questions, and the answers that should worry you

1. What is our definition of a defect, and where is it written?

The answer to worry about is a confident verbal definition that two people state differently. Ask two people separately, and compare.

2. Which of our failures reach the customer, and which do we catch first?

If you cannot split your failures into caught and escaped, you cannot tell whether your checking is doing anything.

3. How would we find out about a problem we are not currently looking for?

Worry if every route runs through a customer complaint. That means your detection lag is the length of a customer's patience.

4. What did we change in the last quarter, and did anything get worse?

Worry if the answer is a list of improvements with no adverse effects at all. Real changes have costs somewhere, and an organization that never notices them is not looking.

5. Which recurring problem has been on the list longest?

The oldest open item tells you what your improvement process cannot handle. Usually it crosses a team boundary, which is the case treated in operating processes.

6. Who can stop the work, and when did they last do it?

Worry if the authority exists on paper and has never been used. That means it does not exist. The relationship between stated authority and real authority is the subject of management foundations.

7. What do we do with a defect after it is fixed?

Worry if the answer stops at the fix. Fixing the instance without asking why it escaped detection leaves the escape route open for the next one.

8. Which of our checks has never found anything?

A check that has never caught a defect is either preventing them beautifully or not being done. Find out which. These are the cheapest items to delete and the hardest to get agreement on.

9. Where does work sit waiting, and for how long?

Worry if nobody knows. Waiting time is invisible in most reporting and is usually the larger part of the total. The counting method is in operating processes metrics.

10. What do the people doing the work think the biggest problem is?

Worry if you have not asked in the last three months, or if the answer surprises you. Both are diagnostic.

11. What are we doing that is no longer needed?

Every quality system accumulates steps added after an incident nobody remembers. Worry if nothing has been removed in a year.

12. If our main supplier or upstream team sent us something wrong today, when would we notice?

Worry if the answer is measured in weeks. Detection lag on inputs is the quietest expensive problem in most operations.

Running the review without it becoming ritual

The failure mode of any recurring review is that the answers stop changing and the meeting continues anyway. Three habits keep it alive.

Three habits that keep reviews alive

  1. Change who answers
  2. Require a specific recent example
  3. Write down one thing to change

Change who answers. The same person giving the same answers every quarter is a signal that the review is measuring one person's opinion rather than the operation.

Require an example. Any answer to the questions above should come with a specific recent case. General reassurance is not an answer, and asking for the last instance is a reliable way to find out whether it is true.

Write down one thing to change, and only one. A review that produces twelve actions produces none. The related discipline of picking one thing at a time is in quality management.

What this checklist deliberately leaves out

It contains no numbers, and it will not tell you whether your quality is acceptable. That is a judgment about your customers and your obligations, and no general list can make it.

It also leaves out details specific to regulated work, and where an external body sets what you must check and how often, its current text governs and changes.

For work under a safety regime, the relevant regulator publishes its own requirements. The US Occupational Safety and Health Administration's published booklet on process safety management shows the level of specificity that can apply.

Read the source, not a summary, and get advice on anything consequential.

To generalize the checklist habit past quality, the wider set of recurring management checks is in management foundations checklist.

Building checks into the work, rather than around it, is a design idea under poka-yoke. That practice makes the wrong action physically or logically impossible instead of warning against it.

Common questions

Should the checklist be signed?

Only if the signature is used. A signature that nobody ever looks at converts a check into paperwork, and people learn quickly which is which. If you do want the record, decide who reads it and when, and say so on the form.

How long should a checklist be?

Short enough to be run at the moment it is needed, which for a pause in live work is a handful of items. The twelve above are a quarterly review rather than an in-the-moment check, which is why they can be longer and slower.

Our team says checklists insult their expertise. What now?

Take it seriously, because a read and do list aimed at experts genuinely is insulting and will be ignored. Ask them to design a do and confirm version covering only what gets missed under pressure. Ownership changes the reception, and their version is usually better and shorter.

Who should own the list?

The team doing the work, with someone outside reviewing it occasionally for items that have quietly stopped mattering. Ownership by a separate function produces a list that grows and never shrinks, because adding an item is safe and deleting one requires someone to take responsibility for the deletion.

More in Costs

Latest from Market Desk