Back to the on-screen lesson ·
A quality check is built from a named failure: the point where it first shows, a pass-or-fail standard, a frequency and a record; the failure rate it finds shows whether changes help.
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.
You will build a quality check from a named failure — where it shows, the standard, how often and what is recorded — tell a real check from a plea or a late one, and compute the failure rate the checks find.
Standard work says what a finished step should look like. A quality check asks whether it actually does, before the customer finds out that it did not.
| Term | What it means |
|---|---|
| Failure | A result that breaks the promise to the customer. |
| Quality check | A look for one named failure, at a stated point, against a standard. |
| Point | The place in the process where the failure can first be seen. |
| Pass-or-fail standard | A test anyone can apply the same way. |
| Frequency | How often the check runs: every item, or one in so many. |
| Failure rate | Failed items over items checked, times a hundred. |
'We need better quality' leads nowhere. 'We must never send a vegetarian box with meat in it' leads straight to a check. Start with the failure, named specifically, and work backwards:
Three things are not quality checks. Waiting for complaints finds failures after the customer has suffered them. 'Everyone be careful' checks nothing. And counting something else, such as boxes packed or minutes taken, cannot see the failure at all, however carefully it is done.
Another way: table
Three failures and the checks built from them.
| Failure | Where it shows | Standard | How often |
|---|---|---|---|
| Meat in a vegetarian box | Before sealing | Label matches order sheet | Every box |
| Hairline crack in a mug | Before packing | Clear ring when tapped, dry base after a minute | Every mug |
| Brakes that do not stop | Before handover | Stops within a bike length on each brake | Every bike |
Another way: steps
The best standards are written so that anyone can apply them the same way. A range does this well: rice portions between 6 and 7 ounces; reheated soup at least 165 degrees Fahrenheit on the probe. 'Looks appetizing' and 'tastes right' depend on who is checking, and two checkers will disagree.
Then count. If Maya's crack check fails 6 of 40 mugs from one firing, the failure rate is 6 ÷ 40 × 100 = 15 percent. Those six mugs never reached a café, which is the point of the check. The rate, recorded firing by firing, does a second job: it shows whether a change to the kiln or the clay made things better or worse. A check is not a promise that nothing will go wrong. It is a way of catching what does, and of learning from how often.
Name the failure. In the customer's terms, specifically: not 'poor quality' but 'a mug with a hairline crack reaches a café'.
Find the earliest point it shows. Walk the process map backwards from the customer. The crack can be seen after the final firing and before packing; after packing, nobody looks.
Write the standard. A test that passes or fails: a clear ring when tapped and a dry base after a minute of water.
Choose the frequency. Every item when one failure is serious or checking is cheap; one in so many when the check is slow and failures are minor.
Record it. A tick, an initial, a line on the sheet: proof the check ran, and the count of failures.
Check the check. Take a few items you know are faulty and see whether the check catches them. A crack check that passes a known cracked mug is not a check. Then compare two checkers on the same items: if they disagree, the standard is not yet clear enough.
Working back from the failure is needed because a check can only find what it looks for. Starting from 'quality' produces checks that look at whatever is easy to see, not at the failure that hurts the customer.
Choosing the earliest point is allowed, and best, because a failure caught early costs less to fix. A cracked mug caught before packing costs one mug; caught by the café, it costs the mug, the shipping, a replacement and some trust.
Computing a failure rate, rather than a count, is allowed because batches differ in size. Six failures in 40 mugs and six in 80 are very different results; dividing by the number checked makes them comparable.
Some checks should run on every item. A vegetarian box, a brake, an allergen label: one failure reaching a customer is serious, and the check takes seconds. Here the frequency is always 'every one'.
Others can run on a sample. Weighing every rice portion at the lunch rush would slow the kitchen; weighing one bowl in ten, and adjusting the scoop if a bowl falls outside the range, catches drift without stopping the line. A sample works when a failure is minor, when results drift slowly, and when one bad result signals that others are likely.
Write the frequency on the check itself, so nobody quietly decides to skip it on a busy day.
The best person to check is usually the person who just did the work, with a clear standard to check against. The packer checks the label before sealing; the mechanic rides the bike before handover. They are at the right point, they can fix a failure at once, and they learn from every one they catch.
For the most serious failures, add a second pair of eyes at a later point: the driver confirms the vegetarian labels against the delivery list before leaving. Two independent checks catch far more than one, because each misses different things. Keep the second check short, though, or it will be skipped when the day gets busy.
A failed check needs two responses. The first is immediate: fix or replace the item before it reaches the customer. The cracked mug is set aside, the brake cable re-tensioned, the box repacked.
The second is to record it, with a short note of what was found. Over a few weeks the notes show patterns: most cracks come from one shelf in the kiln, most brake failures follow one kind of cable. That is the starting point for the root-cause lesson later in this unit, which asks why a failure keeps happening rather than only catching it each time.
Some quality checks are not optional. Food businesses follow the food code adopted by their state or county, which sets minimum cooking, holding and reheating temperatures, such as 165 degrees Fahrenheit for reheated food held hot. Health inspectors check that the business measures and records them. A probe thermometer and a temperature log are a quality check the law requires.
Other trades have their own: brake and safety standards for some vehicle work, safety rules for electrical work. Find the rules that apply from the official source for the business's trade and location, and build the required checks into the process map first.
Checks decay. In the first month everyone taps every mug; by the sixth, on a busy Friday, a few are skipped, and by the ninth the habit has quietly gone. The record is what keeps a check alive. A column of initials with gaps in it shows at a glance when the check stopped running, and the owner can ask why before a failure gets through.
Look at the records once a week, not only the failure rate but the number of checks run. If the kitchen packed 120 vegetarian boxes and the sheet has 90 initials, a quarter of the boxes went out unchecked, whatever the failure rate says. Make the check easy to do and easy to record, and it will keep running when the business is busiest, which is when failures are most likely.
A check catches failures; it does not stop them being made. A bakery that catches ten label mix-ups a month has a good check and a weak process. The goal is a failure rate that falls over time, because the standard work and the process improve, while the check stays in place to catch whatever still gets through.
So read the failure rate as a signal for the process, not a score for the checker. A checker who catches more failures is doing the job well. The question the rate raises is why there were so many to catch, and that is where the next lessons, on complaints and root causes, begin.
A wholesale bakery in Sacramento supplies cafés with muffins and cookies. Twice in one year, a nut-free batch reached a café with a nut-containing label, or the other way round. Each time, the café had to pull the product, and once a child with an allergy was nearly served the wrong muffin.
The owner built a check from the failure backwards. The earliest point the mix-up could be seen was when the label was applied, before the trays went into the delivery crates. The standard: the baker reads the batch sheet aloud, the packer reads the label aloud, and both initial the sheet if they match. Every tray, every time. A second, short check at loading: the driver matches the crate labels against the café's order.
In the following year the label check caught 11 mismatches, about one a month, every one before it left the bakery. None reached a café. The records also showed most mismatches happened on Mondays, when a new packer was on, which led to a change in the training.
The FDA Food Code, adopted in some form by most states, sets minimum temperatures for cooking and reheating food and requires that hot food be held at 135 degrees Fahrenheit or above. Local health departments enforce the version their state adopted, and inspectors expect to see the temperatures checked and recorded.
Customers will tell us if something is wrong. By then the failure has reached them.
'Be careful' is a check. It names no point, no standard and no record.
Any measurement is a quality check. It must be able to see the failure it is for.
A check guarantees no failures. It catches failures and counts them; it does not prevent them.
Compare failure counts across batches. Compare rates, because batches differ in size.
Name the failure.
$\text{moldy berries found at home}$
From the customer's view.
Find the earliest point.
$\text{when the carton is picked up to bag}$
Before it leaves the stall.
Write the standard.
$\text{turn it over: no soft or moldy berries}$
Pass or fail.
Choose the frequency.
$\text{every carton}$
Quick, and a failure spoils trust.
Record the result.
$\text{a tally of cartons set aside}$
The count for the week.
Name the failure.
$\text{brakes that do not stop the bike}$
From the customer's view.
Find the earliest point.
$\text{before handover}$
After the repair, before the customer.
Write the standard.
$\text{stops within a bike length on each brake}$
At walking pace.
Choose the frequency.
$\text{every bike}$
One failure is serious.
Record the result.
$\text{a tick on the job tag}$
Proof the check ran.
Count the failure rate.
$4 \div 25 \times 100 = 16$
All caught in the shop.
Name the failure.
$\text{a cracked mug reaches a café}$
From the customer's view.
Find the earliest point.
$\text{after the final firing, before packing}$
When the crack can be seen.
Write the standard.
$\text{clear ring; dry base after a minute}$
Pass or fail.
Check the check.
$\text{three known cracked mugs: all fail}$
It catches what it is for.
Run it on a firing.
$6 \text{ of } 40 \text{ fail}$
Every mug.
Count the failure rate.
$6 \div 40 \times 100 = 15$
Failed over checked.
Compare with the next firing.
$2 \div 40 \times 100 = 5$
After moving the shelf: better.
Find the earliest point.
$\text{before the box is sealed}$
After sealing, nobody can see inside.
Write the standard.
$\text{label matches the order sheet}$
Pass or fail.
Choose the frequency and record.
Northside Repairs is worried about one failure: a bike handed back with brakes that do not stop it. Which check would catch it?
Complete the worked solution: Northside Repairs' brake check found $7$ failures among $25$ bikes in March and $1$ among $25$ bikes in April, after a new standard for fitting brake cables. Find each month's failure rate and how many percentage points it fell.
Find March's failure rate.
$(\text{failures}) \div (\text{bikes}) \times \text{a hundred} =$ p
Failures out of each hundred checked.
Find April's failure rate.
$(\text{failures}) \div (\text{bikes}) \times \text{a hundred} =$ q
The same measure after the change.
Find how far it fell.
$(\text{March rate}) - (\text{April rate}) =$ d
Percentage points.
Say what the check did in both months.
$\text{caught every failure in the shop}$
The rate tells whether the change helped.
Bright Home Cleaning wants a check for 'an apartment that fails its move-out inspection'. Put the steps of building the check in order.
Number the steps in order (write the number in the box):
Neighborhood Kitchen's draft quality checks. Mark every check that has a pass-or-fail standard.
This task has no paper form; do it on a device.
Maya's crack check tested $40$ mugs from one firing, and $3$ failed. What percentage of the checked mugs failed?
Answer:
A quick-lube shop in Cleveland checks every oil change for a loose drain plug before the car leaves. Last month the check failed $8$ of $100$ oil changes, and each failure was fixed in the bay. Next month the shop expects $200$ oil changes. If the failure rate stays the same, how many loose drain plugs should the check catch?
Answer:
A chili restaurant's standard is $15$ ounces of chili in each bowl, and a bowl passes the check if it is no more than $2$ ounces under or over. For which weights $w$, in ounces, does a bowl pass?
This task has no paper form; do it on a device.
Lesson test: one question per skill, one attempt each, no hints. Your answers are checked when you submit.
Bright Home Cleaning's move-out check found $1$ failures among $25$ apartments in March and $2$ among $25$ in April, after a new standard for cleaning behind the refrigerator. Fill in each month's failure rate, as a percentage.
| Failure rate, percent | |
|---|---|
| March | |
| April |
You can design a check that would catch the failure you are worried about, before the customer does. Tell someone why waiting for complaints is not a check. Next: what a checklist can and cannot verify.
20. Your turn: Neighborhood Kitchen's vegetarian boxes, step 3
$\text{every box; the packer initials the sheet}$
Proof, and a count.