Quality Non-Conformance Log

What this tool does

CM8-84 is a working non-conformance register. You record what failed, how much of it, what you did with it and why it happened. The tool prices the scrap and the rework, ranks the root causes by cost as a Pareto, tracks the open actions against their target dates, and prints a report you can take to a management review, a customer audit or a supplier meeting.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so part numbers, customer names, batch records and cost data never leave the computer you are using.

What to raise as a non-conformance

A non-conformance is any product, part or delivery that does not meet its specification. Raise one record per batch, per defect, per discovery. If the same batch fails for two unrelated reasons, that is two records — otherwise the Pareto blames one cause for the other's cost. If the same defect appears in a later batch, that is a new record with the recurrence box ticked, not an edit of the old one.

Record it whether it was caught internally or by the customer. The split between the two is one of the more revealing figures in the register: a low internal count with a high customer count means your inspection is not where the failures are, and the cost you can see here is only a small part of what those failures are actually costing you.

Write the description of the containment and the corrective action in plain words, as a sequence of events. "The bore was 0.06 mm oversize on 12 of 60 parts" is useful. "Machining problem" is not, and it will read badly if the register is ever produced in an audit.

Quantities, and the trap in them

Three quantities are recorded: affected, scrapped and reworked. Affected is how many units the non-conformance covers — the batch, the delivery, the shipment. Scrapped and reworked are what actually happened to them. The remainder — affected minus scrapped minus rework — is what was used as is, accepted under concession or found to be conforming after a full inspection.

The tool refuses a record where scrapped plus reworked exceeds affected, because that is the arithmetic error that quietly doubles a cost figure. A unit cannot be both scrapped and reworked. If you reworked a batch and then scrapped some of it during the rework, record the units once, under the outcome they ended up with.

How the cost is calculated

Two unit figures drive everything. The unit cost of the item is what one unit has cost you by the point it was rejected: material, plus the labour and overhead already added to it. A casting rejected at goods-in costs you what you paid for it. The same casting rejected after four machining operations costs you a great deal more — use the value at the point of rejection, not the purchase price and not the selling price.

The rework cost per unit is the labour, materials and machine time to put one unit right.

Days open runs from the date raised to the date closed, or to today while the record is still open:

An action is overdue when the record is not yet closed or verified and its target date has passed. The warning window in the settings flags items approaching the date as well:

If a record has no scrap and no rework — a return to supplier before you added any value, or a batch accepted under concession — its cost of poor quality is zero. That is correct, and it is why the register count and the cost never move together. Two records in the sample data are exactly this case.

What the cost figure leaves out

This matters more than anything else on the page, so it is stated plainly: the cost of poor quality here is scrap plus rework, and nothing else. It excludes, among others:

• Lost customers. Usually the largest number of all, and never on an invoice.
• Expedited freight and premium transport to recover a missed delivery.
• Sorting, containment and re-inspection labour — the hours spent checking everything else in stock after a failure is found.
• Warranty, credit notes, replacement product and returns handling.
• Line stoppages and lost capacity, at your plant or at the customer's.
• Management and engineering time spent investigating, meeting and reporting.
• Overtime to make up the shortfall the scrap created.

Studies of quality cost have argued for decades that the hidden portion is several times the visible one. The exact multiple is not knowable from this register and the tool does not invent it. Treat the figure here as the floor of your quality cost — the part you can evidence line by line — and say so whenever you quote it. It is more than good enough to justify spending money on prevention, which is what it is for.

Cost as a share of production value

An absolute cost means little without a denominator. Enter the sales or production value for the period on the Settings tab and the tool shows the ratio:

The one thing you must get right is that the two cover the same period. Filtering the register to one quarter while leaving a full year's production value in the settings divides a quarter's cost by a year's output and produces a ratio four times too flattering. Clear the date filter, or change the production value, but never leave them mismatched.

The target percentage is your own. One per cent of production value is a common first target for a manufacturer with an established quality system, but what is achievable differs enormously between a high-volume pressing shop and a low-volume aerospace fabricator. Set it from your own history, and move it when you beat it.

The Pareto, and the cumulative column

The Pareto table and the first chart rank root causes by cost, highest first. Two columns follow the cost:

Because the cumulative column is a running sum taken down the sorted list, and the total is the final value of that same running sum, the last row always reads 100.0 % — there is no rounding drift to explain away. The chart colours in blue every cause up to the point where the cumulative share passes 80 %: those are the ones worth an engineer's week. The grey ones are not, yet.

Read the cost Pareto against the defect-type donut, which counts records rather than money. They usually disagree, and the disagreement is the point: the defect you see most often is frequently not the one costing you most, because one scrapped batch of expensive machined parts outweighs a hundred relabelled cartons.

If every cause is recorded as "not established", the Pareto is a single bar and tells you nothing. That is a signal about your investigations, not about the tool.

Choosing a root cause

The categories are the traditional six — machine, method, material, measurement, manpower and environment — with supplier, design and "not established" added, because in practice those three account for a great deal of what a register contains.

Choose the category for the cause you would have to change to stop it recurring, not the category of the symptom. A dimensional defect caused by a gauge that was out of calibration is a measurement cause, not a machine one. A part scrapped because the drawing tolerance is tighter than any process can hold is a design cause, however it was made.

Resist "manpower" as a default. Someone made a mistake, but ask why the mistake was possible and why it was not caught: two near-identical parts stored in adjacent bins is a method problem wearing a manpower costume. Leave the cause as "not established" while an investigation is genuinely still open — an honest unknown is more useful than a guess that lands in the Pareto and misdirects the next fix.

Dispositions

The disposition records what was decided about the material. Scrap and rework both drive cost and the tool insists on a matching quantity for each. Use as is and concession mean the material was accepted — the difference is that a concession is a documented, usually customer-approved, deviation, and it should always be in writing. Return to supplier means the material went back.

Two cautions. A return to supplier costs you nothing here only if you added no value and received full credit; if you machined the part first, or paid the freight, that cost is real and the register does not see it. And a run of concessions on the same characteristic is a specification that no longer matches the process — fix the drawing or fix the process, rather than granting the same concession every month.

Containment, correction and closure

Keep the two action fields distinct, because they answer different questions. Containment is what you did that day to stop the problem reaching anyone else: quarantine the batch, hold the shipment, inspect the stock, tell the customer. Corrective action is what stops it happening again. A register full of containment and empty of corrective action explains exactly why the same reference number keeps reappearing with a different suffix.

The status ladder runs open → contained → investigating → closed → verified. Verified is not decoration: it means somebody went back after the fact and confirmed the action worked — the next three batches were checked, the recurrence did not happen. Closing a record the day the action is written down is how a corrective action system quietly stops working.

The tool requires a date closed before it will accept a closed or verified status, and refuses a date closed on a record that is still open. Open items are listed by target date in the Reports tab, with anything past its date marked overdue and anything inside the warning window marked due soon.

Recurrence

Tick the recurrence box when this is the same defect on the same product from the same cause as a record you already hold. The headline figure is simple:

It is the single best measure of whether your corrective actions are real. Cost can fall because volumes fell; the count can fall because people stopped raising records. A recurrence rate that stays high says the fixes are not holding, regardless of what the other numbers do. Anything repeating after a record was marked verified deserves a second look at what was actually verified.

Printing and sharing

Print Report produces a report from whatever the current filter shows: header, headline figures, the four charts, the Pareto table, the open items and the full register, with your closing notes. Print to PDF to circulate it.

The scope line under the title states the filter in force. Clear the filters before issuing anything described as the full period, and check that the production value on the Settings tab covers the same period. Cost data and customer complaint detail are commercially sensitive — decide who should receive the report before you send it, and consider filtering to one product family rather than exporting the lot.

Saving your work

Records, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the register for spreadsheet work, including the columns hidden from the on-screen table and every calculated figure, and it covers every filtered record rather than only those drawn on screen. Reset asks twice, then erases everything this tool has stored. There is no undo.

Quality records often have a retention period set by a customer contract or a management-system standard. Keep archived exports accordingly — a browser is not a records system.

Accuracy & disclaimer

This tool calculates from what you enter and nothing else. It cannot tell whether a unit cost is right, whether the root cause is the real one, whether a quantity was estimated, or whether the records in front of it are all the records there are. Every figure on the page inherits those limits.

The cost of poor quality shown is scrap plus rework only. It is not a full quality cost, not an accounting figure, and not suitable for a financial statement without reconciliation to your own ledger. Requirements for non-conformance control, disposition authority, concession approval and record retention differ by management-system standard, by industry and by customer contract. This is an internal record-keeping and costing aid, not a compliance system and not advice.