botminds.aiRequest a demo
WorkflowQuality & Labeling

Nonconformance (NCMR) Investigation and MRB Disposition

NCMR investigation and MRB disposition that sets the outside records on the table, weighs each cause and option against them, and leaves each seat's signature to that seat.

See the demoBook a demo
Follow one thin-coating NCMR from the table to a CAPA ↓
Watch with transcript →
The problem

The record that settles an NCMR sits outside the quality system

The nonconforming material report holds the description and the hold. What decides it sits elsewhere: the traveler in the execution system and the supplier's certificate in its portal. The CMM report and the program behind it live on a file share, beside the furnace chart, the gauge calibration log, the drawing and the risk file. A quality engineer pulls them by hand for each case.

At the board each item gets a few minutes, so the option that feels right wins and the evidence against it never reaches the table. A use as is gets signed with no design rationale on file. A changed measuring program reads as a bad part, and good parts are scrapped or sent back to a supplier that did nothing wrong.

Built forQuality engineersManufacturing engineersDesign engineersSupplier quality engineers

The Problem Today

Review boards decide on held parts from memory, because the records that settle it sit outside the quality system.

95
FDA 483 observations on nonconforming product, fiscal 2025 our estimate
2 hours
to pull the evidence for one report by hand our estimate

Pulled by hand for each report

  • Traveler
  • Certificate
  • Measuring
  • Risk file
  • Concessions without a reasonUse as is signed with no design rationale on file
  • Good parts scrappedA measuring change reads as a bad part

The Solution

  1. Outside records on the tableTraveler, certificate, measuring report and program, furnace chart, gauge log
  2. Each cause weighedEvidence for and against each cause, and what would rule it out
  3. Use as is held at a gateDesign rationale and a risk file check before it reaches the board
  4. Each seat signs for itselfEach signer opens the records in their own session; none signs their own

Built-in controls

  • A held lot stays blocked until people decide
  • Agents read source systems and never sign
  • Each vote, reason and signature on the trail

What You Get

Boards sign each disposition on the records, and a measuring or supplier cause reaches its owner before the next lot.

$2.93M
of quality engineering time a year, for a big maker our estimate
  • Measuring causes tracedA program change found before good parts are scrapped
  • Concessions with a reasonDesign rationale and risk file check on each use as is
  • Systemic causes sent onA CAPA for our cause, a request for the supplier's
Thursday's board, item 1

An NCMR for 140 coated hip shells: thin parts, or a moved gauge band?

Porous-coated acetabular shells, coated by an outside supplier and measured on the CMM after coating. The sampled shells read thin, and the board meets at 10:00.

demoThe plant, its coating supplier, the shells and the board members are sample data, as is each reading.

  1. 0110:00

    Item 1 of 35: nine of 13 sampled shells read under the 0.80 mm minimum

    Quality engineer, MRB chair

    The session opens on a held lot of 140 porous-coated shells. At op 60 the CMM measured 13 of them, and 9 read below the 0.80 mm coating minimum. All 140 sit in the MRB cage, blocked in the ERP, none shipped. It is day 4 of 10 on the disposition clock, and scrapping the lot would cost $57,680.

  2. 0210:01

    Six records from outside the quality system land on the table, passages marked

    Quality engineer, MRB chair

    She prepares the table and watches each step. First the intake checker confirms the hold against the ERP. Then the evidence puller brings the traveler, the coater's certificate of conformance, the CMM report, the program history, the furnace chart and the gauge calibration log. Last, the risk file checker finds coating thickness verified only from 0.80 mm up.

  3. 0310:02

    Rev D moved the gauge band to 23.0 mm below a coated pole

    Quality engineer, MRB chair

    The CMM program's history is the record that matters. Rev C measured the band 3.0 mm above the machined rim face. Rev D, rewritten for a software update, sets it 23.0 mm below the pole touch point, and no check against a reference part is recorded. On a coated shell that pole sits one coating higher, where the dome is narrower.

  4. 0410:03

    Program datum 81%, thin supplier run 58%, and both can be true

    Quality engineer, MRB chair

    The cause analyst lists four causes, each with what supports it, what argues against it and what would rule it out. Three shells re-run on rev C read 0.11 to 0.13 mm thicker. Yet run 1188's witness coupons sit at the low end, and its sampled shells are the lowest readings. CMM hardware and an oversize substrate are ruled out by their own records.

  5. 0510:05

    Use as is cannot reach the table: coating is verified only from 0.80 mm

    Quality engineer, MRB chair

    For each of six codes the draft lists the evidence for, the evidence against and what would make it wrong. Sort and re-inspect is drafted. Use as is and repair stop at the gate: thickness is risk control RC-07a, verified from 0.80 mm, with no design rationale below it. Rework has no validated re-coat step, and returning the lot fails while the cause may be the program.

  6. 0610:06

    She opens the reference shell check, then signs the sort as its author

    Quality engineer, MRB chair

    Rev E restores the rim-face datum and reads the reference shell at 1.001 mm against a sectioned value of 1.003 mm. She opens that record before the signature box will take her password, and signs as author, never approver. A sort needs quality and manufacturing; she added design engineering because use as is was considered.

  7. 0710:08

    Design engineering opens the drawing and the risk file, and rejects use as is

    Design engineer

    The design engineer switches to her own view and signs in as herself. She opens drawing rev H, with its 0.80 to 1.20 mm band, and the risk file entry for loss of biological fixation, then rejects use as is: below the verified range. Her agreement to the sort is a second signature, and two of three seats are now signed.

  8. 0810:10

    Manufacturing books 28 CMM hours on four machines and signs the sort

    Manufacturing engineer

    The manufacturing engineer opens the rev E reference check himself and approves for his discipline: 140 shells at 12 minutes each, about 7 hours on each of four CMMs. The board has signed. Shells that fail will be scrapped at once; the release of shells that pass waits for a quality engineer who did not work this NCMR.

  9. 0910:11

    A program cause opens a CAPA, and seven more CMM programs wait for a check

    Quality engineer, MRB chair

    The packet and CAPA writer drafts a CAPA with the packet attached: CMM program revisions released without verification against a reference part. Seven other programs changed in the same software update are listed, one already holding up a lot. The look-back finds two earlier lots measured on rev D, clear for the minimum and not clear for the maximum.

Two earlier lots on rev D pass the minimum and miss the maximum

Rev D reads about 0.13 mm thin. That is safe for the 0.80 mm floor, and it hides a shell over the 1.20 mm ceiling.

Earlier lot on rev DShellsHighest sampled reading, plus 0.13 mmAgainst the maximum
The first lot1201.06 + 0.13 = 1.19 mmNot clear
The second lot801.02 + 0.13 = 1.15 mmNot clear

demoSample data, from the working solution. Only 13 shells of each lot were measured, so the rest could sit higher. Both lots are still in finished goods, and the look-back proposes blocking them and re-measuring all 200 on rev E.

Around the board

The agenda, open NCMRs and every signature behind a disposition

What the chair, the seats and the site quality director read before and after a session.

Thursday's agenda35 items ordered by value at stake, then clock, with each seat's state. Eleven routine items sign as a batch once each signer opens a sample of 3.
Open NCMRs1,240 open at the plant, from receiving, in process and final inspection: search by id, part, lot, supplier or words, and filter by stage.
One NCMRWhat is wrong, the hold in its cage and the ERP block, the six outside records, and each agent step and signature in order.
Audit trailEach agent step, vote, reason and signature, its meaning and the session it came from. It exports as CSV, with an evidence file per NCMR.
Codes and signersSix disposition codes and who signs each, the gate on use as is and repair, and the batch rules. A looser batch limit waits for the quality systems lead.

demoSample data, from the working solution.

Four seats, one independent checker, and nobody approves their own draft

Each seat opens its own records in its own session before it signs.

Quality engineer, MRB chair
Sees
The table, the causes and each option's evidence against it
Signs
Authors the disposition, confirms a cause, opens the CAPA
Design engineer
Sees
The drawing, the risk file and the verified range
Signs
The form-fit-function rationale, or a rejection of use as is
Manufacturing engineer
Sees
Travelers, the measuring program check and machine time
Signs
Approval for manufacturing on a sort, rework or scrap
Supplier quality engineer
Sees
Certificates, coupons and the supplier's history
Signs
Sends a corrective action request when the cause is the supplier's
Independent checker
Sees
Sampled re-measure reports on an NCMR she did not work
Signs
The release of passed parts after a large sort
Site quality director
Sees
Clocks, use-as-is calls and overdue items, read only
Signs
Nothing at the table
Who prepares the table

Seven agents pull, weigh and draft, and none of them moves a part

Each cites the passage behind what it says and hands the call to a named person.

Intake checker

Checks a new report's lot, quantity and hold location against the ERP and flags missing containment.

  • Never moves, releases or unblocks stock
Evidence puller

Pulls the traveler, certificate, CMM report and program, furnace chart, calibration log, drawing and risk file.

  • Leaves each source record as it found it
  • Reads only the lot, part and supplier on the report
Cause analyst

Lists possible causes with evidence for and against, a confidence, and what would rule each out.

  • Never calls a cause final
  • Never blames a person without the process reason
Disposition drafter

Drafts the code and rationale, and for every option what would make it wrong.

  • Never drafts use as is or repair past the gate
  • Never signs
Risk file checker

Checks whether the feature is a risk control or design output, and its verified range.

  • Leaves the acceptability of a risk to design engineering
Packet and CAPA writer

Assembles the packet and drafts a CAPA or a supplier request when a cause is systemic.

  • Never opens or sends either without the owner's signature
Clock keeper

Watches the containment and disposition clocks and orders the agenda.

  • Cannot extend a clock

Holds, releases and concessions stay with the material review board

Pulling and drafting is the agents’ share. Each of these decisions carries a person’s signature.

No agent moves stockBlocking, scrapping and releasing parts take a person's signature. A hold or a scrap takes effect at once; a release waits for its second person.
Use as is needs design engineeringWithout the form-fit-function rationale and a passing risk file check, a use as is or repair never reaches the board.
A cause is confirmed by the report's ownerThe cause analyst gives a confidence and what would rule a cause out. The quality engineer who owns the report confirms it.
Typed reasons are read by a personA note or reason typed at the table is never compared by software. The chair opens it before the item can close.
Field returns are not taken inProduct returned from the field is a complaint record for the complaint handling unit. This board takes receiving and production only, and no agent opens a complaint file.

What MRB chairs ask about NCMR investigation

What is an NCMR?

An NCMR, or nonconforming material report, records parts or material that miss their drawing or specification, with the hold that contains them. Here agents pull the records behind each NCMR, weigh the causes and draft a disposition, and the material review board signs it.

Can the agents approve a use as is disposition?

No. A use as is or repair draft cannot reach the table without design engineering's rationale and a passing risk file check, and every seat it needs signs for itself.

How does each seat sign?

In its own session, with its own user ID and password, after opening the records it is asked to read. The author signs as author and never as approver, and a disagreement reopens the item for every seat.

What if the cause is the measuring system?

The evidence puller brings the CMM program's revision history and the gauge calibration log with the readings, so a program change shows as a cause beside the supplier and the part. A sort on a verified program then decides each serial.

What happens to a systemic cause?

The packet and CAPA writer drafts a CAPA with the packet attached, or a supplier corrective action request when the cause is the supplier's. The owner opens or sends it with a signature.

Does it handle product returned from the field?

No. Field returns are complaint records and go to complaint handling. This material review board works on nonconformances found at receiving and in production.

Quality & Labeling

Other quality and labeling work on the plant floor

All 4 solutions →
FDA 483 and Warning Letter Response

Every promise to FDA has an owner, a date and its proof.

1-min demoUDI Data Reconciliation (GUDID and EUDAMED vs Label)

Every database says what the label says, each fix signed before it is sent.

1-min demoCAPA Effectiveness Verification

Every closed CAPA proved with its own sums, and the same cause found where the fix never went.

1-min demo

Bring one held lot
and the records your board would ask for.

We will set them on the table and show what argues against the option you like.

Book a demoBrowse solutions