chart view only · the ranking always scores the full window
scrapreworkparts rejected
Rank 4WatchPlating - Scratch/Pre-Handle
200101-FABKW M8 HDWY HDR, FAB
$194K
Improving · parts -54%7.3% affected volume
Range
chart view only · the ranking always scores the full window
scrapreworkparts rejected
Rank 5WatchPolishing - Weld Seam
200709A-PNTKW 4.5 TRACER TIP,PAINT
$709K
Improving · parts -55%7.2% affected volume
Range
chart view only · the ranking always scores the full window
scrapreworkparts rejected
Rank 6WatchPlating - Burn
04-35552-032-POLPIPE-CRV STK,4IN-5IN,POL SS,32
$407K
Improving · parts -75%6.7% affected volume
Range
chart view only · the ranking always scores the full window
scrapreworkparts rejected
Rank 7WatchPlating - Rack Mark
11180RETAINING RING
$404K
Improving · parts -71%6.5% affected volume
Range
chart view only · the ranking always scores the full window
scrapreworkparts rejected
Rank 8WatchPlating - Stains
1020399-DBRTUNGSTEN SPROCKET DEBUR
$504K
Improving · parts -61%5.6% affected volume
Range
chart view only · the ranking always scores the full window
scrapreworkparts rejected
Card 1 of 8
By line or departmentthe plant first, then the line, then the part · click a line to filter the ranking and the board to it · mock lines until the real reject detail's line field connects
See all 15 opportunities, ranked 1 to 15
1000007
10G CARBON STEEL 120" X 30" · Fab - Handling Damage
The brief: why it ranks here, the trend, and your decision
Why this is #1
Worsening (six straight rising weeks) and its 13.9% fallout rate is above the plant bar of 3.0%. Annualized, $5,232,313.
Loss trend, last 8 weeks
It stepped up in the week of Jul 19 and has not come back down: $55K a week before, $116K a week since, on 3,058 parts rejected in those 6 weeks against 482 in the 2 before.
Financial impact$5.23M / yr$802,766 / 8 weeks
Evidence3,540 parts rejected$227 / rejected part · 204 tickets · 8 wks
Trend↑ 92%Failures are rising · parts rejected · recent half vs prior half of the window
Forecast impact$1.65Mnext 8 weeks · 52,612 units forecast x 13.9% fallout x $227 · projected, not measured
View calculation details →
The other dollar readings
$5,232,313the headline · window pace x 365, annualized
$527,694newest half, rework 40¢
$802,766full 8 wks incl. rework at full value
$2,688,252scrap ahead, NOT measured · 85,495 released units x 13.9% scrap x $227
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run up 2% vs the prior 4 wks · volume +88%
No baseline of its own yet · judged against the plant
What changedweighted loss $582K to $803K · 44 more reject events · direction +9% to +2%
Rank unchangedOn the list 23+ runsin all 8 weeks scored
Last time: dismissed 2026-08-25 13:59Z · 2 notes in the dossier
First move (AI draft, human-gated)
Break the part's rejects down by shift and week to isolate what changed.
Checker tonight: 8 weeks of data, no one-week spike, about an amount the engine could not verify per-unit worsening, and mostly true scrap make this a real new loss.
The brief: why it ranks here, the trend, and your decision
Why this is #2
Worsening (parts rejected rose from 979 to 2,361 between the window's halves, beyond noise at 95% confidence) and its 13.5% fallout rate is above the plant bar of 3.0%. Annualized, $3,290,720.
Loss trend, last 8 weeks
It stepped up in the week of Jul 19 and has not come back down: $45K a week before, $69K a week since, on 2,107 parts rejected in those 6 weeks against 460 in the 2 before.
Financial impact$3.29M / yr$504,878 / 8 weeks
Evidence2,567 parts rejected$197 / rejected part · 204 tickets · 8 wks
Trend↑ 36%Failures are rising · parts rejected · recent half vs prior half of the window
Forecast impact$419Knext 8 weeks · 32,558 units forecast x 6.5% fallout x $197 · projected, not measured
View calculation details →
The other dollar readings
$3,290,720the headline · window pace x 365, annualized
$197,585newest half, rework 40¢
$504,878full 8 wks incl. rework at full value
$680,514scrap ahead, NOT measured · 52,907 released units x 6.5% scrap x $197
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 4% vs the prior 4 wks · volume +36%
No baseline of its own yet · judged against the plant
What changedweighted loss $293K to $348K · 30 more reject events · direction -10% to -4%
Rank unchangedOn the list 23+ runsin all 8 weeks scoredChecker held tonight
First move (AI draft, human-gated)
Break the part's rejects down by shift and week to isolate what changed.
Checker tonight: 8 weeks are present, but the trend is essentially flat at about -3.7% per unit, so this is a steady loss, not a worsening project.
The brief: why it ranks here, the trend, and your decision
Why this is #3
Worsening (parts rejected rose from 409 to 945 between the window's halves, beyond noise at 95% confidence) and its 8.4% fallout rate is above the plant bar of 3.0%. Annualized, $1,705,678.
Loss trend, last 8 weeks
It stepped up in the week of Jul 19 and has not come back down: $20K a week before, $37K a week since, on 859 parts rejected in those 6 weeks against 155 in the 2 before.
Financial impact$1.71M / yr$261,693 / 8 weeks
Evidence1,014 parts rejected$258 / rejected part · 67 tickets · 8 wks
Trend↑ 70%Failures are rising · parts rejected · recent half vs prior half of the window
Forecast impact$430Knext 8 weeks · 19,856 units forecast x 8.4% fallout x $258 · projected, not measured
View calculation details →
The other dollar readings
$1,705,678the headline · window pace x 365, annualized
$164,655newest half, rework 40¢
$261,693full 8 wks incl. rework at full value
$698,622scrap ahead, NOT measured · 32,266 released units x 8.4% scrap x $258
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run up 3% vs the prior 4 wks · volume +64%
No baseline of its own yet · judged against the plant
What changedweighted loss $206K to $262K · 13 more reject events · direction -2% to +3%
Rank unchangedOn the list 19 runsin all 8 weeks scoredChecker held tonight
First move (AI draft, human-gated)
Pull True Cam tank parameters for the line and check rack contact / agitation on the top part numbers.
Checker tonight: 8 weeks are present and mostly true scrap is meaningful, but the trend is only about +3.3% per unit, which is too flat to call a worsening project.
#4 PriorityChronic3.5% affected volumeStableSource One
$1.62M
The brief: why it ranks here, the trend, and your decision
Why this is #4
Steady, but its 3.5% fallout rate is above the plant median and $1,615,953 a year puts it in the plant's top quarter. A steady bleeder is a project, not a fire.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$1.62M / yr$247,927 / 8 weeks
Evidence820 parts rejected$302 / rejected part · 211 tickets · 8 wks
Trend↓ 17%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$257Knext 8 weeks · 46,882 units forecast x 1.8% fallout x $302 · projected, not measured
View calculation details →
The other dollar readings
$1,615,953the headline · window pace x 365, annualized
$78,974newest half, rework 40¢
$247,927full 8 wks incl. rework at full value
$418,291scrap ahead, NOT measured · 76,184 released units x 1.8% scrap x $302
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 18% vs the prior 4 wks · volume -1%
No baseline of its own yet · judged against the plant
What changedweighted loss $188K to $177K · 37 more reject events · direction -26% to -18%
Up 2On the list 23+ runsin all 8 weeks scoredConsolidated into known work
Last time: on hold 2026-08-09 13:42Z · 2 notes in the dossier
First move (AI draft, human-gated)
Pull True Cam tank parameters for the line and check rack contact / agitation on the top part numbers.
Checker tonight: The trend is healing by about 18.4% per unit over 8 weeks, so this is not a still-worsening opportunity.
The brief: why it ranks here, the trend, and your decision
Why this is #5
Big enough ($798,792 a year), but it is healing (six straight falling weeks); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$799K / yr$122,554 / 8 weeks
Evidence469 parts rejected$261 / rejected part · 42 tickets · 8 wks
Trend↓ 44%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$87Knext 8 weeks · 14,469 units forecast x 2.3% fallout x $261 · projected, not measured
View calculation details →
The other dollar readings
$798,792the headline · window pace x 365, annualized
$29,946newest half, rework 40¢
$122,554full 8 wks incl. rework at full value
$141,574scrap ahead, NOT measured · 23,512 released units x 2.3% scrap x $261
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 9% vs the prior 4 wks · volume -39%
No baseline of its own yet · judged against the plant
What changedweighted loss $111K to $84K · 11 fewer reject events
Rank unchangedOn the list 9 runsin all 8 weeks scoredConsolidated into known work
First move (AI draft, human-gated)
Run a cycle-time and fixture check on the polishing cell; confirm operator ramp on that line.
Checker tonight: This defect is already owned by open initiative PRJ-114, so it should roll into that work instead of opening a new ticket.
The brief: why it ranks here, the trend, and your decision
Why this is #6
Big enough ($561,426 a year), but it is healing (failure counts are healing (parts rejected fell from 821 to 304 between the window's halves, beyond noise at 95% confidence); a problem fixing itself is not chased); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$561K / yr$86,137 / 8 weeks
Evidence402 parts rejected$214 / rejected part · 57 tickets · 8 wks
Trend↓ 55%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$267Knext 8 weeks · 37,716 units forecast x 3.3% fallout x $214 · projected, not measured
View calculation details →
The other dollar readings
$561,426the headline · window pace x 365, annualized
$26,784newest half, rework 40¢
$86,137full 8 wks incl. rework at full value
$433,260scrap ahead, NOT measured · 61,289 released units x 3.3% scrap x $214
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 4% vs the prior 4 wks · volume -48%
No baseline of its own yet · judged against the plant
What changedweighted loss $110K to $81K · 12 fewer reject events · direction +3% to -4%
Up 2On the list 23+ runsin all 8 weeks scoredChecker held tonight
Last time: dismissed 2026-07-30 02:38Z
First move (AI draft, human-gated)
Pull True Cam tank parameters for the line and check rack contact / agitation on the top part numbers.
Checker tonight: 8 weeks are present, but the trend is essentially flat at about -3.6% per unit, so this is not yet a worsening project.
Your decision
Add a reason (optional)
Why?
200101-FAB
KW M8 HDWY HDR, FAB · Plating - Scratch/Pre-Handle
The brief: why it ranks here, the trend, and your decision
Why this is #7
Big enough ($129,986 a year), but it is healing (nine straight weeks below its own average); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$130K / yr$19,943 / 8 weeks
Evidence94 parts rejected$212 / rejected part · 17 tickets · 8 wks
Trend↓ 19%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$40Knext 8 weeks · 5,187 units forecast x 3.7% fallout x $212 · projected, not measured
View calculation details →
The other dollar readings
$129,986the headline · window pace x 365, annualized
$8,911newest half, rework 40¢
$19,943full 8 wks incl. rework at full value
$65,303scrap ahead, NOT measured · 8,429 released units x 3.7% scrap x $212
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 18% vs the prior 4 wks · volume -2%
No baseline of its own yet · judged against the plant
NewOn the list 1 runsin all 8 weeks scoredConsolidated into known work
First move (AI draft, human-gated)
Pull True Cam tank parameters for the line and check rack contact / agitation on the top part numbers.
Checker tonight: The trend is healing by about 17.9% per unit over 8 weeks, so this is not a current worsening opportunity.
Your decision
Add a reason (optional)
Why?
11180
RETAINING RING · Plating - Rack Mark
#8 PriorityWatch3.0% affected volumeImprovingSource One
$295K
The brief: why it ranks here, the trend, and your decision
Why this is #8
Big enough ($295,131 a year), but it is healing (failure counts are healing (parts rejected fell from 686 to 172 between the window's halves, beyond noise at 95% confidence); a problem fixing itself is not chased); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$295K / yr$45,280 / 8 weeks
Evidence220 parts rejected$206 / rejected part · 24 tickets · 8 wks
Trend↓ 48%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$50Knext 8 weeks · 25,559 units forecast x 0.9% fallout x $206 · projected, not measured
View calculation details →
The other dollar readings
$295,131the headline · window pace x 365, annualized
$8,644newest half, rework 40¢
$45,280full 8 wks incl. rework at full value
$80,715scrap ahead, NOT measured · 41,533 released units x 0.9% scrap x $206
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 16% vs the prior 4 wks · volume -43%
No baseline of its own yet · judged against the plant
NewOn the list 1 runsin all 8 weeks scoredConsolidated into known work
First move (AI draft, human-gated)
Pull True Cam tank parameters for the line and check rack contact / agitation on the top part numbers.
Checker tonight: This part matches dismissed initiative PRJ-097, and the trend is healing by about 15.9% per unit, so it should not be reopened.
Your decision
Add a reason (optional)
Why?
04-31798-026-POL
5" PIPE-CRV STACK 439 SST NiCR · POL · Plating - Rack Mark
The brief: why it ranks here, the trend, and your decision
Why this is #9
Big enough ($752,595 a year), but it is healing (failure counts are healing (parts rejected fell from 918 to 364 between the window's halves, beyond noise at 95% confidence); a problem fixing itself is not chased); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$753K / yr$115,467 / 8 weeks
Evidence444 parts rejected$260 / rejected part · 49 tickets · 8 wks
Trend↓ 36%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$153Knext 8 weeks · 29,004 units forecast x 2.0% fallout x $260 · projected, not measured
View calculation details →
The other dollar readings
$752,595the headline · window pace x 365, annualized
$45,250newest half, rework 40¢
$115,467full 8 wks incl. rework at full value
$248,624scrap ahead, NOT measured · 47,132 released units x 2.0% scrap x $260
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 13% vs the prior 4 wks · volume -26%
No baseline of its own yet · judged against the plant
What changedweighted loss $158K to $115K · 15 fewer reject events
Down 5On the list 23+ runsin all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
14101869
MEDALLION, BREAKOUT · Material - Dings/Gouges/Dents
#10 PriorityWatch1.8% affected volumeImprovingSource One
$441K
The brief: why it ranks here, the trend, and your decision
Why this is #10
Big enough ($440,827 a year), but it is healing (six straight falling weeks); chasing it would spend a project on a problem fixing itself.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$441K / yr$67,634 / 8 weeks
Evidence276 parts rejected$245 / rejected part · 33 tickets · 8 wks
Trend↓ 69%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$63Knext 8 weeks · 16,835 units forecast x 1.5% fallout x $245 · projected, not measured
View calculation details →
The other dollar readings
$440,827the headline · window pace x 365, annualized
$15,634newest half, rework 40¢
$67,634full 8 wks incl. rework at full value
$102,466scrap ahead, NOT measured · 27,357 released units x 1.5% scrap x $245
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 11% vs the prior 4 wks · volume -62%
No baseline of its own yet · judged against the plant
What changedweighted loss $99K to $62K · 13 fewer reject events · direction -25% to -11%
Down 1On the list 23+ runsin all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
04-35552-032-POL
PIPE-CRV STK,4IN-5IN,POL SS,32 · Plating - Burn
#11 Priority2.9% affected volumeImprovingLine 91
$297K
The brief: why it ranks here, the trend, and your decision
Why this is #11
Below the plant's bars on every rule.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$297K / yr$45,549 / 8 weeks
Evidence206 parts rejected$221 / rejected part · 21 tickets · 8 wks
Trend↓ 63%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$49Knext 8 weeks · 11,411 units forecast x 2.0% fallout x $221 · projected, not measured
View calculation details →
The other dollar readings
$296,880the headline · window pace x 365, annualized
$8,137newest half, rework 40¢
$45,549full 8 wks incl. rework at full value
$80,165scrap ahead, NOT measured · 18,543 released units x 2.0% scrap x $221
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 30% vs the prior 4 wks · volume -60%
No baseline of its own yet · judged against the plant
What changedweighted loss $63K to $37K · 7 fewer reject events · direction -21% to -30%
Rank unchangedOn the list 23+ runsin all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
04-31813-026-POL
5" to 4" PIPE-30DEG CRV 439SST · Plating - Wrong Process
#12 Priority2.3% affected volumeImprovingLine 37
$290K
The brief: why it ranks here, the trend, and your decision
Why this is #12
Below the plant's bars on every rule.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$290K / yr$44,423 / 8 weeks
Evidence157 parts rejected$283 / rejected part · 22 tickets · 8 wks
Trend↓ 57%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$119Knext 8 weeks · 25,617 units forecast x 1.6% fallout x $283 · projected, not measured
View calculation details →
The other dollar readings
$289,544the headline · window pace x 365, annualized
$13,299newest half, rework 40¢
$44,423full 8 wks incl. rework at full value
$193,151scrap ahead, NOT measured · 41,628 released units x 1.6% scrap x $283
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 4% vs the prior 4 wks · volume -41%
No baseline of its own yet · judged against the plant
What changedweighted loss $53K to $37K · 7 fewer reject events · direction -25% to -4%
Rank unchangedOn the list 21 runsin all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
107A-060-1102
7X60,MIT,ST(OD) CHR POL · Plating - Scratch/Pre-Handle
#13 Priority2.2% affected volumeImprovingLine 49
$227K
The brief: why it ranks here, the trend, and your decision
Why this is #13
Below the plant's bars on every rule.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$227K / yr$34,816 / 8 weeks
Evidence184 parts rejected$189 / rejected part · 28 tickets · 8 wks
Trend↓ 56%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$118Knext 8 weeks · 28,502 units forecast x 2.2% fallout x $189 · projected, not measured
View calculation details →
The other dollar readings
$226,929the headline · window pace x 365, annualized
$10,596newest half, rework 40¢
$34,816full 8 wks incl. rework at full value
$191,704scrap ahead, NOT measured · 46,315 released units x 2.2% scrap x $189
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run up 31% vs the prior 4 wks · volume -67%
No baseline of its own yet · judged against the plant
Newin all 8 weeks scored
Last time: on hold 2026-08-07 03:29Z
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
10201584
SCREW,FLANGE,HEX HEAD · Plating - Burn
#14 Priority1.9% affected volumeImprovingLine 37
$416K
The brief: why it ranks here, the trend, and your decision
Why this is #14
Below the plant's bars on every rule.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$416K / yr$63,896 / 8 weeks
Evidence326 parts rejected$196 / rejected part · 47 tickets · 8 wks
Trend↓ 59%Failures are falling · parts rejected · recent half vs prior half of the window
Forecast impact$116Knext 8 weeks · 31,668 units forecast x 1.9% fallout x $196 · projected, not measured
View calculation details →
The other dollar readings
$416,465the headline · window pace x 365, annualized
$18,620newest half, rework 40¢
$63,896full 8 wks incl. rework at full value
$188,384scrap ahead, NOT measured · 51,461 released units x 1.9% scrap x $196
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 10% vs the prior 4 wks · volume -54%
No baseline of its own yet · judged against the plant
What changedweighted loss $92K to $64K · 17 fewer reject events · direction -1% to -10%
Down 4in all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
108A-114-1102
8X114,MIT,ST(OD) CHR POL · Plating - Scratch/Pre-Handle
#15 Priority1.7% affected volumeImprovingLine 91
$339K
The brief: why it ranks here, the trend, and your decision
Why this is #15
Below the plant's bars on every rule.
Loss trend, last 8 weeks
No single step-up in this window: the loss is roughly level week to week.
Financial impact$339K / yr$51,959 / 8 weeks
Evidence199 parts rejected$261 / rejected part · 16 tickets · 8 wks
Trend→ 14%Roughly level · parts rejected · recent half vs prior half of the window
Forecast impact$115Knext 8 weeks · 26,045 units forecast x 1.7% fallout x $261 · projected, not measured
View calculation details →
The other dollar readings
$338,661the headline · window pace x 365, annualized
$24,021newest half, rework 40¢
$51,959full 8 wks incl. rework at full value
$186,538scrap ahead, NOT measured · 42,323 released units x 1.7% scrap x $261
One problem, several readings, each labelled with its scope; the backward-looking ones tie to the dollar. The projection is separate and forward-looking.
The detailed reasoning
Loss per unit run down 12% vs the prior 4 wks · volume -3%
No baseline of its own yet · judged against the plant
What changedweighted loss $59K to $52K · 3 fewer reject events · direction -1% to -12%
Down 8in all 8 weeks scored
Not covered in tonight's AI pass; the numbers are the engine's own.
Your decision
Add a reason (optional)
Why?
Family problems, across partsthe same rule table, grouped by defect description instead of part · catches one root cause bleeding through many small parts
Family · signal on quantity
Fab - Setup - Benderacross every part carrying this defect
Signal: Worseningsix straight rising weeks
$1,049,619above the plant median of $190,691
A family has no release denominator, so there is no rate leg and the
chronic rule cannot fire here; the dollars and the signal stand on their own.
Gut vs evidencethe same parts, three orders: biggest dollars, strongest evidence, biggest volume ahead · the gap between the lists is the point · click any part to see why it ranks where it does
Explain Gut vs Evidence
By cost to the business · the gut list. Every part ranked by total dollars
lost in the analysis window, biggest first. A fair size measure, but it cannot tell a
problem that is growing from one that is already going out.
By evidence · worsening first. The same parts re-ranked by the priority
table, and the fallout rate drives the order: worsening parts lead, rate-first with
dollars breaking ties, and everything else follows the same way, so an improving part
can still appear here, just lower down. The big number on each row IS the fallout rate;
the chip is the signal that placed it. This is the order the tool stands behind.
By what is coming. A projection: the volume customers have already released,
priced at today's scrap rate. It looks forward, so it is projected money, not measured.
Reading them together. Follow one part across the three columns; a big gap is
the story. Red =
big dollars, but already healing on its own, so the evidence list drops it.
Green = getting
worse while the dollar list underrated it.
Yellow = small
dollars today, but a large released volume ahead.
How the ranking picks the order
Each part gets four measured numbers, all computed from the ledger: Impact
(what a year of this problem costs at the current pace), Rate (parts rejected out
of parts released), Signal (did something actually change, or was it just a noisy
week), and Forward (what the volume ahead would cost at today's rate).
Four stated rules turn those into the order, and the fallout rate drives it
(re-specced with the OpEx team, Aug 2026). Chase now: worsening, with a fallout
rate above the plant bar (3.0% this run); a
worsening part with no usable rate still chases when its dollars clear the plant median
($440,827), because a ton of money gets chased
regardless. Chase, chronic: steady, rate above the plant median rate, impact in
the top quarter ($1,207,372). Watch: big
enough, but healing, a one-off, or too new; a one-off above the
$15,000 bar is flagged for
acknowledgement rather than quietly de-ranked. Fine: everything else, still
ranked, never hidden. Within every tier, rate orders the list and dollars break ties.
The bars are recomputed from each run's own parts every night. Python does all the
math; there is no single blended score.
04-31798-026-POL5" PIPE-CRV STACK 439 SST NiCR · POL
Watch
Improving
3.3%
$351,420
$704,772
↓1 vs 8 wk
11
04-31813-026-POL5" to 4" PIPE-30DEG CRV 439SST
Fine
Improving
5.4%
$192,728
$386,515
↑1 vs 8 wk
12
107A-060-11027X60,MIT,ST(OD) CHR POL
Fine
Improving
5.0%
$162,653
$326,200
↑1 vs 8 wk
13
108A-114-11028X114,MIT,ST(OD) CHR POL
Fine
Improving
3.5%
$192,905
$386,869
↑2 vs 8 wk
14
04-31799-014-PLT4" PIPE-CRV STACK 439SST NiCR
Fine
Stable
3.4%
$434,367
$871,121
↓10 vs 8 wk
15
10201584SCREW,FLANGE,HEX HEAD
Fine
Improving
3.0%
$217,347
$435,889
↓1 vs 8 wk
the ledger holds 14 of the 26 weeks this horizon asks for; the view fills out as history accumulates
By cost to the business · the gut listtotal window dollars, biggest first
By evidence · worsening firstthe whole ranking: worsening leads, the rest follows · ordered by the priority table, not the dollars shown
By what is comingreleased volume ahead, priced at today's rate · carries 1 extra part(s) already covered by open projects: their volume is still at risk
big dollars, already healing on its owngetting worse; the dollar list underrated itsmall dollars today, big released volume ahead
Three readings of the same plant. Dollars look backward at totals, evidence looks at
direction of travel, and the third column multiplies how sick each part is per unit by the volume your
customers have already released. A part that is quiet in the first two and loud in the third is the one
worth catching early.
The verdict board. Every call you make on a card lands here:
drafts in review (a worker AI drafts a first move for top parts, an independent reviewer
plus a number check gate each draft), what you put on hold, what you assigned, and what
you dismissed. Nothing is ever auto-approved.
0 waiting on your call0 held for fix
No verdicts and no drafts on this run yet. Rule on any card on the Overview and it lands here.
Your decisions so farevery verdict is stored with a timestamp · a sent-back part stays out of future drafting, so this ledger is its paper trail
The six settingsthe only opinions in the system, shown with tonight's values · open one for the full story
Every number on this portal is measured from the ledger except these six
settings. The dollar bars are recomputed from each run's own parts every night, so no threshold
was hand-picked, and this page always shows exactly what tonight's run used. Changing a setting
is a reviewed change, not a slider.
Money windowtrailing 8 weeksEvery dollar on a card reads the same 8 trailing weeks, so the figures always multiply out from each other.
The chart, the window total, and the annualized opportunity all read
this many trailing weeks. Shorter reacts faster but judges on fewer reject lines per part;
the nightly signal already fires within one run either way. The approved options are 8, 4,
or 2 weeks.
Chase bars3.0% rate · $440,827 tonightChase now needs a Worsening signal and a fallout rate above the plant bar; a huge dollar figure still chases even without a rate.
The fallout rate drives (re-specced with the OpEx team,
Aug 2026): a worsening part chases when its rate (rejected units out of released units)
is above the 50th percentile of the plant's own rates,
3.0% tonight. The dollar leg is the backstop:
a worsening part with no usable rate, or a rate under the bar, still chases when its
annualized loss clears the plant median ($440,827
tonight), because a ton of money gets chased regardless. Both bars recompute nightly
from the run's own parts.
Chronic bars$1,207,372 + 3.0% tonightA steady bleeder is chased when its yearly impact is in the plant's top quarter and its reject rate is above the plant median rate.
Both legs must hold: impact in the top quarter of the plant's
annualized losses ($1,207,372 tonight) AND a reject
rate above the plant's median rate (3.01% tonight).
No rate available means no chronic claim, never a guess.
Big one-off bar$15,000A one-time event above this size is flagged for acknowledgement instead of being quietly de-ranked.
A two-hour line-down that spikes one day should stay quiet; a
fallout across two or three shifts should not (the OpEx team's ask, Aug 2026, with
roughly $15,000 as the working size). A part whose signal reads one-off AND whose worst
single week cost at least this much shows a "Big one-off, acknowledge" flag: confirm
what happened, then dismiss it with a note or assign it. Changing the bar is a reviewed
change.
Signal windowtrailing 26 weeksHow much history the tool reads when it asks: did something actually change, or was it just a noisy week?
Stage one looks for the classic factory patterns in those weeks: a
streak (6 rising or falling in a row), a step (9 on one side of the part's own average), or
one freak week standing alone. Stage two, when no clean pattern shows, asks whether the
count of failed parts truly changed between the first half and the second, beyond luck, so
a steady climb with one dip still reads as the climb it is. Under six weeks of history:
"too new", never a guess. The patterns are the published control-chart rules (Western
Electric 1956, Nelson 1984); the halves check sits at the textbook 95% bar, not a dial
anyone tuned.
Methods and sourcesnamed, not inventedEvery method on this portal is published; none of it was invented here.
Pareto by cost (Juran) picks the vital few; cost-of-poor-quality
accounting (Juran/ASQ) prices them; Western Electric (1956) and Nelson (1984) control-chart
rules read the signal; and the rule table itself is the AIAG-VDA Action Priority shape that
replaced multiplied risk scores in 2019. The full trail with links is kept in the project's
research notes.
Track recordthe tool grading its own past calls; nothing fitted after the fact
How it grades itselfnot measured yetEvery past call is recorded before its outcome exists, then graded against what the ledger did next.
Not measured. This comparison is not yet a fair test of the ranking, so no performance claim is made here. The buckets below show how the data is distributed, not that one method beats another.22 calls graded across 2 weeks. Each one was recorded before its outcome existed, then measured against what the ledger did over the next 4 weeks. Loss rate is dollars per unit run, so a part cannot look worse just for running more volume.
The testwhen the two methods disagreed, whose picks carried the money
Picked by
Parts
Calls
Forward loss rate
Forward cost
evidence only
2
2
$149.74 per unit
$106,794
dollars only
4
6
$200.53 per unit
$50,974
This drops every part both lists agreed on and keeps only the disagreements, which is the cleanest way to ask whether ranking by evidence beats ranking by dollars.
By bucketexploratory, not graded: this data is what remapped the board colors, so confirm it on fresh weeks
Bucket
Parts
Calls
Forward loss rate
Rate change
Got worse
confluence
10
16
$195.24
-11%
6%
dollars only
4
6
$200.53
-13%
17%
What this proves, and what it does not
It tests prediction, not repair
Nothing is being fixed while these calls are graded, which cleanly isolates one question: does the ranking see where loss is heading? Whether acting on it helps is the pilot's job, measured by the approve and send-back scoreboard instead.
It cannot flatter itself
Every call was stored in the runs table before its outcome existed. The grader reads history it cannot edit, so a bad result would appear here just as plainly as a good one.
Part baselinespick a part: how much of what it runs usually ends up scrapped, and what its own normal looks like
Scrap baseline13.9%3,540 of 25,531 units released in the window were scrapped
All rejects13.9%3,540 rejected units (scrap + rework) out of released
Dollar split100% scrapof $802,766 lost in the window; the rest is rework
Against its own historyOnly 4 week(s) of history, needs 26. Judged against the plant, not against itself.
The exact tableproduction rejects only (inventory adjustments held out) · newest first
What is not in this table: 273 inventory-adjustment rows
worth $661,941 were held out of every number on this portal. They post to the same
ledger but carry a reason code instead of a defect code, because they are stock corrections rather than production
defects. Scoring them would rank inventory write-offs as quality problems.
Costs post negative in the export; the metrics above and the SQL both use absolute values.
Click any column header to sort. The excluded / ignore-exception rows (currently 254) never enter any calculation.
KPI registrycomputed inside Azure SQL by the scheduled job · these are the numbers the portal shows
Each number: its stored value, the plain-English formula, and the exact SQL that produced it.
Anyone can rerun the SQL and get the same answer. Nothing here comes from the AI; the maker and checker only ever
receive these computed values.
Total reject cost, rolling window$2,542,502
Sum of the absolute value of reject_cost across scored rows in the rolling analysis window (last 8 weeks back from the newest row; production rejects only, inventory adjustments and incomplete rows held out). Costs post negative in the ledger, so we take absolute values.
Show the SQL
SELECT ISNULL(SUM(ABS(reject_cost)), 0) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output'))
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
True scrap cost (unrecoverable)$2,023,247
Sum of absolute reject_cost where item_type = 'Non-Rework'. Scrap means the part is gone.
Show the SQL
SELECT ISNULL(SUM(ABS(reject_cost)), 0) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output')) AND LOWER(LTRIM(RTRIM(item_type))) = 'non-rework'
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Rework cost (recoverable)$519,255
Sum of absolute reject_cost where item_type is anything other than 'Non-Rework'.
Show the SQL
SELECT ISNULL(SUM(ABS(reject_cost)), 0) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output')) AND LOWER(LTRIM(RTRIM(item_type))) <> 'non-rework'
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Scrap share of all loss79.6%
True scrap cost divided by total reject cost, times 100. Computed in one SQL statement.
Show the SQL
SELECT 100.0 * SUM(CASE WHEN LOWER(LTRIM(RTRIM(item_type))) = 'non-rework' THEN ABS(reject_cost) ELSE 0 END) / NULLIF(SUM(ABS(reject_cost)), 0) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output'))
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Reject events, rolling window1,058
Count of active rows that have both a posting date and a cost, the same rows the engine analyzes.
Show the SQL
SELECT COUNT(*) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output'))
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Average cost per reject event$2,403
Total reject cost divided by total events, in one SQL statement.
Show the SQL
SELECT SUM(ABS(reject_cost)) / NULLIF(COUNT(*), 0) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output'))
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Worst week2026-08-23$360,039
Group scored production rows by the export's week_of column, sum absolute reject_cost, take the biggest. Detail carries which week it was.
Show the SQL
SELECT TOP 1 SUM(ABS(reject_cost)) AS value, CONVERT(varchar(10), week_of, 23) AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output')) GROUP BY week_of ORDER BY SUM(ABS(reject_cost)) DESC
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Group coded rows (not uncoded) by defect_description, sum absolute reject_cost, take the biggest. Detail carries the family name.
Show the SQL
SELECT TOP 1 SUM(ABS(reject_cost)) AS value, defect_description AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output' AND posting_date > DATEADD(week, -8, (SELECT MAX(posting_date) FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0) AND posting_date IS NOT NULL AND reject_cost IS NOT NULL AND LOWER(LTRIM(RTRIM(entry_type))) = 'prod. output')) AND NOT (defect_code IS NULL OR LOWER(LTRIM(RTRIM(ISNULL(defect_description, '')))) IN ('', 'no code selected')) GROUP BY defect_description ORDER BY SUM(ABS(reject_cost)) DESC
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Active ledger rows2,328
Count of rows where the ignore-exception flag is off (excluded = 0).
Show the SQL
SELECT COUNT(*) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 0 AND sim_round IN (0)
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Rows excluded by the ignore-exception flag254
Count of rows personnel flagged out of the math (excluded = 1), IT's centralized-adjustment pattern.
Show the SQL
SELECT COUNT(*) AS value, NULL AS detail FROM dbo.reject_lines WHERE excluded = 1 AND sim_round IN (0)
computed 2026-08-27 00:03Z by run_pipeline.py · stored in dbo.kpis
Connected data
What this portal is reading right now. Every number on every view comes from here.
Database
Azure SQL · opexradar on sql-opexradar-wus2 (serverless, auto-pause)
reject_lines
2,328 active rows · 254 excluded by the ignore-exception flag
known_projects
3 tracker entries the checker cross-references
runs
23 saved runs · every output kept for the pilot comparison
kpis
14 numbers computed IN SQL by the KPI job · refreshed daily at 05:30 UTC by the Azure timer and on every pipeline run
Feed
mock reject-board export · the live version is an IT-owned push from Navision
Window
Jul 5 to Aug 26, 2026
Engine
m0_engine.py · all math in Python · 77 offline tests + an engine self-test + 2 live parity suites