Case study · MX · Automotive & machining
Injection-Molding Plant
Scrap down 31% · $210k/yr verified savings
The results
Daily walks catch oil on floors and damaged guards early; recordables fell a third year over year.
The regrind A3 plus faster leak closures took plant scrap from 4.8% to 3.3%.
Fewer press-down hours and fewer quality holds lifted on-time delivery to 97%.
Scrap reduction, purge recovery, and energy-saving retrofits, all finance-verified in the ledger.
Bilingual capture removed the wall between the floor's Spanish and the report's English — submissions run 2.5× the old program's best year.
“We argued about scrap for two years with theories. The board settled it in six weeks with photographs, and the A3 fixed a regrind station nobody had ever blamed.”
The starting point
The plant molds appliance and automotive plastic components on 42 presses across three shifts. Scrap ran at 4.8% — everyone knew it was too high, and every month produced a different theory: resin lots, mold wear, operators, humidity. Meetings argued theories; nobody owned evidence. Downtime firefighting consumed the maintenance crew, and the same hydraulic leaks reappeared weekly.
Improvement ideas existed — the floor speaks Spanish, the corporate reporting ran in English, and somewhere between the two, most ideas simply evaporated.
What changed, step by step
One connected loop: audit, defect, fix, root cause, kaizen, standard, recognition. Open any step to see what it did at this plant.
Step 1 · Audit · 5SDaily walks that see all 42 presses
The plant mapped zones around press clusters, material handling, and the regrind area. Shift leads walk daily with phones; oil weeps, damaged thermocouple cables, and open material containers become photographed defects in seconds. Coverage tracking ended the 'we only audit the presses near the door' pattern within a month.
Step 2 · DefectsLeaks and specks stopped being anecdotes
Every scrap spike, hydraulic leak, and contamination find goes on one board. The AI's clustering showed the black-specks rejects concentrated on parts running regrind blends — across different presses, different molds, different shifts. The 'it's the operators' theory died in front of a chart.
Step 3 · Fix-PlanMaintenance went from firefighting to a ranked queue
Hydraulic leaks now carry severity, photos, and owners; the AI drafts fix-plans from the press's history — including the seal kit part number used the last time. Repeat leaks per month fell by half, and the maintenance crew reclaimed enough hours to start preventive work again.
Step 4 · Root cause · A3The black-specks A3 found the regrind handling gap
The specks became an A3 with the clustering chart as case context. The fishbone crossed Material and Method; 5-Whys ended at the regrind station: open gaylords, no purge discipline between dark and light regrind, and a grinder screen past its replacement interval. Countermeasures — covered containers, a purge-and-label standard, screen-change schedule — were verified over eight weeks: specks-related scrap on the affected part families fell by two-thirds.
Step 5 · Kaizen · savingsIdeas in Spanish, ledger in dollars
Operators submit ideas in Spanish from the floor; LeanOS translates for the review board and the corporate report. Purge-waste recovery, a nozzle-insulation retrofit, and packaging-density changes lead the ledger: $210,000 a year, finance-verified. Idea submissions run 2.5× the old program's best year — the language wall was the bottleneck all along.
Step 6 · OPL → Skill matrixThe regrind standard trains every new hire
The A3 produced a photo-first OPL for regrind handling and purge discipline, in Spanish. Acknowledgement feeds the skill matrix; material handlers can't be scheduled to the regrind station until their row is green. The countermeasure became part of onboarding, not a memo.
Step 7 · LeaderboardThree shifts, one scoreboard
Walk streaks, closed defects, and verified ideas earn points, and the three shifts watch each other. Shift C — the perennial 'nobody tells us anything' shift — leads the current quarter. The plant manager starts the daily meeting with the scoreboard because the arguments got shorter when the numbers got public.
More case studies
All 9 →Illustrative deployment story: a composite of typical LeanOS rollouts; company and names are fictional.