Roving quality checks from a spare supervisor
Quality

Roving quality checks from a spare supervisor

Financial impact

$4,803USD/ yr EBITDA

Operational impact

~2faults caught / day
medium confidence

The math

Residual defects after upstream cards4,480

112000 garments × 10.0% rate × (1 - 60% upstream)

End-of-line cost avoided$4,803/ yr

residual defects × 13.4% catch rate × $8/defect

Capex

$200

Payback

1 months

Difficulty

2 / 5

Time to implement

~2 weeks

Faults already prevented by the two cheaper fixes
%
Share of leftover faults the walking inspector catches at the machine
%
Money saved per fault caught early instead of late
$
One-time cost of the inspector kit
$

Also uses annual garment volume 112,000, in-line defect rate 10%. Edit on the dashboard.

Description

Today the factory only checks garments for faults at the very end, in a separate QC room (quality control, the check for stitching faults) after everything is sewn, pressed, and finished. The problem is timing: a fault made early, say a skipped stitch at the fifth of twenty steps, has ten more steps sewn on top of it before anyone sees it, so fixing it means unpicking (pulling the stitches back out) through all of them. And there is no fault-tracking on the floor at all: no log sheets, no tally board, no checks out on the line itself, just a thread-trimmer glancing at finished pieces. Faults are only caught once they are already expensive to fix.

Evidence

  • [00:04:08] Striped-polo supervisor pacing the floor empty-handed except for a phone and pen: the one to redirect
  • [00:00:48, 00:02:49, 00:03:12] Yellow-polo supervisor touching bundles, kneeling, re-setting machines: the real dispatcher, leave him in place
  • [00:01:04, 00:02:21] Two more supervisors pacing hands-off with no defined role
  • No on-floor quality station, defect bin, or tally board visible; garments are only checked at the end, in the QC room

Quotes

No formal defect tracking system visible.

Assessor

Deep dive

Expand for detail

Load-bearing, card breaks if wrong

One of the four floor supervisors is genuinely spare and can be moved to full-time checking.

The walkthrough shows four supervisors pacing the floor, and three of them (the white button-down, the striped shirt, and the striped polo) are doing overlapping, hands-off watching. The whole card rests on being able to pull one of those three into a dedicated checking role without leaving the floor short on direction. If all four are actually load-bearing, if they only look idle because the floor is chaotic and they are firefighting, then pulling one starves the sewing lines and you lose more in output than the $5K of quality you gain. What it looks like on the floor if this is wrong: after the switch, the 41 sewing machines drift out of sync, half-finished bundles (stacks of cut pieces that move through the line tied together) pile up at some stations while others sit idle, and someone keeps dragging the new inspector back to route work.

The two cheaper fixes ahead of this one get done first, and this inspector adds catches on top of them.

This card only counts the faults left over after Card 6 (a printed step-by-step work card at each station) and Card 7 (a daily tidy-and-standardize routine: sort, set in order, shine, standardize, sustain) have run and shrunk the fault pool, and it assumes a real share of what the inspector then catches is genuinely new, not faults those two cards would have caught anyway. Two things have to be true: the two cheaper fixes actually go first, and the inspector adds catches on top rather than re-catching the same faults. Run the inspector first, on a floor with no written standards and no tidy-up, and they get swamped re-teaching basic tension and alignment faults a $5 laminated card would have prevented, so the walk becomes a permanent bottleneck instead of a sampling loop. Run them last but with total overlap, and the $5K was never really there. What it looks like on the floor if this is wrong: the inspector is stuck at one or two machines all shift re-teaching the same fix and never finishing a loop, or the logs fill with catches while the final inspection room sees no drop in faults.

Supportive, card weakens but survives

The way operators are paid can absorb being stopped mid-task to fix a stitch.

If operators are paid per finished garment (piece-rate) rather than per hour, then stopping one mid-bundle to unpick and re-sew a seam costs that operator wages, and people fight it or hide faults to protect their pay. The premise is that the factory can smooth this over, with a small quality bonus or by checking completed pieces off the table instead of interrupting a live seam, without triggering a revolt or teaching people to bury their mistakes. What it looks like on the floor if this is wrong: operators arguing over every red mark, or faults quietly stuffed into the middle of a bundle where the inspector will not see them.

The supervisor picked for the job takes it as a step up, not a demotion.

In a garment factory, going from "line boss" to "the person with the clipboard" can feel like being knocked down a rung. The premise is that the factory owner frames it as a promotion to Quality Lead, so the chosen person owns it instead of quietly resisting or quitting. What it looks like on the floor if this is wrong: the clipboard ends up left on a table, the loops get shorter and rarer, and within a couple of weeks the person has drifted back to their old habits.

The wrong supervisor gets picked, and the sewing lines starve

Risk

The tempting mistake is to pick the yellow-polo supervisor at [00:00:48] because he looks the most hands-on. He is exactly the wrong one: he is the floor's hidden production-control system, the person who routes work and clears bottlenecks. Hand him a clipboard and the sewing lines lose their dispatcher; he will drop the clipboard and go back to moving bundles within about 48 hours to fix the mess, and you are left with no inspector and a disrupted floor.

Early Warning

The clipboard is sitting abandoned on a table, and the "inspector" is back to carrying and routing bundles.

Mitigation

Pick the striped-polo supervisor at [00:04:08], whose day is already observational and administrative, not the dispatcher.

Stake

The full $5K benefit, plus a real hit to overall factory output while the lines run without their dispatcher.

Cramped aisles turn the walking checks into a second final inspection

Risk

The floor is tight, with bundles stacked in the walkways. If the inspector cannot physically squeeze between the machines, they default to standing at the end of the row and checking finished bundles as they come off, which is just another final inspection. The entire point of this card, catching a fault at the machine before ten more steps are sewn on top, is lost.

Early Warning

The inspector stands in one spot for more than 15 minutes at a time, receiving finished pieces instead of walking to the machines.

Mitigation

Do this card after Card 3 (removing the steel racks that block the aisles), and make "physically touch the work at the machine" a written rule of the role.

Stake

The saving per fault collapses from $8 toward $0, because the fault is being caught at the end anyway.

Without a written standard, every correction turns into an argument

Risk

If Card 6 (the printed work cards) is not yet in place, the inspector's corrections are just one person's opinion against the operator's. Operators argue, the line slows, and to avoid the daily conflict the inspector quietly stops correcting anything. The role becomes a walk that changes nothing.

Early Warning

Audible arguments on the floor, and the daily log shows the same fault from the same operators five days running with nothing changing.

Mitigation

Make sure the kit includes laminated "good versus bad" photo cards so a correction is objective: "it does not match the photo," not "I think it is wrong."

Stake

No behavior change at all; the faults keep flowing and the $5K never appears.

The faults were already going to be caught by the two upstream fixes

Risk

This is the card's biggest honest weakness. Cards 6 and 7 shrink the fault pool before the inspector ever sees it, so a large share of the 600 faults the inspector catches are faults those two cards would have prevented anyway. Stack all three benefits at face value and you are counting the same faults twice. The inspector's truly unique value is narrow: catching first-run faults in the opening hours of each new order, when operators are least practiced, and the behavior change from immediate feedback at the machine.

Early Warning

The inspector logs plenty of catches, but the fault volume at the final QC room does not actually fall month over month.

Mitigation

Re-measure the fault rate AFTER Cards 6 and 7 land, before committing to this as a full-time role. If the final-inspection fault count does not drop within a month of going live, redeploy the supervisor.

Stake

Up to the full $5K, if the benefit turns out to be entirely overlap with Cards 6 and 7.

A. Which of the four supervisors to pull

Three of the four supervisors are doing overlapping, hands-off watching, so the checking capacity is already on the payroll. The pick is everything: pull the administrative one, never the dispatcher.

SupervisorSeen atWhat he is doingBest guess at his real jobSafe to move to checking?
Striped polo, jeans[00:04:08]Walking the floor empty-handed except a phone and pen. Pure watching.Floor walker / status checkerYes, first choice. His day is already administrative.
Striped shirt[00:02:21]Standing in the sewing aisle, scanning. Not touching work.Unclear, a floaterMaybe. Watch him for a day first.
White button-down, khakis[00:01:04]Hands behind his back, quietly inspecting (again at [00:02:02] and [00:04:16]).Likely the owner, a senior manager, or a visiting buyerNo. Too senior and costly for an hourly loop.
Yellow polo[00:00:48]Touching bundles, kneeling [00:02:49], re-setting machines [00:03:12].Line boss / dispatcher, the real routing systemNever. He is the floor's hidden production control.

The rule: pull the striped polo. Never pull the yellow polo.

B. The walking route and the "traffic-light" check

The standard apparel practice for this is the "traffic-light" check: one inspector covers about 15 to 20 machines, visits each about once or twice an hour (roughly four times a shift), and pulls a small random handful of pieces (say five) at each stop. Each station gets a colored tag: green means it meets the standard, yellow means a minor fault to watch, red means it fails and the operator stops and fixes it. The inspector checks the everyday sewing specs: stitches per inch, seam width, seam allowance (how far the stitch line sits from the fabric edge), trim placement, and a measurement against the approved sample garment.

One 60-minute loop, run four times a shift:

  1. Cutting handoff: check incoming cut bundles are marked correctly, so a mismatched piece never enters the line.
  2. Fusing press (the Lastar / T. Star machine that heat-bonds stiffening into collars and cuffs): check the heat, the dwell time, and that the caked yellow residue is not transferring onto garments.
  3. Jacket line: the multi-layer seams here are the most likely to skip stitches, so check those first.
  4. Center sewing banks: sample one piece from about ten random machines for loose tension and puckering (fabric bunching along a seam).
  5. Specialty finishing: check buttonhole alignment on the Kaj buttonhole machines.

On a catch: spot the fault, hold up the laminated "good versus bad" photo, show the operator, log the cause, mark the bundle ticket with a red "R" for rework, and the operator unpicks it on the spot. Four 60-minute loops leave about four hours a shift for logging, training, and chasing repeat offenders.

C. Why convert a supervisor instead of hiring one

The floor already has four supervisors and three are doing overlapping, hands-off watching, so the checking capacity is already on the payroll, it just has no defined job. Converting one costs $200 in kit and zero new wages. Hiring a dedicated inspector would add a full salary for a benefit the card already caps at about $5K, which would not pay. The one hard rule: convert the right one (the administrative striped polo), never the dispatcher (the yellow polo).

D. The honest overlap: why we book $5K, not $8K

The early-catch principle (why catching a fault early is worth real money)

The rule of thumb in quality work is 1-10-100: a fault costs about $1 to prevent, about $10 to fix inside the factory, and about $100 once it reaches the customer. It is an order-of-magnitude guide, not a measured law, but the direction is solid. In sewing terms: a skipped stitch caught at the machine is a ten-second unpick; the same skip caught at the final inspection means unpicking through every step sewn on top of it; and shipped to a customer on a custom order it can cost the whole garment plus the account.

The catch-rate math (kept deliberately conservative)

  • Garments a year: about 112,000.
  • Baseline fault rate: 10%, so about 11,200 faults a year.
  • Removed first by Cards 6 and 7: about 60%, or 6,720 faults prevented before this inspector ever sees them.
  • Left over: 11,200 minus 6,720 = about 4,480 faults still reaching the floor.
  • Caught at the machine by the inspector: about 13.4% of the leftover = about 600 garments.
  • Saved per early catch: about $11.50 to fix late minus about $3.50 to fix on the spot = about $8.
  • Total: 600 x $8 = about $4,800 a year.

Why not the naive $8K

Ignore Cards 6 and 7 and let the inspector claim the whole 10% fault pool, and the number would look closer to $8K. That would be double-counting: the printed work cards and the tidy-up routine are already removing most of those faults, so booking them again here would count the same faults twice. The $5K is the honest, non-overlapping remainder. The inspector's truly unique value, the part Cards 6 and 7 do NOT cover, is two things: catching first-run faults in the opening hours of each new order (this is a build-to-order shop, meaning it only makes what customers have already ordered, so the product changes every order and operators are on a fresh learning curve every time), and the behavior change from immediate feedback at the machine (nobody wants a red tag). Everything beyond that overlaps, which is why this card sits at rank 11, not higher.

What we did not have in the corpus and would request from the client on day one. The aggregated gaps across all 15 cards become the engagement-letter ask sheet.

  1. 01

    A written breakdown of what each of the four supervisors actually does

    Owner

    Factory Manager

    We need to confirm from the factory, not just the video, that the striped-polo supervisor at [00:04:08] is genuinely spare and is not quietly handling some critical routing job off-camera. Pull the wrong person and the sewing floor loses its dispatcher.

  2. 02

    The factory's own fault records, broken down by station

    Owner

    Production Manager

    Does the factory already know its three worst stations for faults? If yes, the walking route targets them from day one. If no, the first route is a guess. The assessment notes no fault-tracking board or log sheets exist today, so this may have to be built from scratch.

  3. 03

    A stopwatch check of how long a late fix really takes versus an early one

    Owner

    Operations

    This proves or kills the $8 saving. Time unpicking a jacket sleeve seam right at the machine, then time unpicking the same seam after the jacket is fully assembled, pressed, and lined. The gap in labor and wasted material is the number the whole benefit rides on.

  4. 04

    Confirmation of how operators are actually paid

    Owner

    HR

    If operators are paid per finished garment (piece-rate), stopping them to fix a stitch costs them money and the friction will be high from day one. We need to know this to plan the change (a small quality bonus, or checking completed pieces off the table) before it becomes a fight.