End-Line DHU and the Alteration Loop: What Rework Really Costs an Indian Garment Export Unit (₹)

End-line DHU is defects found ÷ garments checked × 100. It counts defects, not bad pieces, so it can run higher than the share of garments that actually fail. Taken alone, a DHU of 6–8 looks cheap: a few hundred alteration minutes a day. The money goes in two places the DHU sheet doesn't show. First, standard minutes your operators spend re-sewing instead of producing. Second, the jump in risk at the buyer's final AQL inspection once outgoing defectives creep past 3–4%. A failed lot means a 100% re-check, a re-inspection fee in dollars and a ship date you may not make.

Every export unit in Tirupur, the NCR belt (Noida, Gurgaon, Faridabad) and Bengaluru runs an end-line table, a DHU board and an alteration pile. This piece builds the rupee chain from those three to the buyer's final inspection, and what a floor camera can and can't tell you.

By the Mama Editorial Team. Mama scopes camera and floor-visibility projects for Indian factories. Last verified: 13 October 2026, against published DHU definitions, the ISO 2859-1 / ANSI Z1.4 sampling plan as reproduced by inspection firms, and a published Bengaluru case study. Wages, SAMs and minutes are illustrative assumptions, not benchmarks.

Key takeaways

• DHU = (total defects ÷ garments inspected) × 100. A piece with three defects counts three times, so DHU and "% defective" are different numbers. Track both.
• An alteration costs more than the minutes of re-sewing. It also needs a second check, someone to carry the piece back, and a few standard minutes the line never earns.
• The buyer's AQL counts defective pieces, not defects. On a 3,201–10,000-piece lot at AQL 2.5 (sample 200, accept ≤10), a lot shipping 3% major defectives passes about 96% of the time. At 5% it passes only about 58% of the time (our binomial calculation).
• Cameras show the flow: alteration piles growing, pieces going back to the line, a checking table backing up. Defect type, root cause and DHU itself still come from QC data.

How end-line DHU is measured, and where it misleads

The standard definition: defects per hundred units = (total defects found ÷ total garments inspected) × 100. In that source's example, a checker inspects 250 garments and finds 35 defects on 20 faulty pieces, so DHU = 14.

On a typical export floor the number is collected at three points:

• In-line (roving) checks at critical operations, often run as a traffic-light system.
• End-line checking, a 100% check of every piece as it leaves the line.
• Internal pre-final audit in finishing or packing, sampled to the buyer's AQL before the buyer's own inspector arrives.

DHU misleads in three ways:

1. Checker-dependent. End-line DHU measures what the checker caught, not what the line made. A strict checker pushes DHU up; a tired one at 5:30 pm pulls it down.
2. Repaired pieces get counted twice. An altered garment comes back to the same table. If the re-check logs it as a new piece, the denominator inflates and DHU looks better than it is.
3. DHU ≠ what the buyer measures. The buyer's AQL inspection counts defective units in a sample. A line can hold DHU at 8 and still fail a final if its escapes cluster on one defect type the end-line checkers aren't catching.

What the loop actually costs: the alteration minute

Follow one altered piece:

1. The end-line checker finds the defect, tickets or marks it and drops it in the alteration bin.
2. A helper carries the bin back to the line, or to a separate alteration table.
3. Someone unpicks the faulty seam. On knits this is slow and can leave needle marks that turn an alteration into a reject.
4. The operation is re-sewn, often by the original operator, who stops sewing fresh bundles to do it.
5. The piece goes back to the end-line table for a second check.

Steps 3–4 come out of standard minutes the line was meant to earn. That is why rework hits efficiency twice, as we set out in where your sewing minutes go.

Indian evidence comes from a 2020 study of a suit manufacturer in Bengaluru (IJERT, Vol. 9, Issue 6). There, the daily rework rate ran above 7%. A traffic-light inspection system with checklists and root-cause analysis cut final-inspection DHU by 47.6%: the total defect percentage fell from 7.28% to 3.82%. Efficiency rose 11.72%, and the cost of poor quality fell from 1.79% to about 1.04%. One factory over a few months is a demonstration, not a benchmark. Note where the fix sat: sewing (40.95%) and seaming (35.02%) defects dominated final inspection, and they were fixed in the process, not at the checking table.

The step most owners skip: from DHU to the buyer's AQL

Most buyers inspect finished, packed goods to ISO 2859-1 (or its US twin, ANSI/ASQ Z1.4). The most common setting is general inspection level II, with AQL 2.5 for major defects and often 4.0 for minor ones. For a lot of 3,201–10,000 pieces, level II gives code letter L, which means a sample of 200. At AQL 2.5 the lot passes with 10 or fewer major-defective units and fails at 11. At AQL 4.0 the minor-defect limit is 14 defective units. QIMA notes critical defects are usually set at 0. Your buyer's manual overrides all of this.

The step few owners take: how likely is a lot to pass, given the share of defective pieces that escape your end-line? Binomial arithmetic on the 200-sample, accept-10 plan:

Our calculation from the binomial distribution, not a table from the standard; real inspections add inspector judgement.

Read it this way. Under about 3% escaped major defectives, a final is close to a formality. At 4%, roughly one lot in five fails; at 5%, it is close to a coin toss on every shipment. Above 6%, expect to fail. A line at end-line DHU 8 can sit in any of these zones, depending on how many defects share a piece and how many the end-line misses. So DHU alone can't tell you your shipment risk; the escape rate at your own pre-final audit can.

Worked ₹ example: one knit line, one export lot

Assume a Tirupur-style knit polo line. All inputs are illustrative; replace them with your own.

• 40 operators on an 8-hour (480-minute) shift; line output 1,200 pieces a day; garment SAM 12 minutes.
• Illustrative all-in labour cost ₹15,000 per operator per month, about ₹1.20 per attended minute (₹15,000 ÷ 26 days ÷ 480 minutes; derivation in what 1% of sewing efficiency is worth). Each state notifies its own tailoring-trade minimum wage, with DA usually revised twice a year; use yours.
• End-line DHU of 8: about 96 defects a day, on an assumed 70 pieces.
• Assume 3 minutes per defect to unpick, re-sew and handle, plus 0.5 minutes to re-check each altered piece.
• Illustrative CM contribution ₹25 per piece. Export lot of 5,000 pieces.

On a 10-line floor the alteration loop is real but tolerable, about ₹2.76 lakh a month, which is why it persists. The first two rows overlap: on fixed pay the wage minutes are spent anyway, so lost CM is the core loss; wages are extra only when alteration tailors or overtime do the rework. A single failed final costs about a quarter of the lot's CM before anyone talks about the ship date. Re-inspection after a supplier's failure is, in inspection-firm practice, "at the supplier's charge". The buyer often pays the inspector and recovers it by debit note, and the supplier also carries late-shipment penalties and air freight. We haven't priced that line; it varies by contract, and it can wipe out the order's margin.

At cluster scale: the Tirupur Exporters' Association reported about ₹45,000 crore of garment exports from Tirupur and Coimbatore in 2024–25.

What a camera can see, and what needs QC data

A floor camera watches flow and places. It does not inspect seams.

The value of the camera side is timing. The DHU board updates when someone fills it in. A growing alteration pile at 11 am is visible at 11 am, and it tells the floor manager which line to walk to. For defect-level inspection with fixed cameras, see catching quality problems on the line with cameras. Piles of returned pieces are also frozen cash; bundles on the floor prices that.

A note on people. Watching the alteration loop is about the flow of garments, not about who made the defect. Keep any camera reading aggregate: piles, trips, queues by line and hour, never operator rankings. Tell workers what the cameras are for. Our worker CCTV notice under DPDP has a template.

What to change on Monday

1. Split your DHU sheet into first check and re-check, so altered pieces aren't double-counted.
2. Log alteration minutes for one week: time out of the bin, time back to the table.
3. Run your pre-final audit to the buyer's exact plan (sample size, AQL, critical/major/minor). Track the defective-unit share, not just DHU.
4. Use the table above. If your pre-final escapes sit at 4%+, don't book the buyer's inspector yet.
5. Pareto the top three defect types per line and fix them at the operation, as the Bengaluru case did, not at the checking table.

Where Mama fits

Mama reads the cameras a factory already has, or a few it adds, and each morning sends the owner a short note on where hours and money leaked the day before. On a sewing floor that includes which lines built up alteration piles, when the end-line table backed up and how long finishing space sat tied up by re-checks. Mama looks at machines, piles and flow in aggregate, never at named operators. DHU and defect types still come from your QC sheets; Mama shows you when and where to look. Leave an email or phone in the form below and Mama will get back to you.

FAQ

What is a good end-line DHU for a garment export unit? There is no published national benchmark; it depends on product, buyer and how strictly checkers grade. What matters is the trend per line and the escape rate at your pre-final audit. In the Bengaluru study cited above, the final-inspection defect percentage fell from 7.28% to 3.82% (a 47.6% fall in DHU) after process fixes.

What is the difference between DHU and percentage defective? DHU counts every defect per 100 pieces checked, so one piece with three faults adds three. Percentage defective counts pieces with at least one fault. Buyer AQL inspections judge defective units, so track both.

Which AQL do buyers use for garments? Many use ISO 2859-1 / ANSI Z1.4 at general inspection level II, with 2.5 for major and 4.0 for minor defects and zero tolerance for critical ones. Your buyer's manual is the authority.

Who pays when a final inspection fails? Commonly the supplier pays for the re-inspection, often through a debit note against payment, and also carries late-shipment consequences. The exact terms are in your purchase order and the buyer's vendor manual.

Can cameras measure DHU automatically? No. A ceiling camera can't read seam quality. It can show the flow around defects: alteration piles, rework trips and checking-table queues. Defect counts and types come from your checkers or from a dedicated machine-vision station.