What's a Good DHU Rate? Garment Factory Benchmarks 2026
"What's your factory's DHU?" is the second question every buyer asks after "what's your monthly capacity?" The expected answer is a single number. If you say 12, they go quiet and start asking different questions. If you say 4, they ask when you can ship a sample order.
This guide answers the question buyers do not ask but should: what is a reasonable DHU for a CMT factory in 2026, by garment type, by country, by line maturity?
The DHU formula — quick refresher
DHU = (Total defects found / Total garments inspected) × 100.
Critical distinction that trips up new QC supervisors: DHU counts defects, not defective pieces. A T-shirt with a broken stitch, a wrong size label, and a fabric hole is three defects in one garment. All three count.
Example: a checker inspects 250 garments, finds 20 defective pieces with 35 total defects. DHU = 35/250 × 100 = 14. Not 20/250 = 8. Twenty out of 250 is the rejection rate — a separate metric.
The universal benchmark scale
| DHU range | Classification | What it means |
|---|---|---|
| ≤ 5% | EXCELLENT | World-class. Most lots ship straight to brand without rework. You earn buyer preference and repeat orders. |
| ≤ 10% | GOOD | Industry standard for established CMT factories with mature QC. Buyer inspections pass routinely. |
| ≤ 15% | ACCEPTABLE | Common for newer factories, fashion knits, complex garments. Buyer inspections pass but margin is thin after rework. |
| ≤ 25% | POOR | Rework eats your margin. Buyers complain. You are at risk of failed inspections. |
| > 25% | CRITICAL | Lot rejection becomes routine. Margin is negative after sorting and discount penalties. |
These thresholds are what we use in our garment-dhu-calculator npm package and what most published industry references converge on. Some sources use slightly different bands (e.g. ≤7% Excellent, ≤12% Good) but the structure is the same.
DHU by garment category
"Good DHU" depends heavily on what you are making. A 10% DHU on a complex tailored jacket is excellent. A 10% DHU on a basic T-shirt is unacceptable. Approximate 2026 benchmarks from CMT factories we cross-checked with:
| Garment type | Excellent | Good | Acceptable |
|---|---|---|---|
| Basic T-shirt (cotton, no print) | ≤ 3% | ≤ 6% | ≤ 10% |
| Printed T-shirt / polo | ≤ 5% | ≤ 8% | ≤ 12% |
| Hoodie / sweatshirt | ≤ 6% | ≤ 10% | ≤ 14% |
| Formal shirt (woven, button-front) | ≤ 5% | ≤ 9% | ≤ 13% |
| Trouser (5-pocket denim) | ≤ 8% | ≤ 13% | ≤ 18% |
| Tailored jacket / blazer | ≤ 10% | ≤ 16% | ≤ 22% |
| Outerwear (3-layer technical) | ≤ 12% | ≤ 20% | ≤ 28% |
| Lingerie / swimwear | ≤ 6% | ≤ 10% | ≤ 15% |
| Activewear (performance knit) | ≤ 5% | ≤ 9% | ≤ 14% |
| Children's wear | ≤ 4% | ≤ 7% | ≤ 11% |
The pattern: more operations = higher DHU. A jacket has 60+ operations vs a T-shirt's 8-12 operations. If each operation has a 0.5% defect rate, the cumulative DHU is much higher on the jacket. This is math, not a quality problem — but it means you cannot directly compare DHU between garment types.
DHU by country / region
Country averages are very rough because factory-to-factory variation within any country is enormous. A top Bangladesh factory beats a poor Sri Lanka factory easily. But for orientation:
| Country / region | Typical DHU range (basic knit) | Notes |
|---|---|---|
| Vietnam | 4–8% | Strong process discipline, mature workforce |
| Bangladesh (RMG, top tier) | 5–9% | Compliance-driven, well-trained at top factories |
| Bangladesh (mid-tier) | 10–15% | Most volume; variable QC investment |
| India (Tirupur knit) | 6–10% | Strong knit cluster, lower for tees, higher for fashion |
| India (Delhi/Noida woven) | 10–18% | Complex garments, more variation |
| Cambodia | 8–14% | Skilled but workforce turnover affects consistency |
| Nepal | 10–15% | Younger industry, growing |
| Sri Lanka | 4–8% | Premium positioning, strong process |
| Ethiopia | 15–25% | New workforce, training-curve heavy |
| China (mid-grade) | 3–7% | Strong process culture, declining workforce |
These are not official numbers — they are based on our cross-checks with buyer QA teams and other factory owners over the last two years. Treat as orientation, not gospel.
DHU by line maturity
The single biggest variable inside one factory is line maturity. A line that has been running the same style for three months produces dramatically lower DHU than a line that just started a new style yesterday.
| Line stage | DHU multiplier vs steady state |
|---|---|
| Day 1 of new style | 3–5× higher than steady state |
| Week 1 of new style | 2–3× higher |
| Week 2-3 of new style | 1.5× higher |
| Week 4+ (steady state) | Baseline |
This is why changeover frequency matters more than absolute DHU. A factory with 10 styles per month and 12% average DHU is probably running better operations than a factory with 2 styles per month at 8% DHU — because the first factory's steady-state DHU after the learning curve is likely 4-5%, while the second's is 6-7%.
If you are a buyer evaluating a factory, ask for DHU broken out by week-of-style. If a factory does not track this, that itself tells you something about their QC maturity.
What drives DHU — in order of impact
From our own data (see the full DHU reduction case study):
1. Machine condition (single biggest factor)
Of our 35 machines, 14 had issues we did not see until we audited: bad timing, dull needles, oil leaks. After fixing all 14, our skip-stitch rate dropped 60% and puckering dropped 45%. Machine maintenance is the highest-ROI DHU intervention.
2. Operator skill matrix
Putting an overlock-skilled operator on a single-needle machine produces 3-4× the DHU of someone trained on that machine. Skill matrix discipline is invisible until you measure it.
3. Pre-production sample issues that did not get fixed
If the sample had a measurement issue and you "fixed it in production", that issue is going to show up at scale. Every PP sample defect is a future DHU contribution.
4. Fabric inspection gate
Fabric defects (slubs, weaving issues, shading) account for 15-25% of our DHU. Investing in a 4-point fabric inspection system (separate post — guide here) catches these before they hit the line.
5. Inspector calibration
Two inspectors looking at the same 100 garments will find different DHUs unless you calibrate them. We run a monthly calibration where 5 inspectors check the same lot — the spread reveals whose count to trust.
Free online DHU calculator
The DHU calculator on our tools page runs in your browser — enter defects + garments inspected, get DHU + classification + benchmark badge instantly.
For factory ERP integration, we open-sourced the math:
- npm:
npm install garment-dhu-calculator - PyPI:
pip install garment-dhu-calculator
Both packages support breakdown by defect type, multi-checker daily aggregation, and benchmark classification. MIT licensed, 27 inline tests cross-verified against OnlineClothingStudy's worked example.