Uniqlo owns no factories: how does it control quality?

Suppose you run a T-shirt brand in Binh Duong with its own sewing plant. Next year's orders exceed capacity by 100,000 shirts. An outside factory quotes below your unit cost. But the customer will return a shirt carrying your label, not the factory's. How do you outsource and still hold quality?

Fast Retailing outsources most of Uniqlo's production. This post looks at what the company says it uses in place of ownership (the Uniqlo and SPA case), then applies the idea to a make-or-buy decision at Vietnamese scale.

Make or buy: what should you compare besides the quote?

Compare the total cost of both options first, then add the strategic factors. That is the test in the make-or-buy framework of Lysons and Farrington (Lysons and Farrington, 2020). I have not been able to open the book, so what follows rests on the summary of the framework in this site's library.

By that summary, making in-house is sensible when suppliers are poor, technology must stay secret, spare capacity exists or the activity is a core capability. Buying is sensible when it is cheaper, capacity is short, a better supplier exists or flexibility matters (Lysons and Farrington, 2020). The price of buying is loss of control, dependence and loss of a technical edge. The price of making is upfront capital and difficulty changing direction (Lysons and Farrington, 2020). The common trap is comparing a purchase price with a unit cost and forgetting transport, stock, import duty and administration (Lysons and Farrington, 2020).

This post is about the line that says "loss of control". When you outsource, you give up the right to give orders straight to the line. What you need to buy back is a different mechanism.

How does Fast Retailing say it holds quality at partner factories?

The public answer is on Uniqlo's business-model page, last updated on 9 March 2026 (Fast Retailing Co., Ltd., 2026). It is the company describing itself. These are the six things the company says it does.

1. Fix materials a year ahead. R&D, merchandising, marketing and materials-development teams hold a concept meeting a year before launch to settle the designs, materials and range for the season (Fast Retailing Co., Ltd., 2026).
2. Negotiate directly with material makers. The page gives two examples: working with Kaihara to source denim to specific spinning standards and dyeing specifications, and partnering with Toray on the fibres in the HEATTECH range (Fast Retailing Co., Ltd., 2026).
3. Order in large lots. Many products are made in lots of roughly one million units (Fast Retailing Co., Ltd., 2026).
4. Put your own people near the factory. Production staff, who manage quality and progress, and takumi (skilled artisans) work from offices in Shanghai, Ho Chi Minh City, Dhaka, Jakarta and Bengaluru. Office representatives visit factories every week to resolve problems and manage quality on the factory floor (Fast Retailing Co., Ltd., 2026).
5. Pass on technique. The page quotes a takumi on teaching dyeing technology at partner factories, and says such visits and technical advice build thorough quality control and trust (Fast Retailing Co., Ltd., 2026).
6. Audit regularly and meet yearly. The company holds an annual conference with factory managers. It audits human rights, working environments and management systems at all garment factories, plus core fabric mills and core spinning mills for Uniqlo cotton products and core auxiliary material factories. Low-scoring factories get help to improve (Fast Retailing Co., Ltd., 2026).

The page gives the scope of the audits in item 6 as human rights, working environments and management systems. It does not mention finished-goods inspection. On my reading, the page ties product quality control directly to items 4 and 5 only. Items 1 to 3 sit under its R&D, merchandising and materials sections.

Lan (2024) judges that the company keeps its supplier list short and gives each supplier larger orders, which strengthens the relationship (Lan, 2024). The supplier figures in Lan's text are inconsistent; see the going-deeper section.

What does the SPA closed loop connect?

SPA is a model in which one company coordinates design, contracted production, distribution and retail The SPA closed loop (Uniqlo). Fast Retailing says product development and improvement stem from customer feedback (Fast Retailing Co., Ltd., 2026). Lan describes SPA as folding design, purchasing, production and sales into a shorter chain (Lan, 2024).

The loop carries two kinds of feedback. The first is quality: the customer centres receive more than 39 million pieces of feedback a year and share them with product departments and stores. The page says that when customers raise quality concerns, the production department responds swiftly and makes improvements where needed (Fast Retailing Co., Ltd., 2026). The second is quantity. Merchandisers watch sales through the season and instruct production to rise or fall. The inventory team checks stock at each store daily. At season end merchandisers and the marketing department pick the items to keep selling and the timing of price changes (Fast Retailing Co., Ltd., 2026).

Lan argues that this structure reduces the bullwhip effect because the company controls almost every step (Lan, 2024). That is an argument, and this post does not measure the effect.

Widia and Hapsari (2025) ran a regression on Fast Retailing's quarterly reports for 2014 to 2023. According to the abstract I read, inventory turnover has a positive effect on revenue growth, market share and profit margin (Widia and Hapsari, 2025). The input variables are operating profit, store expansion and inventory turnover; none measures product quality.

What should a T-shirt brand in Binh Duong keep, and what should it buy?

Do two things in order: work out the total cost of the overflow, then split the stages into "keep" and "buy". Everything below is hypothetical. The firm "Lua Moi" and every number are mine, invented.

A hypothetical example. Lua Moi sells 400,000 shirts a year. Its own plant can make 300,000. For the other 100,000 it has two options: expand the plant, at a unit cost of VND 62,000, or use a partner factory quoting VND 55,000. On the quote, buying is 11 per cent cheaper. Add the items the total-cost test asks for and the picture changes.

The gap shrinks to VND 1,200 a shirt, about 1.9 per cent, or VND 120m across 100,000 shirts. Suppose 3 per cent of the lot, 3,000 shirts, has to be scrapped outright, while the own plant scraps none. That costs 3,000 times 60,800, or VND 182.4m, which wipes out the saving. In this example, one bad lot outweighs the gap in quoted price.

The first two rows with a substitute (fabric, sewing the overflow) take their ideas from how Uniqlo describes itself: materials fixed in advance, your own people at the factory, regular visits. The written defect standard in the last row is my suggestion; Fast Retailing's page does not mention it. This suits a firm that has someone qualified to station at the partner and enough volume to be a priority. It does not suit a firm with small, shifting orders and nobody able to sit at the factory: there, a 1.9 per cent gap is thinner than one bad lot.

Going deeper: what each source tells us, and what I could not check

I have not read Lysons and Farrington's textbook, only the library summary. I cannot give the origin or the critiques of the make-or-buy framework.

Fast Retailing's page describes the approach without using the word "SPA". That term comes from academic writing: Lan says the Gap group first adopted the model (Lan, 2024). I read Lan's PDF through an automatic converter, and some passages came out garbled, so I use only the passages that read cleanly.

Lan sets out the model's limits. By cutting back on middlemen, the company carries risk alone: external risk (environment, customer demand) and internal risk (reliance on one production region, overseas operations, exchange rates) (Lan, 2024). Lan uses the temporary closure of plants in China in early 2020 as an example and argues Uniqlo should add suppliers elsewhere. Direct stores are also costly: high rent, frequent renovation, and all expansion funded by the company itself (Lan, 2024). In the parts I could read, Lan does not give the cost of the production-office and takumi system.

Two things deserve doubt. First, Lan's text says the company keeps about 70 suppliers, then later says China alone has more than 200 (Lan, 2024). The two figures do not agree, so this post uses no supplier count as a fact. Second, Lan states as a judgement that SPA guarantees quality at the lowest possible cost, and the parts I read contain no quality measure to test it against (Lan, 2024).

The last limit belongs to this post. Fast Retailing's page is the company describing itself and gives no defect rate or inspection result. I have not read the company's integrated report or checked the works Lan cites. The company describes a set of mechanisms in place of ownership. These sources do not say how well those mechanisms hold quality.