How Inditex uses AI and data: RFID and a demand-driven supply chain

KEY TAKEAWAYS Summary by the editors

1. Inditex's demand-driven model rests on data infrastructure more than on disclosed AI: item-level RFID, full stock integration, twice-weekly store deliveries and production largely close to its Spanish head office.
2. By 2016 Inditex had RFID running in 1,542 stores in 64 markets and planned to complete the Zara rollout by the end of that year, using chips inside reusable security tags.
3. In 2019 Inditex reported stock down 5% over nine months while sales rose 7.5%, which its chairman attributed to RFID and full stock integration.
4. Zara Try-on, a generative AI feature launched in December 2025, had recorded more than 7 million sessions in 43 markets by March 2026.
5. Inditex has not published how, or whether, machine learning drives its replenishment or production decisions, and it has not attributed financial results to AI.

Inditex runs a demand-driven supply chain by collecting granular data on what sells, where and in which size, and by producing and shipping in small, frequent batches. Its documented foundation is item-level RFID, full stock integration and production largely close to Spain; AI appears more recently in customer features such as Zara Try-on. Inditex has not disclosed how machine learning is used in its replenishment or production decisions.

What has Inditex built?

The core asset is data, not a single AI system. The public record shows four layers:

• Item-level RFID: by March 2016 RFID was operational in 1,542 stores across 64 markets, with the Zara rollout due to be completed by the end of 2016. The chip sits inside the reusable security tag, which can be used more than 100 times.
• Integrated stock: in 2019 then chairman Pablo Isla credited full stock integration, together with RFID, for letting the company run with even lower inventory.
• Invisible alarms and automation: at the July 2026 AGM, Inditex said an invisible alarm system was in place in over 90% of its products, enabling self-checkout and automated processes that connect fitting rooms with warehouses.
• Generative AI for customers: Zara Try-on, launched in December 2025, lets customers create an avatar from photos and see it wearing products.

At the 2026 AGM, Inditex also committed to new solutions driven by artificial intelligence as part of its innovation pillar, without detailing them.

Why did Inditex invest in data and AI?

Inditex's model depends on keeping inventory low and selling at full price. An analysis by Walter Scott, published by Macquarie in November 2024, describes most manufacturing as taking place near head office (Spain, Portugal, Morocco and Turkey), with twice-weekly deliveries to stores meaning more is sold at full price. That only works if the company knows precisely what each store holds and sells. Former chairman Pablo Isla summed this up in 2019: the business model had always been based on very low inventory, and RFID with full stock integration let the company run with even less.

Loss prevention was a second motive. In 2016 FashionNetwork reported that shrinkage cost Inditex almost 170 million euros a year, about 0.8% of sales, and the RFID chip was built into the security tag precisely to address both stock accuracy and theft.

Read also

How Uniqlo and Fast Retailing use AI and data in the Ariake Project

How does it work (data, models, process)?

The following describes what is documented and, where marked, our analysis. Each garment carries an RFID identifier that is read in distribution centres and stores. This gives real-time visibility of stock by item and location, which, combined with full stock integration, supports replenishment and inventory counts. Combined with frequent deliveries and short-run production near Spain, the data allows the company to react to sell-through during a season rather than committing everything in advance.

Analysis: this is the data foundation any demand-sensing or allocation model needs. Accurate item-level stock removes one of the main sources of error in forecasting, because a model cannot learn true demand from sales if it does not know when items were out of stock. Inditex has not said which forecasting or optimisation methods sit on top of that data.

What results has Inditex reported?

The most specific figures relate to the data infrastructure rather than to AI. In December 2019 Fortune reported that Inditex's stock-in-trade fell 5% in the nine months to October while sales rose 7.5%, and that a new Bilbao flagship sold more than four previous boutiques combined with 20% less inventory. For generative AI, Inditex said in March 2026 that Zara Try-on was live in 43 markets and had recorded more than 7 million sessions; it has not reported effects on conversion or returns.

Inditex also discloses investment levels: according to its 2026 AGM coverage, ordinary investment was 1.8 billion euros a year in 2024 and 2025, and 900 million euros a year went to strengthening logistics in the same period, with 2.3 billion euros committed for 2026. No breakdown of AI spending has been published.

What are the limits and open questions?

Editorial analysis: Inditex is often cited as an AI success story, but its public record supports a narrower claim: it is a data and operations success story. The open questions are:

• how far machine learning, rather than experienced commercial teams using RFID data, drives buying and replenishment;
• whether generative features such as Try-on reduce returns or simply add engagement;
• how much of the advantage depends on proximity sourcing, which competitors with longer supply chains cannot copy through software.

Read also

How Levi Strauss & Co. uses AI: the data programme and its lessons

What can other fashion companies learn?

1. Fix stock accuracy before buying forecasting software. Item-level visibility is the precondition for any reliable demand model.
2. Design data projects with two payoffs. Inditex's RFID tag served stock accuracy and loss prevention at once.
3. Shorten the loop between sell-through and supply; data creates value only if production and logistics can respond.
4. Integrate stock data across locations so that demand signals and replenishment draw on one view of inventory.
5. Be precise about claims: separate what data infrastructure delivers from what AI models add.

Frequently asked questions

Does Zara use AI to predict demand?

Inditex has not publicly described the forecasting models behind Zara's buying and replenishment. What has been documented is the data foundation: item-level RFID, full stock integration and twice-weekly store deliveries.

How does Inditex use RFID?

Each garment carries an RFID chip, originally inside the reusable security tag, that is read in distribution centres and stores. This gives real-time visibility of where each item is and also helps reduce theft, which cost Inditex almost 170 million euros a year in 2016.

What is Zara Try-on?

Zara Try-on is a generative AI feature launched in December 2025 that lets customers create an avatar from photos and generate images of it wearing Zara products. By March 2026 it was live in 43 markets with more than 7 million sessions.

Why is Inditex's supply chain called demand-driven?

Inditex produces a large share of its goods close to its Spanish head office and delivers to stores twice a week, so it can adjust supply to what is selling during the season instead of committing all stock months in advance.

GuideThe complete guide to AI in the fashion supply chain and sustainabilityRead the complete guide