Fashion tech: digital product passport, RFID, AR try-on, 3D design
The document helps decide which technologies to adopt and in what order: digital product passport, RFID, AR try-on, 3D design. For each it counts what it delivers (stock accuracy, fewer returns, fewer samples, compliance), what it costs and when it pays back, and how a pilot on a few sites differs from a network-wide rollout.
Why this document
The technology and operations directors read this when a vendor proposal is on the table and the question is whether to pay. It matters because technology gives nothing by itself: RFID pays back where stock accuracy is low and costs sales, AR try-on where size-related returns are high. So the effect is first measured in money on a pilot and only then scaled. The digital passport should be planned separately, as a regulatory and market requirement, not as an image project.
At a glance
- What is this document and why is it needed?
- The document helps decide which technologies to adopt and in what order: digital product passport, RFID, AR try-on, 3D design. For each it counts what it delivers (stock accuracy, fewer returns, fewer samples, compliance), what it costs and when it pays back, and how a pilot on a few sites differs from a network-wide rollout.
- Who owns the document and how often is it rebuilt?
- The decision is prepared by the digital or operations director with the finance director; projects are framed as investments with a pilot, success criteria and a decision date. The owner sees payback and pilot risks.
- What mistakes are the most common?
- Adopting technology for its own sake -> costs without effect -> start from a problem in money. Assuming sales uplift without a pilot -> overstated payback -> a pilot with a control group. Dirty master data -> tags and passport do not match the product -> clean data first.
For whom
- Brand
- A brand benefits more from technology that works for the product: 3D design (fewer physical samples and faster approval), a digital passport and composition and origin data, RFID at production and warehouse for accurate shipping. The effect appears in cost and collection time to market.
- Retail
- A retailer benefits more from technology on stock and the customer: RFID in stores (accurate stock, quick counts, fewer losses), AR try-on and smart mirrors (fewer returns, more conversion), data for omnichannel. The effect appears in availability and returns cost.
What it consists of
1. Digital product passport (DPP)
- Contents: product identifier, material composition, origin, suppliers, care and repair rules, recycling, carbon footprint; access by link or code on the item.
- What it gives: compliance for export markets, customer trust, a basis for resale and repair; what it needs: clean master data and supplier data.
2. RFID
- A tag on each item, readers at warehouse, entrance and in store, handheld scanners for counts; integration with the accounting system and WMS.
- What it gives: stock accuracy (up to 95–99% when set up correctly; a notional figure to be validated by a pilot), counts many times faster, item visibility for web orders, fewer losses.
3. AR try-on and smart mirrors
- Retailer: Virtual try-on on the web and in the app, mirrors with size and complementary item suggestions in store.
- What it gives: higher conversion and fewer fit returns; the effect is measured with a control group, not by comparing periods.
4. 3D design
- Brand: Digital samples in 3D garment software (CLO, Browzwear and similar): fit and drape on a virtual avatar, color and material approval without a physical sample.
- What it gives: fewer physical samples and iterations, faster time to market, sales content before production; condition: a library of materials and avatars based on real measurements.
5. Payback calculation and pilot
- For each technology: annual effect (profit from sales, savings on returns, samples, labor), one-off cost, annual running cost, payback in months.
- Pilot: 2–4 stores or part of web traffic with a control group, 8–12 weeks, success criteria set in advance.
6. Brand and retailer: where each technology works
- Brand: Brand DPP: the owner of the passport data (composition, suppliers, repair, recycling); data are collected along the chain from the material, so the source of truth should be in PLM, not a spreadsheet; incomplete supplier data are closed by contract.
- Retailer: Retailer DPP: receives data from brands and shows them to the customer via QR; for a private label it collects data like a brand; control: the share of cards where the brand lacks mandatory fields.
- Brand: Brand RFID: the tag is applied at sewing or packing, the shipment is checked against the ASN without recounting, shipment accuracy rises; the gain is fewer wholesale claims and faster receiving at their end.
- Retailer: Retailer RFID: a store count in hours (an example assumption of 160 against 64 hours), “find on the floor”, theft protection, one stock view for online; tags are read at receiving by portals.
- Brand: 3D and digital samples: fewer physical samples (an assumption of 2 samples per style at RUB 9,000), a digital showroom for wholesale clients before manufacturing; the condition is shared material and fit standards in PLM.
- Retailer: AR try-on and smart mirrors: conversion growth and lower online returns; for a multi-brand retailer a catalog of brand 3D models is needed, so reach is limited to the private label and partners.
- Brand: Brand pilot: one factory or one DC, 8-12 weeks; criteria: shipment accuracy, receiving time at the client, share of samples saved.
- Retailer: Retailer pilot: 2-4 stores and the same number of control stores, 8-12 weeks; criteria: stock accuracy, count time, size sales, online returns.
7. Brand and retailer: what to count in payback
- Brand: Brand effect in money: saved samples, fewer wholesale client claims, a collection launch sped up by weeks; the added tag cost and supplier data work go into costs.
- Retailer: Retailer effect in money: sales growth from size availability, lower losses, count labor savings, fewer online returns; costs: tags, readers, integration, store training.
How to read the result and which decisions follow
- The bridge shows a net RFID effect of RUB 13.9 m per year before running cost of RUB 1.2 m on an investment of RUB 24.0 m; payback 22.7 months. The best category by effect to revenue is “Accessories”. Decision: introduce RFID in stages, starting with categories with a high effect per rouble of revenue.
- Brand: a digital passport and a tag on the item are needed for authenticity, repair and recycling; passport data (composition, suppliers, care) is gathered in PLM in advance. Decision: first put composition and batches in order, then launch the tag; legal dates for textiles are not asserted without checking (unverified).
- Retailer: RFID pays through the speed of counting and the accuracy of stock in stores; AR try-on and 3D give an effect at the other end of the chain. Decision: compare technologies by payback and by impact on returns and availability, not by novelty.
- Brand: 3D design cuts the number of physical samples; the example effect is 120 styles x 2 samples x RUB 9,000. Decision: count the effect by dropped samples and cycle speed, not by the number of virtual items.
- Retailer: the tag as a percentage of price for kids and accessories is higher than for women; for cheap items the tagging decision is taken per batch, not per item.
Control checks
- Check: the net effect of RUB 13.9 m equals the sum of the bridge columns by category (the code reconciles with the table).
- Brand: the tag failure rate and read accuracy (95-99% in the example) are conditional; measure them in a pilot and do not carry them from promotional data.
- Retailer: sales growth from availability (0.8-1.0% in the example) is an assumption; check it with a control group of stores.
- Brand: product passport data contains commercial information about suppliers; decide what to publish to the client and what to keep inside.
- Retailer: RFID tagging and movement tracking do not replace mandatory labeling where the law requires it; both systems work in parallel (unverified).
Where the data comes from
| Data source | Fields |
|---|---|
| Stock and count data from accounting | Count variances, losses, labor spent on counts |
| Web shop and returns logistics | Orders, return rate and reasons, return handling cost |
| Product and samples | Number of styles, samples per style, sample cost, approval times |
| Supplier and certificate data | Composition, origin, certificates, data for the product passport |
How to build it step by step
- Step 1. State the problem in money (stock variance, size returns, sample count), not “adopt technology”.
- Step 2. Size the problem from own annual data: lost sales, return costs, sample spend.
- Step 3. For each technology define the effect mechanism and the share of the problem it removes; the share is an assumption until the pilot.
- Step 4. Compute costs: one-off (equipment, integration) and recurring (tags, licenses, support, staff).
- Step 5. Build the payback table and sensitivity to the key assumption (for example sales uplift).
- Step 6. Plan a pilot with a control group and metrics; define in advance the values at which rollout expands or stops.
- Step 7. Prepare data: clean master data, barcodes and codes, composition, supplier data.
- Step 8. Run the pilot, compare with control and recompute payback on actual values.
- Step 9. Decide whether to scale; plan rollout in waves, staff training, support.
- Step 10. Reconcile actual and plan for effect and cost twice a year.
- Brand: Step 11. Brand: start with PLM and data quality (composition, suppliers), then the shipment tag; do not launch a DPP without a data owner for each material.
- Retailer: Step 12. Retailer: start with stores that have the largest stock mismatches and online fit returns; compute the effect by category, not on average.
Formulas
| Measure | Formulas | Note |
|---|---|---|
| Payback, months | Payback = one-off cost / (annual effect - annual running cost) x 12 | If the difference is not positive, the technology does not pay back. |
| Retailer: Effect of stock accuracy | Profit = revenue x sales uplift from availability x gross margin | The uplift is tested in a pilot; overestimating it is the main calculation error. |
| Effect of return reduction | Saving = orders x (return rate before - after) x cost per return | Add preserved margin if a return cannot be resold. |
| Retailer: Tag cost | Tag cost = items per year x tag price | The tag is part of item cost: for cheap items its share is visible. |
Formulas in notation
- Payback
Payback = Capex / (Effect − Run) × 12
Effect is the annual effect, Run running cost per year; in months; if Effect ≤ Run it does not pay back.- Profit from sales growth
G = Rev × u × GM
u is the sales uplift from stock availability, GM gross margin.- Shrinkage saving
L = Rev × (s_0 − s_1) × (1 − GM)
s_0 and s_1 are shrinkage shares before and after; losses are valued at cost.- Tags per year
Tags = Rev / P × t
P is the average item price, t the tag price.- Tag as % of price
τ = t / P × 100
Shows for which categories RFID is expensive.
Worked example
Exercise: RFID effect in 40 stores by category, RUB m a year
| Category | Revenue | Items, thousand | Profit from sales uplift | Lower losses | Count labor | Tags per year | Net effect before fixed costs | Tag, % of price |
|---|---|---|---|---|---|---|---|---|
| Women | 1,280 | 284 | 7.2 | 1.7 | 0.69 | 2.28 | 7.3 | 0.18% |
| Men | 640 | 128 | 2.1 | 0.6 | 0.35 | 1.02 | 2.0 | 0.16% |
| Kids | 400 | 160 | 1.7 | 0.6 | 0.22 | 1.28 | 1.2 | 0.32% |
| Footwear | 560 | 86 | 1.4 | 0.6 | 0.30 | 0.69 | 1.6 | 0.12% |
| Accessories | 320 | 107 | 1.6 | 1.0 | 0.17 | 0.85 | 1.9 | 0.27% |
| Total | 3,200 | 765 | 13.9 | 4.4 | 1.73 | 6.12 | 13.9 |
Notional: tag RUB 8, count labor savings spread by revenue (RUB 1.73m in total). Sales uplift differs: from 0.5% in footwear to 1.0% in women; losses before the rollout range from 1.0% to 2.0% of revenue in accessories. Total net effect is RUB 13.9m a year before fixed running costs of RUB 1.2m; with one-off costs of RUB 24m payback is about 23 months. The best effect per rouble of revenue is in Accessories: the tag is cheap against item price while losses are higher. AR and 3D were not split by category. All inputs are exercise assumptions, not a market benchmark; the sales uplift is tested in a pilot.
Figures
- positive
- negative
- total
Data as a table
| Step | Change | Level |
|---|---|---|
| Sales growth | +13.9 | 13.9 |
| Lower shrinkage | +4.4 | 18.3 |
| Counting labor | +1.73 | 20 |
| Tags per year | −6.12 | 13.9 |
| Net effect | — | 13.9 |
RFID effect bridge per year. Profit from sales growth, lower shrinkage and counting labor minus the cost of tags give the net effect before fixed costs.
- Effect minus running cost
- One-off cost
Data as a table
| Measure | Effect minus running cost | One-off cost |
|---|---|---|
| RFID in 40 stores | 12.5 | 24 |
| AR try-on in the web shop | 1.3 | 4 |
| 3D design | −0.8 | 1.5 |
Annual effect net of running cost by technology. What remains per year after running costs: RFID and 3D give a plus, AR try-on does not at these assumptions.
Data as a table
| Measure | Tag, % of price |
|---|---|
| Women | 0.18 |
| Men | 0.16 |
| Kids | 0.32 |
| Footwear | 0.12 |
| Accessories | 0.27 |
Tag as a percentage of price by category. A tag at RUB 8 is more expensive for lower-priced categories: kids and accessories carry a larger share of the price.
Digital product passport: where data comes from and where it goes. Data is gathered in PLM and from suppliers, published by a code on the item and used by the client, service and recycling.
- Suppliers: material, origin → PLM and ERP: composition, batch
- PLM and ERP: composition, batch → Passport via a QR code or tag
- Passport via a QR code or tag → Client: care, authenticity; Repair and resale; Recycling: composition and disassembly
- Client: care, authenticity
- Repair and resale
- Recycling: composition and disassembly
Specifics by category
- Apparel
- RFID: the more SKUs and counts, the bigger the effect; a tag in a label or seam costs RUB 8 against an item price of RUB 4,500, i.e. 0.18% (an assumption). 3D design of fit and drape cuts 1-2 samples per style. AR try-on is more useful for outerwear and dresses where fit is critical. DPP: composition and care are already in the tech pack, so a passport pilot starts with one product group, not the whole range.
- Footwear
- The tag is usually put on the box or insole: on the box data are stable but the mix-up risk is higher (a pair not in its box); inside the pair metal and moisture interfere. AR sizing from a foot measurement (length in mm, Mondopoint) cuts fit returns, the costliest in footwear; EU/UK/US are derived from length. 3D is limited by the last and materials: a digital last is a separate asset. A tag at RUB 8 against a pair price of RUB 6,500 is cheap (0.12%), so the limit is technology, not price.
- Accessories
- RFID pays on small expensive items: loss reduction and stock accuracy matter more than sales uplift; the example assumption is losses falling from 2.0% to 1.2% of revenue against 1.2% to 0.9% in apparel. A tag on earrings and thin belts is hard to place and metal shields the signal, so hang-tags are used. AR try-on of glasses, jewelry and hats is strong. Bag authenticity: a serial number and repair history in the passport; precious metals need the GIIS DMDK circuit and RFID does not replace it.
- Kids
- RFID helps with seasonal peaks and high turnover; a tag in a child’s item must be safe (no skin irritation, no small detachable parts), a decision for the technologist (tag requirements under TR CU 007/2011 were not checked, unverified). AR is limited: size depends on height, a 6 cm step, and the child grows. A child’s data is not collected without guardian consent. In a DPP for infants composition and care matter: for children under 1 year products are made of natural materials, except fillings.
- Men
- RFID pays through basics stock accuracy and auto-replenishment; sales uplift is more modest than in women (an assumption of 0.6% against 1.0%). AR has less effect, but size matching for shirts and suits from measurements cuts returns on sleeve and girth. 3D is useful for basics: one pattern set runs for years, so sample savings pay back more slowly than with fast-changing fashion. DPP is simpler with repeating SKUs: data are reused across seasons without a new collection effort.
- Women
- A higher share of fashion, fit returns and online: AR try-on and size matching have a bigger effect; a control group is mandatory, otherwise a conversion rise is easily blamed on the season. 3D cuts samples with frequent collection changes. RFID gives a larger uplift through size availability and “find it on the floor” (an assumption of 1.0%). Vanity sizing makes measurement-based matching stronger than number-based. In a DPP each style has more materials and suppliers, hence more data work.
Common mistakes
- Adopting technology for its own sake -> costs without effect -> start from a problem in money.
- Assuming sales uplift without a pilot -> overstated payback -> a pilot with a control group.
- Dirty master data -> tags and passport do not match the product -> clean data first.
- Rolling out to the whole network at once -> errors replicate -> waves and stop criteria.
- Confusing an RFID tag with the marking code -> breach of mandatory marking -> apply the marking code by the rules; RFID complements it.
Russian specifics
- Light industry goods are subject to mandatory code marking (marking rules, an expanding list of product groups): the marking code is required regardless of an RFID tag and is not replaced by it. Rule: Decree 1956 of 31.12.2019: Government Decree of 31.12.2019 No. 1956 (marking rules for light-industry goods).
- New marking waves change the list of goods and deadlines; check current deadlines against the norm registry at the decision date. Rule: Decree 2129 of 25.12.2025: Government Decree of 25.12.2025 No. 2129 (amendments to Decree No. 1956; fourth marking wave).
- Brand: EU digital product passport requirements (the ecodesign regulation) matter for exports to the EU; application dates for textiles and the data format need separate checking. Rule: EU Regulation 2024/1781 of 13.06.2024: Regulation (EU) 2024/1781 (ESPR, ecodesign for sustainable products) (rule check not completed).
- Brand: the marking code is applied to the item at release or import; an RFID tag does not replace the code, and both must be tied to one SKU in the ledger (format and dates unverified).
- Retailer: at sale and return the marking code is read at the till while RFID portals serve counting; the withdrawal-from-circulation procedure and stock handling are checked against current rules (unverified).