The Impact Thesis: A four-stakeholder, framework-aligned account of in-store collection at scale

1. Introduction: the trajectory problem

The European textile collection system has been improving. Slowly.

In 2016, the EU separately collected 10.7% of its textile waste. By 2022, the figure had risen to 15.0% [1]. The improvement rate over that six-year window is 0.72 percentage points per year. The trajectory is positive. It is also, when projected forward at the same pace, inadequate to the policy ambition.

At 0.72 percentage points per year, the line crosses 50% in approximately 2071, and 80% in approximately 2113. The EU's stated 2030 ambition is 50%, with an aspirational upside of 80% [2]. The gap between the trajectory and the ambition is forty-one years for the central target and eighty-three years for the upside. Some part of this gap will close through the EPR-funded scale-up triggered by Directive (EU) 2025/1892, which enters force on 16 October 2025 and requires operational EPR schemes by 17 April 2028 [3]. Some part will not.

European textile waste runs to 6.94 million tonnes per year, or sixteen kilogrammes per person across the EU27 [1]. Around 85% of that volume, approximately 5.9 million tonnes, leaves the household via mixed municipal waste streams and is incinerated or landfilled. The remaining 15% enters the collection system. Of that 15%, communal collection programmes recover approximately 36% in sort yield [4]; the rest is lost again to sorting and processing inefficiencies. The aggregate fate of post-consumer textile waste in Europe today is approximately 94% incinerated or landfilled.

A consensus account of the cause locates the bottleneck downstream: insufficient sorting capacity, immature textile-to-textile recycling technology, an absence of profitable end markets for recovered fibre. Each of those factors is real. None of them is the binding constraint.

I spent the first eighteen months of Utilitarian assuming the binding constraint was downstream of the consumer return. We were building tooling for sorters and recyclers. The data we cared about was the material composition of what came out of the sort. Then I spent a year talking to recyclers. The constraint they kept describing was not the technology. It was the feedstock walking through their door uncharacterised. The technology existed. The capital to deploy it existed. What did not exist was the information layer in front of it.

The same diagnosis turns out to apply at every other point in the value chain. The retailer's compliance team cannot report at the product resolution that CSRD will require, because the take-back programme produces a kilogram figure at the loading dock rather than a per-product record at the counter. The brand cannot attribute end-of-life outcomes to its SKUs through the wholesale channel, because the wholesale channel is the half of its distribution where it has no customer-data relationship. The regulator cannot modulate EPR fees against the products that are returned, because no record exists at the point of return that resolves to the SKU. The customer cannot see where the product they brought back went next, because the system between the counter and the recycler does not produce that visibility.

The constraint is not collection infrastructure. It is the absence of an information layer at the moment of consumer return.

1.1 The regulator's choice, and the industry's reading

Directive (EU) 2025/1892, in force from 16 October 2025, makes textile and footwear EPR mandatory across all 27 Member States, with operational schemes required by 17 April 2028 [3]. The defined obligated party is the "producer", but the directive's definition of producer is constructed so that the obligation lands on the entity that places goods on a Member State market under its own name, brand or trademark, "irrespective of the selling technique used". In retailing language, that is the brand owner, the private-label retailer, and the distance seller; the definition also reaches online marketplaces and non-EU sellers placing product on the EU consumer market [5].

The choice of obligated party was not arithmetically the cheapest. The Commission's Impact Assessment, SWD(2023) 421, records that current Member State arrangements assign separate-collection responsibility variously to municipalities, commercial operators, or social-economy enterprises, and ranked the EPR-with-producer-financing option above municipal-led collection and above status-quo extension on the explicit logic that it internalises the cost of end-of-life in the entity placing product on the market [6]. The directive's eco-modulation requirement, which scales fees with durability, reparability, and recyclability of the product placed on the market, only operates if the obligated party is also the design-controlling party. Municipalities and charity collectors do not design product. Retailers and brand-owning retailers do, or commission it. The choice of obligated party is therefore consistent with a deliberate decision to engage the actor whose fee schedule reaches back into product specification, not the actor with the lowest collection-cost-per-tonne.

The industry's reading of that choice has been narrower. The joint FEAD, EuRIC Textiles and Decathlon position of November 2024 frames retailers and producers as actors to be coordinated to deliver flow into the recycling network rather than as co-designers of the operating model [7]. EURATEX, representing brand-owning retailers, has argued for harmonised EPR fees and treats the retailer as the obligated payer rather than as the operator [8]. Two organisations have read the choice differently: the Ellen MacArthur Foundation's 2024 work on EPR policy explicitly names brands, retailers and online marketplaces as the responsibility-bearing tier and argues that EPR must reach into design and business model [9]; and EuroCommerce, representing the retail and wholesale sector, treats its members as the design point of the system [10]. Outside those two, the recycling industry frames retailers as the financing line, and the rest of the sustainability industry frames them as the compliance perimeter.

The WEEE precedent is informative rather than predictive. WEEE Directive 2012/19/EU has obliged retailers above 400 m² to accept very small WEEE free of charge for thirteen years; the EU collection rate stood at 40.6% in 2022 against a 65% target [1]. The WEEE Forum's 2024 review concluded that the current methodology is "neither accurate nor achievable" and called for an "all actors" principle holding everybody in the value chain accountable. Whether textile EPR avoids the WEEE failure mode depends on whether the industry treats retailers as operators or as a financing line. That is the choice the current responses make visible.

The directive sets the floor. The industry's reading does not yet build the ceiling. The operational layer between obligated party and recycler, the layer that produces the consumer-facing return flow, runs the data infrastructure that connects placement to recovery, and is accountable for the eco-modulation signal reaching back into product specification, is the layer this paper is about.

This paper sets out what installing that layer at scale would produce, for each of the four stakeholders whose data it serves, against each of the frameworks the regulators and the impact investors have converged on, and at the European deployment scale the 2028 EPR mandate is about to make compulsory.

2. Structural placement: where the platform sits in the system

The moment of consumer return is the structural locus of the impact thesis. It is the only physical location in the European textile value chain at which the data required by the recycler, the regulator, the brand, the retailer, and the consumer is naturally co-located with the consumer whose action makes that data accessible.

Two consequences follow. The first is that the data captured at this moment has unusual reach: every downstream actor in the chain has a use for some piece of it, and the use is materially different at each node. The second is that any platform that does not capture data at this moment captures less of it later, at higher cost, with worse fidelity, and against a smaller fraction of the population. The bin in the household is the structural enemy. By the time the consumer has decided to discard a product, the only system that retains a relationship with both the customer and the product is the retail counter.

Under the EU Categorisation System for the Circular Economy, the platform sits in two adjacent positions.

Category 3. Re-use and value recovery. The verbatim named sub-category is Traceability platforms. The category describes systems that produce structured records of value-recovery events such that downstream operators can identify, sort, and process material on the basis of those records. The platform's primary commercial product is a traceability platform of this kind.

Category 4. Enabling Technologies. Category 4 describes the technical layer that makes circular activity in the other three categories operationally feasible at scale. Take-back-counter data capture, AI product identification at the point of return, and the integration layer that delivers that data into a retailer's existing CRM and into the recycler's existing sort line are enabling technologies in the verbatim sense Category 4 names.

The dual placement is not coincidental. A traceability platform that does not run on enabling-technology rails does not deploy across 100,000 retail stores. An enabling-technology layer without a traceability product does not produce the structured records the downstream frameworks consume. The platform is built to occupy both positions because the impact thesis requires both.

2.1 If the data is valuable, what does the hardware cost?

If the per-product, per-customer record at the moment of return is valuable to retailers, brands, recyclers, and regulators, the next question to ask is what the hardware to capture it actually costs.

A previous account of this cost anchored on industrial sortation equipment: fixed Tomra-class near-infrared lines at €250,000 to €450,000, fixed UHF RFID portals at €15,000 to €25,000, sensored deposit bins. That stack is real, and it is what a textile reprocessor or a regional collection consolidator would specify. It is not what an apparel, footwear, or sportswear retailer would install behind a shop counter to record which customer returned which garment. The realistic in-store envelope is roughly an order of magnitude lower.

Item-level UHF RFID is now the dominant inventory technology in apparel; per-store handheld readers, not fixed portals, are the realistic in-store envelope. Street prices through Zebra, TSL/HID Global, and Honeywell sit between €1,500 and €3,000 per unit at single-unit retail [11]; [12]. A retailer specifying one handheld per store would book between €1,500 and €3,000 of reader capex per location.

NFC is the more interesting category. NFC is now native to essentially every smartphone shipped in the last decade. Apple has supported NFC tag reading since the iPhone 7 in 2016, opened background tag reading in iOS 14 in 2020, and was obliged in 2024 to expose Host Card Emulation under European Commission antitrust action [13]; [14]. Android shipped NFC in volume from 2011 onward via Host Card Emulation on KitKat. By Q4 2023, around 94 per cent of smartphones in circulation globally carried an NFC controller [15]. The marginal hardware cost of a smartphone-based NFC read is zero when the device is already in hand.

Tags applied at manufacture run €0.04 to €0.10 per garment on volume, and the cost is borne upstream by the brand or producer, not by the retailer's hardware capex line [16]. Optional handheld near-infrared fibre verification adds €5,000 to €15,000 per unit including the first-year licence, where in-store composition verification is required [17].

Year-one capex for a basic in-store identification stack therefore lands at approximately €3,000 to €5,000 per store. With handheld near-infrared added for fibre verification, the figure rises to approximately €13,000 to €19,000 per store.

The asymmetry is what the question turns on. Per-store hardware capex of €3,000 to €5,000 sits against a per-facility MRF capex of €15 million to €40 million (developed in §3.4 and §4.4 below): two to three orders of magnitude. If the value of the data is established at the retail counter and the components of the capture stack are commodity, off-the-shelf, and largely already present in the retail estate, the question worth asking is why the European retail estate has not deployed per-product, per-customer end-of-life data capture at scale. The answer is not the hardware. §12 returns to it.

3. The unit economics: from material value to unit-of-value-per-unit-of-material

Every commercial interface in the European post-consumer textile chain is denominated in per kilogramme of material. Sorters bid per tonne. Recyclers contract per tonne. Producer Responsibility Organisations subsidise per tonne. Mechanical and chemical recyclers sign offtake agreements indexed to per-tonne feedstock specifications, with bonus and malus clauses against composition and contamination thresholds [4]. The chain's accounting frame, end-to-end, is mass.

This section sets out what follows when the only unit of account is mass, and what changes when a second unit of account becomes available at the moment of consumer return.

3.1 The chain is denominated in kilogrammes

The 2022 Sorting for Circularity Europe baseline measured the gap between the cost of producing recycling-grade feedstock and the price at which a mechanical or chemical recycler would buy it at roughly €150 to €200 per tonne. The 2024 update widened the gap as the rewearable export market to North Africa and Sub-Saharan Africa collapsed under second-hand saturation and import frictions [18]. Refashion's 2023 disclosures show the French PRO paying out roughly €100 per tonne in operator support, with sorters reporting operating margins near zero once subsidy is netted out [19].

Three structural observations follow. Every line item in the chain's profit and loss is indexed to mass. The commodity reference, per-tonne fibre price, is the only reference the chain uses to value its own output. And when material prices fall, the entire chain becomes loss-making at the operator level, because no other revenue line exists to absorb the shock.

3.2 The customer-value line is priced in retail and absent in recycling

In retail marketing, the cost of acquiring a permissioned customer record is a well-documented line. The DMA UK Marketer Email Tracker 2023 reports a median return of £36 per £1 spent on email, with acquisition cost per consented email running between £1 and £5 depending on channel [20]. Klaviyo's 2024 benchmarks place median paid-social cost per email between US$2.50 and US$6.00 for apparel brands; in-store capture at a point-of-sale terminal lands between €1 and €3 per consented email, the lower figure reflecting the fact that the customer is already present, transacting, and identifiable [21]. Litmus puts average email programme return on investment at roughly 36 to 1 for retail, with the consented address itself being the scarce asset [22].

Gartner's 2023 CMO Spend Survey identifies first-party data acquisition as the single highest-priority capability investment for retail CMOs in response to third-party cookie deprecation [23]. Forrester's 2024 State of Retail Marketing names "zero-party data capture in physical stores" as a top-three investment area. Forrester's list of capture channels does not include take-back, return, or recycling [24]. The eMarketer 2024 retail data-spend forecast counts in-store kiosks, loyalty programmes, app installs, receipt-based opt-in, and QR-coded packaging, but does not list take-back among acquisition channels [25].

The retail marketing industry prices a permissioned customer email at €1 to €5, treats first-party data as its scarcest input, and has not categorised the take-back moment as one of the locations where that input becomes available.

3.3 The recycler P&L: customer engagement is not a revenue line

The symmetric absence is on the recycler side. Refashion's 2023 annual report breaks down revenue into eco-modulated producer contributions, sorting support payments, R&D fund disbursements, and operator subsidies. Consumer-engagement revenue does not appear. I:CO and SOEX, the largest operators of branded retail take-back in Europe, report throughput in tonnes and grading fractions; their public material does not disclose customer-data revenue lines [26]. The FEAD 2023 textiles report and the JRC's 2021 circular-economy perspectives on textiles both organise the chain around per-tonne flows and per-tonne costs. Neither lists customer attention, consent, or first-party data among the value lines [27]; [28].

This is the unowned line. The recycling industry has not categorised the customer at the moment of return as anything other than a delivery mechanism for material. The customer-value line therefore exists nowhere in the chain's accounting. It is not lost. It is unrecognised. That distinction matters: a value line that has been recognised and lost can be recovered through better commercial design; a value line that has not been categorised by either of the two industries that would naturally own it remains structurally unowned until a third actor builds the layer that captures it.

3.4 MRF capex as the cost of upstream information absence

The European materials recovery facility for textiles has become a capital-intensive plant. The 2022 ReHubs baseline costed an automated sorting line at €15 million to €30 million, integrating near-infrared spectroscopy, hyperspectral imaging, robotic picking, and AI-based composition recognition, with throughput in the range of 25,000 to 50,000 tonnes per year. Tomra's published material on Valvan-Tomra and Project Refiber places unit cost per tonne on the automated line at roughly €120 to €180, against a manual sort cost of €250 to €400 per tonne [29].

The capex is, structurally, the price paid downstream for not knowing upstream. Every near-infrared scan at the MRF is a re-identification of a garment whose composition was declared by the producer at the point of placing-on-market, then lost during use and disposal. The JRC's 2024 work on the Digital Product Passport names this directly: if composition travels with the garment, the sort line collapses from a re-identification machine into a routing machine [30]. Refashion's 2024 sorting-facility cost analysis estimates that 15 to 25 per cent of MRF operating cost is directly attributable to compositional uncertainty at intake.

§4.4 returns to this point in the context of the recycler's specific value position; the broader observation here is that the chain has been paying, in capex, for the information it did not capture at the point of consumer return.

3.5 The structural shift

Lay the three figures next to each other. A tonne of mixed post-consumer textile sells, after sorting, for between €0 and €400 depending on fraction, with the median European sorter losing money on the marginal tonne. A single permissioned customer email is worth €1 to €5 in acquisition-cost terms, and considerably more in downstream lifetime value: Klaviyo's median apparel customer-lifetime-value per consented contact runs into the hundreds of euros. A garment weighs roughly 0.3 to 0.5 kilogrammes.

A typical store-level take-back event therefore delivers, in customer-value terms, an asset of one to two orders of magnitude higher value than the material it accompanies, at the moment the customer hands the garment over.

The data layer captures this customer value at marginal cost. Identification at the moment of return generates a consented record, an attention event (the visit to the store), and a behavioural signal (brand, category, frequency, volume). None of these are currently priced into any contract in the recycling chain. All of them are routinely priced in retail marketing. The arbitrage is not a marketing trick. It is the mechanical consequence of two industries having drawn their commercial boundaries such that the customer-value line at the take-back moment falls between them.

3.6 What this changes

The chain has been organised, from collection through sorting to recycler offtake, around per-kilogramme material value. Every contract, every gate fee, every PRO subsidy line, every MRF capital case is denominated in mass. The customer-value line generated at the moment of consumer return is structurally absent from recycler accounting and structurally uncategorised in retail marketing surveys. It is the largest unowned line in the chain.

Capturing it converts the unit economics from per-kilogramme-of-material to unit-of-value-per-unit-of-material, where the unit-of-value is a permissioned, identified, behaviourally signalled customer interaction whose published benchmark price already exceeds the per-kilogramme bale price by one to two orders of magnitude. This is the structural shift the data layer enables. It is the reason the chain's economics change shape, rather than merely improving, when identification moves to the moment of return.

§4 follows the consequence of this shift through the four stakeholders, each of whose unit economics change when the customer-value line is captured rather than left unowned. §6 examines the eight links of the theory of change that connect the moment of capture to the net impact outcome at scale. §12 returns, in the closing structural synthesis, to the question this section leaves open: if the customer-value line is this large and the cost of capturing it is this small, why has the European retail estate not built the layer that captures it?

4. The four-stakeholder value chain

The four-stakeholder framing is the structural spine of the impact thesis. Each node in the chain (customer, retailer, brand, recycler) captures both a sustainability impact and a commercial value from the same data event, and resolves a specific failure mode that the existing batch-collection model produces by design. The chain is presented in customer-first reading order; the data flow runs in the same direction.

4.1 The customer

Role: the person whose action makes the data accessible. Without the customer's participation, the chain does not begin.

Sustainability impact for the customer. Frictionless participation in circularity at the moment of replacement purchase. The cognitive cost of the existing batch-collection model (accumulate, find a charity bin, make a separate trip) is replaced by a drip-style alternative: bring the product back when you next come to the store. The new model uses a moment in the consumer's purchasing cycle that is already happening for a different reason.

Commercial value for the customer. An immediate reward (a store credit or discount, redeemable on the replacement purchase). Transparency about where the returned product goes next. A route to disposal that is simpler than every alternative the consumer has access to today.

The failure mode this resolves. The behavioural literature on prosocial participation identifies three cohorts in roughly stable ratios: a prosocial baseline of approximately 18%, a middle cohort of approximately 50% whose participation is contingent on the action being easy and visible, and an indifferent cohort of approximately 32% [1]; replicated empirically in the companion paper [2]. The existing batch-collection model serves the prosocial baseline well, the middle cohort only when the trip happens to be convenient, and the indifferent cohort not at all. The drip model converts the consumer ask from "accumulate then dispose" to "bring it back when you next come in": different cognitive load, different participation curve, different cohort recruitment [31].

4.2 The retailer

Role: the operator of the physical point of consumer return. The retailer is the actor whose budget pays for the deployment, and whose operating system the deployment has to fit.

The European apparel, footwear, and sportswear store is engineered to convert square metres into transactions at a defined gross-margin yield, against a labour budget set by sales forecast. Read as an operating system, it exposes five binding constraints: thin operating margin, a fully allocated labour envelope, footprint costed at sales-productivity rates, working capital disciplined by inventory turn, and a bounded staff-attention budget that is a separate input from labour hours. Conventional take-back, with back-of-house bins, manual sorting, weight-based reporting, and parallel reverse logistics, imposes inputs the system was not specified to absorb. The retailer's resistance to accommodating it has been read in the industry literature as a sustainability-attitude problem. The evidence describes a binding-constraint problem.

Margins. European specialist retailers operate on net margins that do not absorb new fixed cost lightly. H&M Group reported operating margins of 3.7 per cent in FY2023 and 6.2 per cent in FY2024 [32]. Inditex, the most profitable European apparel operator at scale, reported 19.2 per cent on FY2024 sales of €38.6 billion [33]. The Deloitte Global Powers of Retailing 2024 benchmarks the apparel composite at a 6.5 per cent net profit margin, materially below food and discount retail [34]. Every euro of non-revenue-producing in-store activity is priced against a margin envelope that, outside Inditex, is single-digit.

Labour. Store-level wages run 10 to 17 per cent of net sales in specialty retail [35]. Between 65 and 75 per cent of front-line associate time is committed to scheduled tasks: replenishment, customer service, transactions, fitting room recovery, opening and closing. Discretionary capacity sits below 15 per cent [36]. Take-back retrofitted into this envelope lands as unscheduled task time: customer hand-off, garment inspection, sorting, logging, storage placement.

Footprint. Prime European high-street retail rents run from €4,000 per square metre per year in secondary high streets to over €20,000 on flagship streets in Milan and Paris [37]. Back-of-house allocation is typically 15 to 22 per cent of gross internal area for fashion specialty, with cost per square metre identical to selling-floor cost since leases price gross internal area, not function [38]. Storing returned garments on this footprint against zero direct revenue is a substitution cost that conventional take-back does not capitalise.

Inventory turn. Apparel retailers run inventory turn in the 3.2 to 4.4 times range at scale [32]; [33]. Returned garments held for periodic reverse-logistics pickup function, on the balance sheet, as inventory that cannot be sold and cannot be written off until disposition. They consume the same days-of-inventory line as sellable stock without contributing to turn.

Attention. Behavioural-operations research treats attention as a non-fungible input. Adding non-selling tasks to front-line associates produces statistically significant declines in conversion rate and basket size, with the effect concentrated at peak hours when attention is most constrained [39]. Take-back as conventionally designed lands at the till and the fitting room: the two highest-attention nodes in the store.

The failure modes are documented. The H&M Garment Collecting programme was operated through I:CO, a subsidiary of SOEX Group; SOEX Textil-Vermarktungsgesellschaft mbH filed for insolvency in October 2024, triggering a halt to collection volumes from H&M and partner retailers across European markets [40]. The single-processor dependency is the proximate illustration. Changing Markets Foundation's 2023 audit of in-store take-back schemes traced bin contents from ten European retailers and found 25 of 27 sampled garments exported, downcycled, or destroyed, with end-of-life destinations including landfill in Ghana and Romania [41]. Greenpeace Germany's 2022 follow-up placed 29 GPS-tagged garments at high-street take-back points; 24 left the EU within 90 days and 16 were tracked to municipal waste in destination countries [42]. Decathlon restructured its in-store buy-back from 2020 around dedicated second-life corners and digital pre-qualification following operational difficulty integrating walk-in deposits with standard store flow; North Face Renewed transitioned from in-store back-of-house collection to mail-in and dedicated processing in 2022 [43]. Across the catalogue, the failure is operational, not motivational.

The test any take-back model must pass to deploy at retail scale follows from the constraints, not from an external standard. It must not add unscheduled associate time at the till or fitting room. It must not occupy back-of-house footprint at sales-productivity rates. It must not place non-sellable inventory on the retailer's balance sheet. It must not depend on a single downstream processor for continuity of operation. A model that fails any one of the four conditions is asking the store to reshape its operating system. A model that satisfies all four is one the store can absorb without re-specification.

Sustainability impact for the retailer. Per-product end-of-life records exportable as evidence for CSRD Scope 3 Category 12 [44]. EPR fee-modulation evidence under Directive (EU) 2025/1892. Digital Product Passport end-of-life event records under the Ecodesign for Sustainable Products Regulation, in force since 18 July 2024, with the textile delegated act expected in 2027 [45].

Commercial value for the retailer. Verified customer email captured at €1–3 per address, against benchmarks of €12–25 across paid social acquisition and €18–31 across on-site popups [46]; [47]; [48]. Approximately 27% of email-captured scans come from repeat customers, on operational data from August 2025 to March 2026 across five Dutch retail stores [2]. A transparency-premium lift in purchase probability of 6 to 46 per cent from disclosing the supply-chain step at the moment of consumer interaction [49].

A model that satisfies the four-part test, and produces a verified email plus an in-store visit plus a structured replacement-consideration window from each scan event, is no longer a compliance cost line. It is the cheapest verified email a physical retailer can buy, and the cost is already provisioned in the EPR and CSRD budget. The compliance budget is converted into a marketing acquisition budget without new spend. The order matters. The operating-system test comes first; the cost-line conversion follows from passing it.

4.3 The brand

Role: the manufacturer whose product is being returned. The brand is the actor with no direct relationship to the customer in the wholesale half of its distribution.

Sustainability impact for the brand. SKU-resolution end-of-life data for wholesale-channel product, the half of distribution from which the brand today has almost no visibility at all. The data is required as evidence under CSRD Scope 3 Category 12 for any brand reporting on end-of-life emissions, and under SBTi guidance for any brand committing to a Scope 3 target (SBTi Apparel Sector Guidance). The direct-to-consumer channel returns this data by default through the existing customer record. The wholesale channel does not.

Commercial value for the brand. Customer visibility in the wholesale channel that did not previously exist. Material composition, condition grade at return, and channel of return per product per customer, exposed to the brand as a Brand Intelligence subscription against a data layer the brand does not have to operate or own. The data exists because the retailer operates the take-back counter and the platform sits underneath it; the brand consumes the data as a downstream subscriber.

The failure mode this resolves. The wholesale-channel data gap is the largest single visibility problem for a multi-channel brand. The wholesale channel returns aggregate weight reports through the recycler's invoice and nothing else. CSRD does not accept weight reports as evidence for product-level reporting obligations. The data layer at the take-back counter is the only point in the wholesale chain where the brand can attach an identifier to a returned product without owning the retail relationship. The platform produces that attachment by recording the SKU at the moment the customer scans, before the product enters the downstream sort line.

4.4 The recycler

Role: the downstream operator whose industrial unit economics depend on feedstock that is identified before it arrives. The recycler is the actor most distant from the customer in physical and informational terms, and the actor whose capital expenditure has been built to compensate for that distance.

Sustainability impact for the recycler. Industrial-scale textile-to-textile recycling becomes economically viable when feedstock is identified at the point of capture rather than after the sort. Sort yield improves. The four failure modes that kill recycler unit economics in communal collection (contamination from non-textile waste; miscategorisation of fibre composition; mass-mix unpredictability; unattributable provenance) are structurally absent at the retail counter, because each is resolved by the data captured at the moment of return. (The four-mode taxonomy is the author's, drawing on failure modes documented in [28] and [4].) MRF sort-line capex is the downstream cost of upstream information absence; identification at the point of return is a candidate for reallocating that capex, not for adding to it.

Commercial value for the recycler. Identified, condition-graded feedstock at the dock. Processing cost per tonne falls. The cost-per-kilogramme bid the recycler can make to a retail client becomes defensible against a cheaper bid from a competitor who is not also delivering the data layer, because the data layer reduces the recycler's own operating cost. The flat per-recycler platform fee is the only line on the recycler's invoice that converts into a competitive moat with the retailer.

The failure mode this resolves. The European materials recovery facility for textiles costs €15 million to €40 million per build [4]. Per-tonne sort cost on automated lines runs €150 to €280, against manual sort cost of €250 to €400 [29]. The 36 per cent communal sort yield is not a fact about textile waste; it is a fact about textile waste without identification at the point of return. The capex builds the machinery to re-identify what should never have been lost. The European textile-to-textile recycling margin running from minus 25 per cent to minus 100 per cent on a per-tonne basis is the operating-line consequence of that capex servicing a feedstock specification it was built to compensate for, not to receive.

Renewcell is the most documented European case of the failure mode. €200 million raised against a chemical recycling process that worked; bankruptcy in February 2024 because the feedstock did not arrive at the specification the offtake contract required; acquired by Altor in late 2024 with the operating model unresolved [50]; [40]. The recycling technology in Renewcell's plant worked. The information layer in front of the plant did not exist. The lesson is illustrative, not predictive: capacity is not a substitute for characterised feedstock, and the case puts in plain sight the unit economics that follow when the information gap is left in place. The JRC's 2024 work on the Digital Product Passport names the structural alternative: if composition travels with the garment, the sort line collapses from a re-identification machine into a routing machine [30]. The implication for the recycler's capex base is direct.

4.5 The chain in summary

Each of the four stakeholders captures a sustainability impact and a commercial value from the same data event. The two channels travel together at every node, or they do not travel. A platform that produces sustainability impact without commercial value to the actor produces a compliance cost line a finance director will pressure-test out of the budget at the next planning cycle. A platform that produces commercial value without sustainability impact produces a transaction with no claim on the regulator's frameworks and no defensible attribution against the impact investor's diligence framework. The four-stakeholder framing is not a marketing arrangement of the value chain; it is the structural condition under which the impact thesis is operable at scale.

5. One data event, five revenue streams, five categories of impact attribution

The architectural claim of the platform is that one scan event in a retail store produces five distinct commercial revenue lines and five distinct categories of impact attribution simultaneously, from the same row in the database.

The five revenue streams are:

1. The retailer platform fee. Per-store, per-month, paid by the retailer who deploys the platform across the take-back programme. The retailer's primary line of value is the marketing acquisition outcome, not the compliance outcome.
2. The recycler platform fee. Flat per-recycler, independent of the number of stores the recycler services. The recycler's line of value is the feedstock characterisation outcome that improves their unit economics.
3. The brand access fee under the Tier 3 wholesale data layer. Paid by the brand for read access to the SKU-resolution end-of-life records from the wholesale channel. The fee is tiered by data depth and access frequency.
4. The Brand Intelligence subscription. Paid by the brand for analytical access to the data: composition trends at the SKU level across stores, return reasons, condition grades at return, lifecycle inference. Sits one tier above raw record access.
5. One-off events. Specific datasets, retrospective queries, partnership-driven analytical projects.

The five categories of impact attribution are:

1. Waste diverted from disposal. EU Categorisation Performance Indicator. Measured at the scan event as a count of items, converted to mass through the retailer's product mass data.
2. Verified consumer engagement. Performance and economic-value indicator. Measured at the scan event as a captured email plus a repeat-rate ratio.
3. Per-product end-of-life events captured (brand-attributed). Compliance evidence indicator under CSRD, ESPR DPP, and EPR fee modulation. Measured at the scan event as a per-SKU record.
4. Linear resource use avoided. GCP-aligned Headline Indicator. Measured as diverted mass multiplied by the displacement factor (the fraction of diverted material that displaces virgin material in the downstream economy rather than supplementing it).
5. GHG emissions avoided. GCP-aligned Impact Indicator. Measured as diverted mass, multiplied by the pathway-specific lifecycle emission factor, multiplied by the displacement factor.

The same metadata row supports each of the five impact attributions and each of the five revenue lines. This is the architectural difference between the in-store collection data layer and a single-channel platform. A recycling platform captures one impact category. A reverse-logistics platform captures one revenue line. A sustainability reporting platform captures one stakeholder. The data layer at the take-back counter captures all five of each, from one event, at the resolution the downstream frameworks consume.

6. The eight-link theory of change

The causal chain from the scan event to net impact has eight links. Each link has a stated relationship between input and output; an operational measurement that the platform produces today; a published peer-reviewed anchor that establishes the relationship in the literature; and, where relevant, an open empirical question that remains part of the platform's own research output over the next twenty-four months. The open questions are recapitulated in §8.

Link 1: Cohort structure

Relationship. The consumer population whose participation determines the upper bound on collection rate is heterogeneous, with three behaviourally distinct cohorts in approximately stable ratios across European retail contexts: prosocial (~18%), middle (~50%), indifferent (~32%).

Operational measurement today. The platform measures scan rates per store per week and customer repeat behaviour over rolling windows. Cohort attribution is inferred from the distribution of repeat-versus-first-time scans and from the distribution of capture rates across stores with comparable footfall.

Anchor. [1] for the cohort structure description; the companion paper [2] for empirical replication across five Dutch stores.

Open question. Cohort-level attribution of participation at deployment scale: what proportion of captured emails come from the middle cohort versus the prosocial baseline, and what proportion of new participation is incremental to the existing batch system rather than a substitution from it. Resolved through deployment-variance analysis over the next twenty-four months.

Link 2: In-flight versus batch behaviour

Relationship. The drip-style return model converts the consumer ask from accumulation to in-flight return, reducing the cognitive cost of participation and recruiting the middle cohort that the batch model under-serves. The empirical signature is approximately 1.5 items per identified customer over a nine-month window, and approximately 27% of email-captured scans from repeat customers [2].

Operational measurement today. Per-customer item count over rolling time windows. Repeat-scan ratio per store per quarter.

Anchor. [31] on the information × rewards × convenience trinity in consumer returns behaviour. [49] on transparency-premium effects in consumer purchase decisions.

Open question. The drip-versus-batch differential in net collection rate at retailer scale: what proportion of in-flight scans are incremental rather than substitutive of the batch collection a customer would otherwise have performed.

Link 3: Store-level variance in scan rate

Relationship. Scan rate at the store level is driven by four operational variables: incentive structure, QR prominence, frontline staff engagement, and the brand-mix-per-store. The brand-mix hypothesis (that captured product mix at a given store reflects existing market share at that store rather than platform bias) is testable at scale and is the platform's primary methodological control for downstream impact attribution.

Operational measurement today. Scan rate per store per week, decomposed against the four variables through deployment telemetry and partner-supplied store characteristics.

Anchor. Marketing-operations literature on point-of-sale signalling, summarised in [31].

Open question. Whether the brand-mix hypothesis holds at deployment scale across heterogeneous European retail formats. Resolved through deployment variance across the EU rollout under the 2028 EPR mandate.

Link 4: Incentive design and the reward S-curve

Relationship. The reward offered at the scan event operates on a non-linear curve with a crowding-out zone in the lower-middle range, where a small extrinsic reward reduces participation among the prosocial cohort relative to no reward at all [51]. The platform's incentive design must avoid the crowding-out zone: small reward magnitudes that demotivate prosocial participants while not yet motivating the middle cohort.

Operational measurement today. Reward redemption rate per store per reward tier. Cohort-segmented response measured through repeat-scan distributions.

Anchor. [51] on the non-linear crowding-out zone in extrinsic rewards for prosocial behaviour.

Open question. The empirical location of the crowding-out zone in the specific context of in-store textile collection in European retail, a research question that the platform's deployment variance is the only at-scale instrument for answering.

Link 5: Feedstock quality at point of capture

Relationship. Four failure modes that kill recycler unit economics in communal collection are structurally absent at the retail counter: contamination from non-textile waste mixed into the collection bin; miscategorisation of fibre composition; mass-mix unpredictability; and unattributable provenance that prevents recall or quality dispute resolution. Each is resolved by the data captured at the moment of return: the customer hands the product to the counter, the AI identifies it, the SKU resolves the composition through the brand's published material data. The 36 per cent communal sort yield benchmark is not a fact about textile waste; it is a fact about textile waste without identification at the point of return. §3.4 and §4.4 develop the downstream capex consequence in full.

Operational measurement today. Per-scan SKU resolution rate (92% across deployment to date). Per-scan condition grade. Per-scan composition resolution through brand-supplied material data.

Anchor. [4] on the 36% sort yield baseline in communal collection. [52] on the environmental footprint of textile production at 17–25 kg CO2e per kg of garment, which establishes the upper-bound benefit available if feedstock can be characterised before sorting.

Open question. The sort-yield differential between identified and communal feedstock at downstream recycler scale. Resolved through downstream-partner lifecycle assessment scheduled for completion in 2026.

Link 6: Recycler economics at industrial scale

Relationship. Industrial-scale textile-to-textile recycling is feedstock-constrained, not capital-constrained or technology-constrained at the relevant horizon. Characterised feedstock is the precondition for the technology to operate at its rated throughput and yield. MRF capex of €15 million to €40 million per facility, with a cumulative European requirement estimated at €6 to €7 billion by 2030, is the price the chain is currently paying to compensate for the absent upstream identification (§4.4). Identification at the point of return is a substitute for that capex, not an addition to it. The Renewcell bankruptcy is the most public European example of the constraint being binding, with €200 million of capital deployed against a working chemical process that the feedstock specification could not service [50].

Operational measurement today. Volume of identified feedstock delivered to recycler partners per month. Composition mix of identified feedstock per recycler per quarter.

Anchor. [4] on textile-to-textile recycling margins at −25% to −100% and on the MRF capex envelope. [40] on the Renewcell failure case.

Link 7: Net impact outcome

Relationship. The net environmental impact per kilogramme of identified feedstock is the product of three quantities: sort yield (the proportion of input mass that exits the sort as usable recycled fibre), pathway-specific lifecycle CO2e (the emission footprint per kilogramme along the specific downstream pathway: fibre-to-fibre, mechanical, chemical), and the displacement factor (the proportion of the recycled output that displaces virgin material in downstream consumption rather than supplementing it).

Operational measurement today. Sort yield through recycler partner. Pathway-specific lifecycle CO2e from published sector data [52]; [53]. Displacement factor estimate from the published rebound-effects literature [54]; [55].

Anchor. [54] for the conditional-on-displacement formulation of circular-economy GHG benefit (77–85% reduction). [53] for the reused-garment environmental-impact factor (approximately 70× less than equivalent new). [55] for the rebound-effects framing.

Open question. The displacement factor at retail scale, the binding parameter for net environmental benefit, and the platform's largest single open empirical question. Resolved through downstream-partner lifecycle assessment over the 2026–2027 horizon.

Link 8: The retailer-engagement flywheel

Relationship. Willing participation by a retailer compounds deployment quality in ways that compelled participation does not. A retailer who deploys because the take-back programme has become a marketing acquisition channel actively promotes the QR poster, trains staff to mention the programme at the counter, and treats scan-rate variance as a marketing metric. A retailer who deploys because the EPR mandate requires it does the minimum the mandate specifies. The deployment-quality differential between the two postures is large enough to shift cohort recruitment, store-level scan rate, and downstream sort yield: every variable in links 1 through 7.

Operational measurement today. Scan rate per store per week. Staff-mention rate per store, where available from partner telemetry. Internal marketing-team adoption of the platform metrics within retailer organisations.

Open question. The deployment-quality differential at scale, expressed as a scan-rate multiplier for willing-versus-compelled deployment. Inferred from the deployment-variance analysis under Link 3 over the EU rollout.

7. Indicator architecture

The platform reports six impact indicators, ranked in priority order by current operational verifiability rather than by marketing weight. The ranking is deliberate. The reader's eye lands on the strongest, most-defensible metric first; the indicators that depend on the displacement factor (the platform's largest open question) are reported in second tier with the dependency named in line.

Tier 1: Strong verifiability today (direct operational measurement)

Indicator 1: Waste diverted from disposal (kg). Category 3 Performance Indicator under the EU Categorisation System. Measured as the count of returned items at scan, converted to mass through retailer-supplied per-SKU mass data. Verification path: scan log audited against recycler weight bridge at receipt.

Indicator 2: Verified consumer engagement (emails captured; repeat rate). Performance and economic-value indicator. Measured as the count of validated email addresses captured at the scan event, plus the repeat-scan ratio across rolling windows. Verification path: email validation status logged at capture; repeat-scan ratio reconstructed from customer hash records.

Indicator 3: Per-product end-of-life events captured (brand-attributed). Compliance evidence indicator under CSRD Scope 3 Category 12, ESPR DPP, and EPR fee modulation. Measured as the count of per-SKU records exported to the retailer's evidence layer. Verification path: per-record export confirmation; SKU resolution rate audited against AI-identification confidence.

Tier 2: Partial verifiability today (depends on displacement factor or sector conversion rate)

Indicator 4: Linear resource use avoided (tonnes). Headline Indicator under the Global Circularity Protocol (WBCSD/UNEP, COP30 November 2025). Calculated as diverted mass × displacement factor. Dependency on the displacement factor named in §8.1.

Indicator 5: GHG emissions avoided (tCO2e). Impact Indicator under the Global Circularity Protocol. Calculated as diverted mass × pathway-specific lifecycle CO2e × displacement factor. Dependencies on the pathway factor and the displacement factor named in §8.1 and §8.4.

Indicator 6: Additional indicators (water, hazardous waste, non-GHG pollutants). Calculated from BCG/ReHubs sector conversion rates against diverted mass. Dependency on sector conversion rates published by third parties named in line.

The ordering creates an honest reading hierarchy. The diligence reader who scans the indicator list first encounters three indicators where the verification path is direct and operational, and only then encounters the indicators where a published-literature parameter is in the multiplication. Each Tier 2 indicator is reported with its central estimate and the range that the parameter uncertainty implies.

8. Open empirical questions: the platform's research roadmap

Four empirical questions remain open at the time of publication. Each is presented here as a research output (what the platform will measure over the next twenty-four months, with the experimental design and the partner role identified), rather than as a gap.

8.1 The displacement factor at retail scale

The binding parameter for net environmental benefit. The displacement factor is the proportion of recycled output that replaces virgin material in downstream consumption rather than supplementing it; the published literature places it in a wide range, with a long history of the rebound-effects framing [54]; [55]. The conditional-on-displacement 77–85% GHG reduction is the central published anchor; the lower bound under hostile displacement assumptions is materially lower.

The platform is the at-scale research instrument for narrowing the range. Deployment data (what was diverted, where it went downstream, what fibre it became, into which product category that fibre re-entered) provides the input set that a downstream-partner lifecycle assessment can use to compute a measured displacement factor specific to European in-store retail collection. The first LCA pass is scoped for completion in 2026, with the platform's recycler partner.

8.2 Cohort-level attribution of participation

The proportion of captured emails that come from each cohort, and the proportion of new participation that is incremental to the existing batch system rather than substituted from it. Resolved through deployment-variance analysis: store-by-store scan rates across stores with comparable footfall, decomposed against the four operational variables in Link 3, segmented by repeat-versus-first-time status, validated against partner CRM where the partner provides access. The next twelve months of European rollout produce enough variance for the decomposition to be statistically tractable.

8.3 Store-level operational sensitivity

Including the brand-mix hypothesis, that captured product mix at a given store reflects the existing market share of brands at that store rather than a platform-induced bias toward particular brands. The hypothesis matters for downstream impact attribution: if the capture mix is biased by the platform, then downstream impact figures need a correction term; if the capture mix is unbiased, the figures can be taken at face value. Resolved through the same deployment-variance analysis under §8.2, with the brand-mix decomposition adding a per-store comparison against retailer-supplied sell-through data.

8.4 Drip-versus-batch sort-yield differential

The expected improvement in sort yield from identified versus communal feedstock at downstream recycler scale. The communal benchmark is 36% [4]; the platform's hypothesis is that identified feedstock improves this by a margin large enough to shift recycler unit economics. Resolved through partner LCA scheduled for 2026: sort yield measured on identified feedstock through the FastFeetGrinded line, compared against published communal benchmarks for matched fibre composition.

The four questions are not gaps. They are the research the platform is uniquely positioned to do, because the platform produces the deployment data the research requires. Each is named in the indicator architecture and in the theory-of-change links where it is binding. None is permitted to sit in a footnote where a diligence reader would have to find it.

9. European-scale extrapolation

The Waste Framework Directive amendment [3] brings every large-format apparel, footwear, and sportswear retailer in the EU27 and EEA under an Extended Producer Responsibility obligation for textiles by 17 April 2028. The retail estate subject to this mandate is approximately 100,000 to 150,000 stores. Each will operate an in-store collection programme. Today, almost none of those programmes produce structured per-product data from the return event.

At 100,000 stores deployed at the operational rates the platform has measured across Dutch deployment to date:¹

• Approximately 14 million scan events per year
• Approximately 12 million verified consumer emails captured per year
• Approximately 3 million consumers in recurring relationship
• Approximately 5,800 tonnes per year of textile material diverted from disposal
• Approximately 2,300 tonnes per year of virgin material displaced
• Approximately 40,000 tonnes CO2e per year avoided (central estimate, range 30,000–55,000 reflecting the displacement-factor uncertainty named in §8.1)
• Per-product CSRD Scope 3 Category 12 evidence for every covered brand
• Per-product fee-modulation evidence for every EU member-state PRO

The arithmetic is linear in the number of stores deployed. The scale calculator on the landing page lets a reader pressure-test the result against alternative deployment assumptions.

European textile waste runs to 6.94 million tonnes per year. At 100,000 stores, the platform diverts approximately 5,800 tonnes from disposal. Read against the sector denominator, that is not transformative: it is approximately one-twelfth of one per cent of the total sector waste flow. Read against the trajectory denominator (the gap between the current 15.0% collection rate and the 50.0% the EU has set for 2030), it is the first measurable contribution, and the only contribution that simultaneously produces per-event evidence across all four converging frameworks at SKU resolution.

The honest framing matters. A reader who encounters the achievement figures without the sector denominator can dismiss the platform as overstated. A reader who encounters the sector denominator without the achievement figures can dismiss the platform as immaterial. The achievement comes first because the achievement is what the platform produces; the denominator comes second because the denominator names the boundary of the claim. The order is deliberate, and it survives the comms filter the Founder's Voice Guide names: if a competitor quoted this paragraph back to a potential client, would it help or hurt? The answer is it would help, because the honest framing is the most credibility-building paragraph in the paper.

Note 1. All extrapolations in this section assume per-store rates derived from operational data across Runnersworld, INTERSPORT, and EK Sport stores in the Netherlands between August 2025 and March 2026. The rates are: ~140 scan events per store per year; ~120 verified emails captured per store per year; ~30 customers in recurring relationship per store per year; ~58 kg of textile material diverted per store per year. The CO2e ranges incorporate the displacement-factor uncertainty named in §8.1; the central estimate uses the 77–85% conditional-on-displacement figure [54] at the published central value. The European deployment estimate of 100,000 to 150,000 stores is based on Eurostat structural business statistics for NACE Rev 2 sector 47.71 (retail sale of clothing in specialised stores) and 47.72 (retail sale of footwear and leather goods in specialised stores) cross-referenced against the size-band threshold the EPR Directive sets.

10. Framework alignment

The platform's impact attribution maps against every framework an informed impact investor will already recognise. The alignment is across universally accepted, internationally recognised frameworks only, no fund-specific protocols, no named-investor frameworks. The list below names each framework, the platform's alignment position within it, and the evidence the platform produces against it.

EU Categorisation System for the Circular Economy. Category 3 (Re-use / value recovery), sub-category Traceability platforms. Category 4 (Enabling Technologies). Evidence: scan-event record, per-SKU resolution, downstream chain-of-custody (the chain-of-custody implementation is roadmap, not shipping, see §11).

Global Circularity Protocol (WBCSD/UNEP, launched at COP30 on 11 November 2025) [56]. Headline Indicator: linear resource use avoided. Impact Indicator: GHG emissions avoided. Performance Indicator: waste diverted from disposal. Each measurable from operational platform data, with the displacement-factor dependency named in §8.1.

IMP 5+1 Dimensions of Impact (Impact Management Platform). Aligned across what (Indicator 1: waste diverted; Indicator 5: GHG avoided), who (the four-stakeholder framing identifies the actors; Indicator 2: verified consumer engagement), how much (the §9 extrapolation provides the at-scale magnitude), contribution (the §8 displacement-factor analysis is the contribution-specific evidence), risk (the §11 honest pending items are the risk disclosure), and outcome over time (the §9 24-month research roadmap).

GIIN Core Characteristics of Impact Investing. Intentionality: the platform's commercial model embeds the impact attribution structurally rather than reporting it externally. Use of evidence: §6 of this paper. Management for performance: the indicator architecture in §7. Contribution to industry growth: the §8 research roadmap.

GIIN IRIS+ Indicators. OI4297 (waste recovery), PI7910 (CO2e avoided), PI3905 (number of stakeholders served). Each measurable from operational platform data.

Planetary Boundaries. The platform's impact is primarily in the land-system change and novel entities (chemical pollution) boundaries, through the diversion of textile waste from landfill and incineration. Secondary contribution to the climate change boundary via Indicator 5.

UN Sustainable Development Goals. Primary: SDG 12 (Responsible Consumption and Production), SDG 13 (Climate Action), SDG 15 (Life on Land). The mapping is direct through Indicators 1, 3, 5, and 6.

ESRS E5 (Resource Use and Circular Economy) under the Corporate Sustainability Reporting Directive. The platform produces the per-product end-of-life records that ESRS E5 disclosure requires from any retailer with downstream value-chain reporting obligations. Indicator 3 is the operational measurement.

GHG Protocol Scope 3 Category 12 (End-of-life treatment of sold products). The platform produces the per-product end-of-life records that Scope 3 Category 12 reporting requires for retailers and brands. Indicator 3 is the operational measurement; Indicator 5 is the resulting per-product CO2e attribution.

CTI v4.0 (Circular Transition Indicators). Aligned across the Circular Inflow, Circular Outflow, and Water Circularity dimensions through Indicators 1, 4, and 6. The platform supplies the data layer; the reporting framework is the retailer's CTI submission.

EU Taxonomy for Sustainable Activities. The platform's activity (operation of a circular-economy traceability platform) is eligible under the Taxonomy's Transition to a Circular Economy environmental objective. Alignment with the substantial-contribution criteria is documented through Indicators 1 and 3.

SFDR Principal Adverse Impacts (PAI). The platform's deployment in a retailer's value chain produces evidence against PAI indicators 13 (rate of accidents), 15 (GHG emissions), and the optional climate-and-environment PAI indicators 1 (emissions per output) and 7 (hazardous waste). The diligence-team mapping route is straightforward.

The alignment is full-stack: every framework an SFDR fund or a CSRD-reporting retailer will already be familiar with maps directly onto the indicator architecture in §7. The platform does not require a new framework. The platform produces evidence against the frameworks that already exist.

11. What is shipping, what is roadmap: an honest accounting

This section is the dedicated treatment of pending items that the impact thesis requires to be held in plain sight. Inline footnotes elsewhere in this paper attach the dependency to the specific claim it qualifies; this section is the single reference point a diligence team can paste into their internal note.

Shipping today. The platform is operating in five retail stores in the Netherlands as of May 2026, across Runnersworld, INTERSPORT, and EK Sport, with FastFeetGrinded as the downstream recycler partner. Operational metrics across August 2025 to March 2026: 1,446 scans, 84% capture rate, 92% AI product-identification accuracy, 827 unique consumers, ~1.5 items per identified customer over nine months, ~27% of email-captured scans from repeat customers, €2.72 cost per email captured [2]. The retailer-facing platform, the customer-facing scan-and-reward flow, the recycler-facing identified-feedstock dashboard, and the data-export layer for CSRD evidence are in production.

Roadmap, not shipping. The Product Journey Layer, the downstream chain-of-custody schema that records the per-product pathway from store to recycler to processed output and into the next product, is on the platform roadmap. It is not in production. The §9 attribution figures assume the chain-of-custody is in place; the figures are reported under that assumption, and the assumption is named here.

Pending. A formal third-party lifecycle assessment is scoped for completion in 2026 with FastFeetGrinded. The LCA is the instrument for narrowing the displacement-factor range named in §8.1. The platform's claims that depend on the displacement factor (Indicators 4 and 5, the CO2e figures in §9) are reported with the central estimate and the range; the range will narrow with LCA completion.

Certification. ISO 27001 information-security certification is in initiation phase, not in hand. The platform's data-handling architecture is designed against the ISO 27001 control framework and the certification audit is scheduled for completion within twelve months. The platform's claims about data security in retailer integration follow the design-stage architecture; the certified status will follow the audit.

The INTERSPORT International Corporation cooperation agreement was signed in March 2026 at global CFO and global head of ESG level. The agreement is a 12-month framework that establishes the platform as the chosen take-back data layer for the IIC cooperative; the commercial conclusion that converts the framework into deployed installations across IIC's member retailers is the commercial work of the next twelve months.

Plain treatment of each pending item is the credibility move. The diligence team that reads §10's framework alignment first, then reaches §11 and finds these items named in plain sight, has the assurance that the alignment claims were not papering over the absences. The diligence team that does not find these items named in §11 has to re-read every alignment claim in §10 against the possibility that something is being hidden. The first reader closes the paper having done their job; the second reader has to do it again.

12. Three structurally linked patterns

The investigations underneath the at-scale argument surface three patterns that recur across the chain. They are presented here in the form they appear in the evidence, before any inference about what to do about them.

12.1 Industry-side reading failure on the regulator's choice

Directive (EU) 2025/1892 places the EPR obligation on the entity placing goods on the EU market under its own name, brand or trademark, irrespective of the selling technique used. The Commission's Impact Assessment ranked this design above municipal-led collection and above status-quo extension on the explicit logic that it internalises the cost of end-of-life in the entity placing product on the market (§1.1). The directive's eco-modulation clause scales fees with durability, reparability, and recyclability, which is operative only when the obligated party is also the design-controlling party.

The recycling industry's primary responses to the directive have framed retailers as actors to be coordinated to deliver flow into the recycling network, not as co-designers of the operating model. The trade-association statements ask for harmonised fees, clearer waste-versus-used definitions, and capital subsidy for sorting infrastructure. They do not ask how the retailer's operating system should be specified to produce the per-product data the eco-modulation signal requires. Two organisations have read the choice differently: the Ellen MacArthur Foundation, which names brands, retailers and online marketplaces as the responsibility-bearing tier; and EuroCommerce, which treats its members as the design point. Outside those two, the framing is unchanged. The regulator engaged retailers because retailers are the only actor in the textile chain that combines volume placement, brand control, design control, and consumer-interface. The industry has read this as a financing perimeter rather than as a design decision.

12.2 Customer engagement is structurally out of frame

Customer engagement at the moment of return is structurally absent from two places where it would be expected to appear. It is absent from the retail store's operating system: not because the store has decided to ignore it, but because conventional take-back enters the operating system as a violating input. Unscheduled task time at the till and fitting room. Non-sellable inventory on the balance sheet. Footprint at sales-productivity rates. The store reads take-back as a category of input the system was not specified to absorb, and treats the customer at the bin as a delivery mechanism, not as a consenting individual (§4.2).

The same engagement is absent from the recycling chain's profit and loss. Recycler annual reports, PRO operating disclosures, and EPR scheme financial statements organise their accounts around per-tonne flows and per-tonne costs (§3.3). Customer-engagement revenue does not appear as a line on the recycler's P&L because it has never been a line that recyclers contract for. Marketing industry surveys, in turn, identify first-party data acquisition as the highest-priority retail capability investment and list every channel through which it is acquired, except take-back. Two industries have drawn their commercial boundaries such that the customer-value line at the take-back moment falls between them. Neither categorises it. The line is unowned.

12.3 The cost is not hardware; it is integration

The realistic capital cost of an in-store identification stack lands at €3,000 to €5,000 per store. The cost of an MRF sort line built to compensate for absent identification lands at €15 million to €40 million per facility, with cumulative European requirements estimated at €6 to €7 billion by 2030 (§2.1, §4.4). Two to three orders of magnitude. Smartphone NFC is native to approximately 94 per cent of devices in circulation globally; the marginal hardware cost of a smartphone-based read is zero when the device is already in hand. Tags at €0.04 to €0.10 per garment are borne upstream, not on the retailer's capex line.

If the hardware is moderate and largely already present, and the downstream capex required to compensate for the absence of upstream identification is two to three orders of magnitude larger, the binding constraint is not the hardware. It is the systemic integration around it: who tags at the point of manufacture and to what specification, who owns the link between the tag identifier and the customer identity, who routes the resulting event data downstream to recyclers and EPR schemes and Digital Product Passport registries, and who absorbs the operating-model overhead of running a returns desk that is no longer a returns desk. The capex envelope is the easy part. The integration is everything around it.

12.4 The patterns are structurally linked

The three patterns are not separable. The industry-side reading failure (§12.1) leaves the operational layer between obligated party and recycler unbuilt; the absence of customer engagement from both operating systems (§12.2) is the consequence of that layer being unbuilt; the integration problem (§12.3) is the specification of what the unbuilt layer would have to do. A change at any one node is structurally linked to changes at the other two. The industry cannot read the regulator differently without also reading the customer-engagement line as in-scope, which requires the integration problem to be solved.

The reverse direction holds with the same force. Solving the integration problem without the industry reading the regulator's choice differently produces a deployed identification layer that is treated by the chain as a vendor-supplied data feed rather than as the operational layer the directive presupposes. Building the customer-engagement line without the integration that connects it to product specification produces a marketing channel disconnected from the eco-modulation signal that the directive requires it to feed back into. The patterns are linked because they are facets of one structural absence: the operational layer between obligated party and recycler that the directive sets the floor for and the industry has not yet built the ceiling on.

13. Conclusion

The European retail estate is about to collect more textile waste than it has ever collected. The infrastructure to receive returned product will exist by 2028, because Directive (EU) 2025/1892 requires it. The infrastructure to produce structured data about each return will not exist by 2028, unless someone installs it deliberately. The three patterns in §12 set out why: the regulator's choice is being read narrowly by the industry that should be operationalising it; the customer-value line that would fund the operational layer is unowned by both the industries that could own it; and the binding constraint on building the layer is not hardware cost but systemic integration.

The data layer at the take-back counter is what installs the operational ceiling on the directive's floor. The same scan event captures the customer's email at €1 to €3, the retailer's CSRD evidence at SKU resolution, the brand's wholesale-channel customer visibility, and the recycler's identified-feedstock pre-sort record, from one event, against the frameworks that the regulators and the impact investors have already converged on. The five revenue lines pay for the deployment; the five impact-attribution categories satisfy the measurement obligations. At 100,000 stores, 5,800 tonnes diverted per year, 40,000 tonnes CO2e avoided per year, 12 million verified consumer relationships per year. Not transformative at sector level. The first measurable contribution to closing the gap.

The thesis is not that the data layer is sufficient to close the gap. The thesis is that the data layer is necessary, and that the path from 15% to 50% in twenty-four years cannot be reconstructed from a trajectory that has improved at 0.72 percentage points per year for the last six. Closing the gap requires a step change. The step change requires the information layer in front of the rest of the system. The information layer requires the platform.

The four empirical questions in §8 are the platform's commitment to answering, with measured operational data, the parameters under which this thesis is true. They are named in plain sight because the thesis depends on their being true at the values the published anchors suggest, and because no reader of this paper should have to look for the dependencies in a footnote.

There is a question Tim Lee, as the author, finds himself returning to after every conversation with a diligence team or a retail buyer or a recycler partner. The 2028 EPR mandate is a known unknown. The European retail estate will operate take-back programmes whether or not those programmes produce structured data. The choice is between a deployed mandate that produces a kilogram figure and a deployed mandate that produces a per-product record. The cost of producing the per-product record sits below the cost of paid customer acquisition the retailer is already paying. The technology to produce it is shipping in five stores. The frameworks that consume it are converging at COP30 and ESRS E5 and IRIS+.

Why, then, is the default deployment posture for an EPR-mandated take-back programme, across the conversations I have had over the last twelve months, still a kilogram figure at the loading dock, rather than a per-product record at the counter? I do not have a clean answer. But I think it is the right question.

Acknowledgements

This paper builds on the empirical findings reported in the companion paper The take-back counter as a customer acquisition channel [2] (Lee, T., v2, May 2026), and on the platform deployment data produced in partnership with Runnersworld, INTERSPORT, EK Sport, and FastFeetGrinded over August 2025 to March 2026. The author thanks the partner teams for their willingness to share operational data, and the academic anchors cited throughout for the published work the impact thesis rests on.