Pricing second-hand fashion: what changed this cycle on Tutus

A 2019 Isabel Marant Étoile coat listed in Lyon at €210 sits unvisited for six weeks because the seller anchored her price against a three-year-old retail receipt. Forty kilometers away in Grenoble, a buyer searches for the same silhouette with a hard cap of €160. Until this cycle, marketplace software treated this mismatch as a messaging problem—coaxing sellers with push notifications to lower prices by arbitrary percentages.

On Tutus, WEVONE’s second-hand fashion universe, we retired static category averages three months ago. The spring product release replaces static depreciation curves with dynamic micro-cohort liquidity modeling. Here is how the pricing engine actually works, what the data showed during the beta, and where the math still breaks.

The Depreciation Fallacy in Resale

Most peer-to-peer fashion platforms calculate suggested prices using naive linear regression: original retail price minus a fixed percentage per year of age, adjusted slightly for brand tier. This model fails because secondary garment value does not decay linearly. A mid-tier trench coat loses 45% of its monetary value the moment tags are removed, plateaus for eighteen months, drops sharply when the brand releases a revised cut, and occasionally spikes if a specific aesthetic trends on social feeds.

Sellers, meanwhile, suffer from endowment bias. They price garments based on historical acquisition cost and emotional utility rather than regional buyer velocity. In Q4 of last year, 38% of inventory listed on Tutus sat unbid for over 60 days. The capital was locked, buyers migrated back to fast-fashion alternatives, and shipping corridors went cold.

Our objective for this cycle was not to force fire sales, but to identify the precise threshold where clearance speed maximizes net payout for the seller after accounting for platform hold times and shipping fees.

How the New Engine Calculates Price Ranges

When a seller uploads a garment to Tutus, Mia’s visual and textual parsing pipeline extracts six core vectors: brand tier, material composition, structural condition, seasonal urgency, current regional inventory density, and historical clearance rates across comparable SKUs.

Rather than returning a single suggested price—which sellers routinely ignore—the interface now surfaces a three-tier liquidity corridor:

1. Fast-Liquid (3 to 7 days): Targeted at sellers prioritizing rapid ledger clearance.
2. Fair-Market (14 to 21 days): The statistical sweet spot where seller margin meets average buyer search bounds.
3. Patience-Bound (45+ days): High-margin pricing with a low probability of immediate conversion, requiring long-term listing storage.

The calculation relies on local liquidity density rather than global averages. A Sézane knit garment in Lille has a significantly higher search frequency per capita than the same item in a rural postal code. The pricing engine weights nearby buyer activity within a 150-kilometer radius to calculate transit-efficient velocity, minimizing regional freight emissions while accelerating match speed.

A Worked Example: The €150 Trench Coat

Consider a real transaction logged during the April rollout. A user in Antwerp listed a navy Burberry cotton trench coat (Condition: Very Good, circa 2018).

Under the legacy system, the default suggestion was €220 based on global brand averages. The item would have joined 114 identical listings competing solely on photo quality.

The updated engine evaluated three active parameters:

• Local density: Seven active buyers in Flanders had set alert parameters for premium outerwear under €180 in the preceding 14 days.
• Micro-cohort trend: Structured searches for mid-length cotton outerwear were up 22% week-over-week in the Benelux zone due to seasonal weather shifts.
• Historical clearance: Similar condition coats priced between €140 and €155 cleared in an average of 3.2 days.

Mia suggested a Fair-Market band of €145–€160, explicitly showing the seller that pricing at €150 yielded a 78% probability of completing escrow release within 96 hours. The seller accepted the €150 recommendation.

The listing went live at 11:14 AM. An alert triggered for a buyer in Ghent. The transaction closed at 2:03 PM at full asking price. Escrow cleared 48 hours after courier delivery confirmation.

The WEVONE Architecture: Escrow, Ledgers, and Contribution

Pricing accuracy directly feeds WEVONE’s underlying settlement architecture. When a seller accepts a recommended liquidity band, the transaction routes through the Tutus transactional escrow mechanism.

Funds from the buyer are locked in the universe ledger upon order confirmation. Because pricing alignment reduces dispute rates—buyers receiving accurately priced, realistically described items open 64% fewer item-not-as-described tickets—the standard dispute window contracts from five days to 48 hours for verified sellers.

Furthermore, sellers who consistently list within the Fair-Market or Fast-Liquid corridors accumulate Tutus Contribution Points. Higher contribution scores reduce transaction processing fees by up to 1.2% and grant priority placement in local discovery feeds without requiring paid promotional boosts. The system rewards velocity and pricing honesty over margin speculation.

Limitations and Unresolved Edge Cases

The model is far from complete. We are explicit about its current technical boundaries:

First, archive and vintage garments (pre-2000 production) break the parsing model. A 1994 Helmut Lang denim jacket does not follow standard depreciation curves; its value is dictated by collector scarcity rather than utility density. When the system detects vintage metadata tags, algorithmic price suggestions are suppressed, and the user is redirected to community appraisal threads within the Trust universe.

Second, sparse postal codes create cold-start distortions. In markets where local buyer density drops below threshold limits (fewer than 50 active weekly sessions per province), the engine defaults to national macro-trends, which overprices items in low-density regions and underprices them in high-density urban zones.

Finally, photo quality remains an unquantified variable. Mia can verify garment authenticity and structural attributes from images, but cannot fully predict how poor lighting or unironed fabric depresses buyer willingness to pay against calculated market value.

What Ships Next

The current release is live across France, Belgium, and Germany for standard apparel categories. Shoes and leather accessories are currently in closed beta testing, where wear patterns on soles present higher variance in buyer acceptance thresholds.

We do not predict that algorithms will eliminate negotiation in second-hand trade. Haggling is part of the social fabric of resale. But by replacing arbitrary listing prices with structural market data, Tutus lowers the friction of the initial offer, keeping garments moving through wardrobes rather than stagnating in closets.