How does ThredUp use AI to list and sell second-hand fashion at scale?

KEY TAKEAWAYS Summary by the editors

1. ThredUp said in March 2024 that AI generates product descriptions and attributes from item photos, which it says speeds up categorisation and improves listing accuracy.
2. In August 2024 ThredUp launched natural-language search, image search and an AI 'style chat' that builds complete outfits from prompts such as 'outfit for a fall wedding'.
3. ThredUp said in August 2024 it had processed more than 200 million unique second-hand items from 60,000 brands across 100 categories.
4. ThredUp's chief executive said in March 2025 that sessions using image search converted at 85% higher rates, and the company also cited AI-powered automated digital measurements.
5. In Q2 2026 ThredUp reported revenue of $90.8 million, up 17%, and a record 1.77 million active buyers, up 21%, without attributing the growth to specific AI features.

ThredUp uses AI at both ends of its resale model. On the supply side, AI generates descriptions, attributes and measurements for second-hand items so they can be listed faster and more accurately. On the demand side, natural-language search, image search and an AI style chat help shoppers find specific pieces in an inventory of millions of one-off items. The company argues that resale benefits more from AI than conventional retail because every listing is unique.

Why is AI especially useful for online resale?

A conventional retailer receives product data from the brand. A resale platform receives a bag of used clothes. Each item must be identified, categorised, measured, photographed, described and priced individually, and the resulting catalogue is too varied for simple keyword search. ThredUp said in 2024 that it had processed more than 200 million unique second-hand items from 60,000 brands across 100 categories, with millions of unique items available at any time.

Chief executive James Reinhart has repeatedly argued that generative AI 'disproportionately benefits marketplaces like thredUP', because it can create the structured product data and search experience that one-off inventory otherwise lacks.

How does ThredUp use AI to process and list items?

On its March 2024 earnings call, as reported by EMARKETER, ThredUp said AI generates product descriptions and attributes from item photos. The company said this enriches its inventory database, speeds up categorisation and improves listing accuracy, which in turn supports search and personalisation. Reinhart also said that because second-hand items lack model photography, he saw generative AI eventually replacing manual photography, a stated ambition rather than a reported deployment.

In March 2025, according to PYMNTS, ThredUp cited AI-powered automated digital measurements to improve the accuracy of listings, and AI to improve inventory management.

Read also

How does The RealReal use AI for authentication, pricing and intake?

What AI shopping tools has ThredUp launched?

In March 2024 ThredUp added an AI-powered search tool that lets shoppers search by occasion, vibe and attribute, for example a 'Sunday brunch dress'. EMARKETER reported that the company cited early increases in searches per session, add-to-cart rates from search and click-through rates, without figures.

In August 2024 ThredUp rolled out three tools to its site, according to Chain Store Age:

• Natural-language search, from simple terms such as 'floral dresses' to detailed requests such as 'swimsuit for a triathlon'.
• Image search, where shoppers upload a photo, take a picture of a garment or share an Instagram post to find matching or similar items.
• Style chat, an AI chatbot that builds complete outfits from prompts such as 'outfit for a fall wedding', which shoppers can then adjust by colour, style and coverage.

In March 2025 Reinhart said: 'The sessions with image searches result in 85% higher conversion rates.' The same report noted new customer volume up 32% year on year. As with most feature-level figures, image-search users are self-selected and likely to be high-intent shoppers, so the comparison does not prove causation.

What results has ThredUp reported since?

ThredUp launched its AI shopping tools during a difficult period: in Q2 2024 revenue fell 4% to $79.8 million amid weak European business, as MediaPost reported, and ThredUp sold most of its European operations in Q4 2024, according to PYMNTS. By Q2 2026 ThredUp reported revenue of $90.8 million, up 17%, a record 1.77 million active buyers, up 21%, and 1.87 million orders, up 22%. The release did not attribute this growth to specific AI tools, and the company lowered its second-half guidance, citing economic uncertainty for price-sensitive buyers.

How does AI support ThredUp's work with brands?

ThredUp also runs resale programmes for other retailers and brands, which it calls resale-as-a-service. Chain Store Age listed partners including Journeys, The Container Store, Toms, American Eagle, H&M, Kate Spade and Francesca's. These programmes rely on the same processing pipeline: items are received, identified, described, measured and listed by ThredUp, then sold through branded resale shops.

For a brand, this makes the quality of ThredUp's AI-generated data part of its own customer experience. A wrong attribute or measurement on a branded resale listing reflects on the brand as much as on the platform. Brands entering resale partnerships should therefore ask how item data is generated, what share is reviewed by people, and how errors are corrected.

ThredUp has not published details of how, or whether, AI sets its resale prices, so pricing should not be assumed to be automated. Reinhart has framed the overall approach as deliberate, saying in March 2025 that 'our AI strategy was developed with a deeply calculated approach to build and sustain a competitive advantage over time'.

Read also

Resale and recommerce: how second-hand reached the brands

What are the risks and lessons?

The main risk in AI-generated listings is accuracy: a wrong fabric, size or condition note on a one-off item leads to returns and disappointment that cannot be fixed by restocking. Generated imagery raises a similar trust question. Image search and style chat also depend on how well items are tagged, so errors in processing flow straight into discovery.

1. Start with structured data from photos, because every later feature depends on it.
2. Measure items automatically where possible; measurements matter more in resale than size labels.
3. Offer visual and conversational search, since shoppers often know a look rather than a product name.
4. Label generated imagery clearly and keep real photos of the actual item.
5. Track conversion and returns by feature with proper controls before scaling claims.

Frequently asked questions

How does ThredUp use AI?

ThredUp uses AI to generate product descriptions, attributes and measurements from item photos, and to power natural-language search, image search and an AI style chat for shoppers. It also says it uses AI to improve inventory management.

What is ThredUp style chat?

Style chat is an AI chatbot launched in August 2024 that builds complete outfits from prompts such as 'outfit for a fall wedding'. Shoppers can then refine the suggestions by colour, style and coverage.

Does ThredUp's image search increase sales?

ThredUp's chief executive said in March 2025 that sessions using image search converted at 85% higher rates. Because shoppers who use image search are often already high-intent, the figure shows association rather than a measured causal effect.

Does ThredUp use AI-generated images?

ThredUp's Q2 2026 results release notes that 'some images are AI generated'. Its chief executive said in 2024 that he saw generative AI eventually replacing manual photography for second-hand items, which lack model photos.

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