How is AI used in fashion customer service, and where does it fail?
AI chatbots and agent assistants can answer order, return and sizing questions around the clock. Customers still want a human when it matters, and companies stay liable for what their bots say.
KEY TAKEAWAYS Summary by the editors
- AI in fashion customer service handles routine contacts such as order tracking, returns, exchanges and size questions, and assists human agents with drafting and summarising.
- A Gartner survey of 5,728 customers conducted in December 2023 found that 64 percent would prefer companies did not use AI for customer service, with difficulty reaching a human the top concern.
- In Moffatt v Air Canada (February 2024), a Canadian tribunal held the airline liable for incorrect information given by its chatbot, rejecting the argument that the bot was a separate entity.
- Article 50 of the EU AI Act requires AI systems that interact with people to inform them that they are dealing with AI, unless this is obvious, from 2 August 2026.
- Reliable service AI depends on accurate order, policy and product data and a clear, fast handover to human agents.
AI in fashion customer service answers routine questions about orders, deliveries, returns and sizing, and helps human agents work faster by drafting replies and summarising cases. It works well for high-volume, well-documented requests, but it fails when it guesses, blocks access to a human or gives answers the company has not verified, and the company remains responsible for those answers.
What can AI do in fashion customer service?
Fashion e-commerce generates large volumes of similar contacts: where is my order, how do I return this, can I exchange for another size, how does this fit. Many are answerable from order systems and policy documents. That makes them suitable for automation. According to McKinsey's State of Fashion 2026, more than 35 percent of fashion executives report already using generative AI in areas including online customer service.
| Use case | AI role | Data required | Automation suitability |
|---|---|---|---|
| Order status and delivery | Retrieves tracking and explains delays | Order management and carrier data | High |
| Returns and exchanges | Explains policy, starts returns, offers size exchange | Return policy, order history, stock by size | High, with rules for exceptions |
| Size and fit questions | Uses garment measurements and fit notes | Size charts, product specifications, return reasons | Medium |
| Product questions | Answers material, care and availability questions | Complete product attributes and care data | Medium |
| Complaints and damaged goods | Triage, summarise and route to an agent | Case history, photos | Low, human-led |
| Agent assistance | Drafts replies, summarises history, suggests next steps | Knowledge base, CRM | High, human remains in control |
How do generative AI chatbots differ from older bots?
Earlier chatbots followed decision trees and keyword rules. They were predictable but rigid, and customers quickly hit dead ends. Generative AI chatbots understand free text, handle several languages and can answer questions they were not explicitly scripted for. That flexibility is also the risk: a language model can produce a fluent answer that is wrong.
The standard mitigation is retrieval-augmented generation: the bot is restricted to answering from approved sources such as the returns policy, product data and the customer's order record, and it is instructed to hand over when it cannot find an answer. Actions such as issuing refunds are executed through controlled system integrations rather than left to the model's judgement.
Do customers want AI customer service?
Customer attitudes are a real constraint. Gartner surveyed 5,728 customers in December 2023 and found that 64 percent would prefer companies did not use AI for customer service, and 53 percent would consider switching to a competitor if they learned a company was going to use AI for it. The top concern was that AI would make it harder to reach a person, followed by job losses and incorrect answers. Gartner's recommendation was that AI chatbots must clearly offer a route to a human agent and carry the conversation over so customers do not have to repeat themselves.
This does not mean customers reject fast self-service for simple tasks such as tracking a parcel. It means they object to AI used as a barrier.
Fashion adds its own nuances. Many contacts are about fit and returns, where a customer wants reassurance as much as information. A bot that can check the order, explain the return window and immediately offer an exchange in the right size resolves the issue; a bot that answers with a generic link to the returns policy tends to generate a second contact, often more frustrated than the first. Measuring re-contact rates is therefore more useful than measuring how many conversations the bot "handled".
Who is liable when a chatbot gives wrong information?
The company is, at least on the evidence of the most cited case. In Moffatt v Air Canada, decided by British Columbia's Civil Resolution Tribunal on 14 February 2024, the airline's chatbot had wrongly told a customer he could apply for a bereavement fare discount after travel. Air Canada argued that the chatbot was a separate legal entity responsible for its own actions. The tribunal rejected this, held the company liable for negligent misrepresentation and said it remained responsible for all information on its website, whether from a static page or a chatbot. The amounts were small, but the principle matters for any retailer whose bot explains return windows, warranties or promotions.
In the EU, Article 50 of the AI Act adds a transparency duty from 2 August 2026: AI systems intended to interact directly with people must inform them that they are interacting with AI, unless this is obvious from the context. For most retail chat interfaces, a clear label at the start of the conversation is the simplest way to meet this, and it also manages customer expectations about what the bot can do.
How should a fashion brand implement AI customer service?
- Analyse contact reasons: identify the high-volume, low-complexity intents that are well documented.
- Fix the knowledge base: policies, size charts and product data must be correct and consistent before any bot reads them.
- Integrate systems: connect order, returns and stock systems so answers reflect the actual case.
- Set boundaries: define topics the bot must never decide alone, such as goodwill refunds or complaints about harm.
- Design the handover: pass the conversation history to the human agent automatically.
- Monitor quality: review samples of conversations, track resolution and re-contact rates and customer satisfaction, not only deflection.
What are the limits of AI in customer service?
AI handles structured, repetitive requests well and complex, emotional or unusual cases badly. Luxury and premium brands, where service is part of the product, may decide to use AI mainly behind the scenes to support human advisers. Costs also go beyond licences: integration, knowledge management, quality assurance and ongoing tuning require people. The most robust approach treats AI as a way to give customers faster answers on simple matters and give human agents more time for the cases that need judgement.
Frequently asked questions
How do fashion retailers use AI in customer service?
Mainly for order tracking, returns and exchanges, size and product questions, and to help human agents draft replies and summarise cases. Complaints and complex cases are usually triaged by AI but handled by people.
Do customers like AI chatbots in customer service?
Many are sceptical. A Gartner survey of 5,728 customers found 64 percent would prefer companies did not use AI for customer service, mainly because they fear it will be harder to reach a human. Clear handover options help.
Is a company liable for what its AI chatbot says?
In Moffatt v Air Canada (2024), a Canadian tribunal held the company liable for incorrect information from its chatbot and rejected the idea that the bot was a separate entity. Retailers should assume they are responsible for their bots' answers.
Do chatbots have to disclose that they are AI in the EU?
Yes. From 2 August 2026, Article 50 of the EU AI Act requires AI systems that interact directly with people to inform them they are interacting with AI, unless this is obvious from the context.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
SOURCES
- Gartner: Survey finds 64% of customers would prefer that companies didn't use AI for customer service
- American Bar Association, Business Law Today: BC Tribunal confirms companies remain liable for information provided by AI chatbot
- EU Artificial Intelligence Act: Article 50, Transparency obligations
- McKinsey & Company: The State of Fashion 2026
