A Chatbot for Fashion Brands That Answers Size Questions and Keeps Exchanges under Control

The Challenge

The brand sells through its own stores, its website and a busy Instagram account, and most customer messages arrive as Instagram direct messages and WhatsApp chats. The same questions came again and again: does this style run small, which size should I take, is the medium still available in black, and can I exchange it if it does not fit? Store staff answered from their own phones between customers, and on busy days replies were slow or never sent.

Exchanges were the bigger problem. Each store handled them its own way. Some staff exchanged items without a receipt, some set aside a replacement size for a customer who never came back, and returned items often sat in the stockroom without being received back into stock. Stock was tied up in limbo, staff spent their time searching for receipts and holding items, and finance could not match credit notes to what had physically come back.

• Size and availability questions answered slowly, from personal phones
• No single exchange process across stores and website orders
• Replacement sizes held for customers with no record and no expiry
• Returned items left in stockrooms instead of being received back into stock

Our Approach

Discovery brought together customer service, store managers, the warehouse and finance. With personal details removed, we read through a sample of recent chats, grouped the questions by topic, and walked through every way an exchange happened in practice: in store with a receipt, in store without one, and for website orders delivered to the customer's door. The pattern was familiar from other Lifestyle and Fashion brands: the questions were predictable, but the exchanges were not.

From that map, the brand's management agreed one written exchange policy covering the exchange window, the condition of the item, proof of purchase and the cases that need a manager. The fit-gap review showed that StarBot could answer size and product questions from an approved knowledge base and look up live stock and order records in Odoo, while Odoo Enterprise Helpdesk, Inventory and Point of Sale covered the exchange itself through configuration and a small amount of Studio work. No custom module was needed.

We worked in phases: the knowledge base and size guidance per style first, then a StarBot pilot on the website widget, then WhatsApp and Instagram, and finally the Helpdesk exchange process and the Point of Sale changes in every store. Testing covered sample conversations in Arabic and English, every exchange path, and the stock and accounting documents each path creates. Training was short and role-based for customer service agents, store staff and supervisors.

The Solution

Size and Product Answers within Seconds

StarBot answers on WhatsApp, the website widget and Instagram direct messages from one knowledge base built on the brand's size charts, fit notes, fabric and care information. Customers describe what they need in Arabic or English, and StarBot replies with the size guidance for that style, product cards with images, and live availability by size and color from Odoo Inventory. When a customer wants a second opinion on fit or styling, the chat passes to a colleague in store or in customer service with the transcript attached.

One Exchange Request, Captured Once

For website orders, a guided flow asks for the order number, the item, the reason and the size wanted, checks the order in Odoo and explains the policy in plain words. Requests that fit the policy create a ticket in a dedicated Odoo Helpdesk team with the order already linked, and the customer can ask StarBot where the request stands at any time. Requests outside the window or without proof of purchase go to a person rather than being refused by the bot. Customers who bought in store are told what to bring and how an exchange works at the till.

Website Exchanges through Odoo Helpdesk

The Helpdesk team uses the after-sales Return action on the ticket, which creates a reverse transfer into a dedicated returns location where items are checked before they go back into sellable stock. Because a size exchange means a different variant, the new size goes out on a new sales order once the returned item has been received and checked, so stock is reserved only for exchanges that are really happening. When a customer wants the same item again, for example because of a fault, the Return for Exchange option creates the replacement delivery in the same step. Any refund is issued as a credit note from the same ticket. A Studio approval rule adds a manager's sign-off before a ticket tagged as a policy exception can be returned or refunded, and the return, the credit note and each approval can be traced from the ticket.

Exchanges at the till

In store, an exchange starts from the original receipt in the Point of Sale orders list, so the returned item is refunded against the sale it came from, Odoo generates the credit note automatically and the replacement is sold as an ordinary sale. Point of Sale access levels let sales associates sell but not refund, so every refund passes through a supervisor, and an exchange without a receipt is treated as a policy exception. When the customer prefers credit to a refund, the amount is issued as a gift card. Returned items go back into the store's stock through the Point of Sale instead of waiting unrecorded in the stockroom.

The Results

• Customers get size guidance, fit notes and live availability within seconds on WhatsApp, the website and Instagram, in Arabic or English.
• Store staff no longer answer routine size questions from personal phones between customers.
• Every exchange follows one process, whether it starts in a chat, at the till or from a website order.
• Replacement sizes are reserved only for confirmed exchanges, so stock is no longer held for customers who never collect it.
• Returned items are received into a returns location and checked, so stock records match what is physically on hand.
• Finance can match every credit note to a ticket or receipt, and to goods that actually came back.
• Exceptions are decided by a manager or supervisor and recorded in Odoo, instead of being agreed informally on the shop floor.

Lessons for Similar Projects

1. Write the exchange policy before configuring anything. A chatbot and an ERP can only enforce rules that management has agreed.
2. Keep size guidance per style rather than one generic chart, and refresh it with every new collection.
3. Let the bot hand exceptions to a person instead of refusing them. Customers accept a firm policy more easily from a colleague.
4. Separate the person who takes goods back from the person who approves the refund, both in store and online.