IKEA Chatbot

Conversational self-service for customer support

The problem

Customers contacted IKEA for questions they could often answer, or solve, themselves. A chatbot could handle the bulk of these conversationally.

My role

Joined as UI designer, became sole UX/UI designer after six months. UX and UI design, prototyping, user testing. 1.5 years on the project.

Approach

Working from real chat transcripts across selected markets, we covered the most common intents like order status and product availability. I designed Billie, a gender-neutral chatbot offering conversational self-service, links to supporting pages, answers pulled from IKEA's knowledge base, and a graceful handover to a co-worker when needed.

I built the UI framework for presenting conversations, including non-conversational content like external pages, and prepared all elements in Figma for handoff. Intent analysis ran on Google Dialogflow.

Learnings

My first conversational design project, a different process built around flows in Miro, constantly bounced off the backend developers to keep them buildable. Rule-based and pre-LLM, but the core problems are exactly what AI assistants face today: reading intent, knowing the limits of self-service, and handing over well.

Get in touch

Galveston AB
Murbruksgatan 8
216 46 Limhamn
Sweden
Phone: +46732439720
Email: johan@galveston.se
LinkedIn: @galveston