Designing Difficult UX Journeys

Role: Conversation Designer & UX Writer

During my time at the Conversation Design Institute, I have worked on projects for large banking and international retail clients, focusing on improving complex journeys.

Banking and retail customers regularly interact with company chatbots for serious concerns, trusting them to provide assistance or guide them to the right help. Common queries like "I don’t recognize this payment", "I want a refund", and "My card was declined" have appeared frequently in my projects. These issues are complex and jargon-heavy, often leading to customer frustration, increased abandonment rates, and expensive handovers—situations where the chatbot should ideally excel.

To address these challenges, I analysed and annotated customer flows, combining quantitative data (CSAT, NPS scores, and abandonment rates) with qualitative insights. I interviewed key stakeholders and synthesized customer data from handovers, feedback, and reports, before embarking on rewriting and optimising the relevant flows - often as a collaboration with either CDI colleagues or client team members. In addition to this, I have also regularly conducted workshops to help in-house teams create smoother customer journeys.

The results have been entirely positive with noticeable improvements in customer satisfaction. Most recently, I optimized a large bank’s "My card was declined" journey, reducing bad handovers and abandonment rates from 30% to 13.8% after four months.

My Top 5 Design Tips for Difficult UX Journeys

1. Steer clear of Business Jargon:
Never assume customers understand internal technical terms or jargon. Confusing language leads to wrong decisions, abandonment, and bad handovers.

2. Did You Help the Customer Today?
Conversation designers should critically evaluate flows, always asking if they effectively helped the customer achieve their goal through the chat.

3. Harness Customer Data
Understanding customer journeys is crucial. Analyse the intents that customers use to start a chat, where they abandon conversations, and what they say when they get through to live agents. This context is essential to optimise flows.

4. Avoid Assumptions
Empathy requires careful wording. Phrases like "Don’t panic" or "It’s easy" can unintentionally amplify concern or talk down to the customer.

5. Don’t give extra work to the customer
Worried customers don’t want to be directed to long FAQs or be told to dial a phone number and wait for a live agent. If you can't offer a direct solution, then it's much better to manage expectations early and be upfront with the chatbot can do.