Cazoo Ltd
Chatbot: Yext Integration
The Problem
The Customer Sales team were handling a high volume of pre-sale queries via chatbot — many of which could have been resolved through self-service content.
Customers engaging with the "I'm looking to buy a Cazoo car" dialogue often requested to speak to an agent, missing the help content already available. This suggested a disconnect between what users needed and how the information was being presented.
-
Containment rate for pre-sale dialogue was just 23.8%
-
Each agent handover cost a minimum of £25, leading to significant operational expense
It was clear we needed to improve the chatbot experience to better serve customers, reduce unnecessary transfers and drive efficiency.
Discovery Process
UX Skills Applied:
-
Research
-
Content Design
-
Service design
-
Iterative UX design
-
Prototyping
-
Usability testing
-
UI design
Discovery Focus Areas:
-
Identified dialogues causing the most user dissatisfaction
-
Flagged dialogues with high transfer request rates
-
Analysed user input just before transfers in the "I'm looking to buy a Cazoo car" flow
-
Mapped gaps in the current UX that blocked users from self-serving
-
Conducted market comparison to benchmark best practices and identify quick wins
​
Discovery Solutions
-
Surface more help articles within the chatbot to give users a clearer path to self-serve and reduce the need for agent handovers
-
Improve bot responses by tailoring replies based on user input at key points in the journey, for example, surfacing finance-related answers when relevant to increase containment
-
Redesign UX flows to move away from a rigid, linear experience, giving users more flexibility to explore different options and find what they need more easily
Market Analysis Snapshot
Platforms like Intercom, Squarespace, Zoom, Samsung, Apple and Tidio all followed a common pattern, surfacing help content and support links directly within chatbot responses. This approach made it easier for users to find answers quickly and encouraged self-serve behaviour, reducing reliance on live agents. It also set a clear best practice benchmark for improving containment and customer satisfaction.
Technical Considerations
The chatbot was built on the Salesforce platform, which came with limitations around serving dynamic content. To work around this, close collaboration with engineers was essential.
During discovery, we identified that YEXT, already in use for help content search, had the potential to be integrated into the chatbot. This opened up new possibilities for surfacing relevant articles and answers directly in bot responses, improving the user experience and increasing containment.
Concepts & Usability Testing
Hypothesis: If the chatbot can serve support content directly from the help site via Yext integration, we should be able to address most pre-delivery queries and increase containment.
​Testing the Concept
We created a prototype to validate the idea through usability testing. Feedback was positive, customers showed a clear willingness to self-serve when relevant help content was delivered by the bot before needing to speak to an agent.
Refined Solution
The validated concept from usability testing allowed us to simplify and strengthen the chatbot experience.
-
Improved UX flow enabled the bot to surface multiple recommended support articles, increasing relevance and usefulness
-
Enhanced understanding of user input allowed the bot to deliver more accurate responses, creating a smoother experience
-
Introduced a clear route back to the main menu, giving users more control and encouraging self-serve
-
Refreshed the UI to align with the brand and design system, making the experience feel cohesive and polished
Summary
​The new solution was launched as an A/B test to measure its impact against the existing experience. This allowed us to compare key metrics like containment, engagement and user satisfaction and validate the effectiveness of the changes before a full rollout.
​
Success Metrics - week 2:
-
Contact deflections (containment) - 250
-
CS cost saving of £875
-
Containment increased by 10%
​
