/projects/messaging Booking.com

Booking.com offered a messaging service for traveller-accommodation owner communication, but its effectiveness varied across accommodation types. Our objective was to improve this chat service to offer convenient access to information and requests for all travellers.

I spearheaded the process and interface design for the Customer Service-facing product. I collaborated closely with a team of 5 designers and cross-functional team members, including a Front-end Developer, 2 Backend Developers, and a Product Manager.

↑6 designers collaborating across 4 teams, each working within their own cross-funtional teams.

Navigating uncharted waters

My initial challenge was to understand the evolving role of the Booking Assistant within the chat landscape, a concept still taking shape. Collaborating with cross-functional teams, we eventually settled on positioning the Booking Assistant as the pivotal point of conversations, orchestrating messages among the various participants, guest, accommodation, and customer service.

↑The Booking Assistant (BA) was positioned as the anchor in the 4-way chat, using topic detection to identify the best recipient and forwarding the messages on the sender's behalf.

The project's outset was marked by a sea of unknowns. The first version of Messaging for CS would be experimental, seeking answers to questions such as how message handling would align with existing CS channels, the potential length of message threads, and the preference for handling multiple simultaneous threads. To gain insights, I began by gathering references from group chat interfaces.

Designing V1 for the CS Specialist pilot team

The creation of V1 was a journey of experimentation. My objective was to replicate the CS Specialist workflow as closely as possible while introducing the ability to manage multiple chat sessions concurrently. Daily collaboration with the CS Specialist pilot team was instrumental in fine-tuning the interface, slowly leading to clarity of what constitutes an ideal workflow and interface for Customer Service.

The uncertainty of whether CS Specialists would need to continually switch between the messaging page and Reservation page for every booking was a recurring question. In response, we investigated ways to streamline processes by importing information from the Reservation page into V1 messaging.

To facilitate a deeper understanding of where breakdowns in the chat threads were happening I organised shadowing sessions with the CS team, accompanied by colleagues from the tech and product teams. Additionally, regular user tests in our in-office research lab allowed us to share knowledge among designers, pinpoint areas for improvement, and maintain a holistic perspective.

Key discoveries from V1

• The distribution of messages posed challenges for Workforce Management's scheduling. 1.
• Messages occasionally pertained to different reservations, leading to confusion. 2.
• The Booking Assistant's categorisation of messages sometimes fell short. 3.
• The Reservation page remained a pivotal tool for handling messages. 4.

Crafting V2: scaling the impact

As we prepared to scale the pilot team to 40 specialists across multiple locations, including London and Grand Rapids. Leveraging the knowledge gained from V1, I set out to design a comprehensive process for CS messaging that could be scaled to the entire CS organisation.

↑The V2 flow including new actions added to address key discoveries from V1.

Crafting V2 involved tackling essential aspects, including message distribution, forwarding messages to different reservations, and integrating the reservation page.

↑Message distribution was integrated with the existing CS ticket distribution tool called Picklist.

I travelled to London with a few of my team to host workshops with CS Specialists who were joining the growing pilot team for the first time. Their fresh perspectives confirmed the changes we were proposing to the interface and process for V2 were moves in the right direction.

After successfully scaling the pilot team, we launched messaging for the entire Customer Service organisation. Here, I presented our work to the other CS development teams and prepared an interactive platform in our development environment to showcase various message types as part of my handover.

Impact

• A new communication channel successfully implemented to the entire Customer Service organisation, more than 12000 Specialists at the time.
• The usage of templated Quick replies by CS Specialists helped train the Booking Assistant’s topic detection and automated replies.
• The range of quick reply templates grew to cover more than 70% of messages sent by CS Specialists, helping them respond to messages faster and with high accuracy.
• Certain ticket types saw on average a reduction of approx. 20% when handled via Messaging versus other communication channels.
• Customers had faster access to information regarding their upcoming reservations leading to higher satisfaction.

Key performance metrics

• Tagged Topics and Quick Replies Usage
• CS Tickets per Customer/Concurrency
• Average Ticket Handling Time
• Accommodation Partner Messaging and Reply Rates
• Tickets Solved by the Booking Assistant Topics
• CS Operational Impact

Reflection

This project not only drove improvements in user experiences but also marked a significant chapter in my career. It reinforced the importance of a user-centric approach, continuous collaboration, and stakeholder engagement. It's a testament to the value of understanding the end-users' needs and fostering strong interdisciplinary partnerships to deliver innovative and impactful solutions.

Booking Assistant press release