Conversation Studio
Externalizing our already trusted travel reputation to make it easier for Partners to deploy Virtual Assistants across Companies and across travel Brands.
Jul 2022 - Dec 2024
TLDR
In an initiative to externalize already proven internal products, Expedia sold our internal virtual assistant-building tool to hotel travel partners. The design work was challenging and extremely complex as we made iterative changes and improvements while onboarding new partners. I relied on my strong communication skills and the ability to understand conversational design and chat systems.
My Role
Make a tool for building chatbots so easy that a hotel administrator or technologist could build and deploy a chatbot on the hotel website. The task was easy to understand but difficult to execute. Our internal tool was complicated with technical language and had a user interface that required the average user to write coding logic. As the lead designer, I had to find multiple ways to simplify the product and make it easy for anyone to build and deploy a virtual assistant...think VoiceFlow but specifically for travel.
What
Our team at Expedia had a chatbot that we deployed across all Expedia products. The bot handles a great deal of intents specialized for travelers and hotelers. The UI was also built for low-code users to make changes within the conversation without the need to code. Externalizing this product and making it a low code / no code product would help hotels deploy a bot quickly and make changes that would help travelers get questions answered and reduce call volume.
Why
87% of hotels in the US do not have enough staff and 36% describe the shortage as severe. Ways to increase self-service in hospitality alleviate the burden of these shortages and lead to more satisfied customers. Virtual assistants or chat bot offer quick ways to find answers to questions without calling the hotel, searching online, or connecting with customer support. Thus, saving time, mitigating cost, and improving customer experiences.
How
We conducted generative research to understand the problems faced by our potential partners in the hotel chat space. User interviews were conducted to gain insights into the tools that were being used, strengths, and weaknesses. I led workshops with our product partners to incorporate unique features, such as variables that can collect, store, and recall relevant information within the bot. We tested new wireframes and versions internally and received feedback on features, layout, and usability from our virtual assistant building team.
What we learned
-
Creating virtual assistants requires the use of complex logic. Initially, we intended to target a wide audience, but eventually shifted our focus to hotel development teams. We aimed to make it easier for these teams to deploy virtual assistants on their sites and make changes over time. The next version of the product will target further simplification of the application so that business operations or managers can also make changes to the virtual assistant. Using a phased approach helped narrow our focus and use an agile approach.
-
Our generative research revealed that current tools for building VA’s were extremely labor intensive and our travel-specific intent library expedited deployment and offered an advantage over our competitors.
