Key Takeaways From a Design Case Study of a Conversational Interface.
During the IxDA week, I attended the workshop “Designing New Futures with AI and Conversational Architecture” by R/GA. While giving the presentation, the team talked about the design process of a chatbot they developed called Rose.
Context:
Cosmopolitan, Las Vegas, is a luxury hotel-casino that also hosts many other amenities like restaurants, bars, spa treatments, events, art galleries, etc. The hotel faced a problem; though it was a luxury hotel with regular guests, its other amenities were unknown to their guests. This was primarily because the cosmopolitan was a skyscraper and when amenities are vertically distributed which makes it tougher for guests to be aware of the facilities. This led to guests to not use those facilities reducing revenue collected by the organization. For this, the clients approached the marketing firm R/GA. [1]
Design Summary:
The team decided to design Rose, a Hotel Chatbot, that could help guests be aware of the many perks of being at the cosmopolitan. Over the course of 12 weeks, they met with every department within The Cosmopolitan to learn the secrets and surprises the typical guest wouldn’t find on their own. With a ton of information, the team leveraged the user experience team to identify key conversation categories that would help guests experience the property through their interests. Once the team designed the unique voice of Rose, over 1,000 conversation threads were created to offer guests ways to book experiences such as restaurant reservations, spa treatments, event tickets, and spontaneous adventures like self-guided art tours. To help drive guests to bars and clubs, Rose gives them insider information, like secret menu items, and offers to help increase spend on the property. Hotel guests are introduced to Rose by receiving her card with their key at the registration desk and are encouraged to ask Rose for anything during their stay. [1]
This article will walk through the design process while giving key takeaways that I learned for designing a conversational interface.
1. Giving a backstory to the chatbot
While most conversational interfaces have a generic persona, Rose was given a rich backstory. Rose was an upbeat and flirtatious woman who had a dream of making it in Hollywood. When she couldn’t realize her dream of being a star, she one day made her way to Vegas and decided to stay there for the many joys the city had to offer. Rose is a woman with secrets, and she knows her way around the cosmopolitan. If there’s anything you want, Rose was the one with the answers.
This gave the conversational designers fauna to work on while designing the conversations rose would have with guests. Her sultry, coy nature made her responses more engaging to users, which helped users see a chatbot in a completely different way. Having a backstory has some advantages beyond just having a rich experience for the user which will be discussed throughout the article.
2. Keeping the conversation going
Most interactions have a user flow that starts at what the user wants and completes at the user achieving what he/she wants. These interactions are designed to be short, and for good reasons. However, Rose was designed in a way to keep a guest engaged in a conversation. This helped rose make guests aware of special offers, event tickets, and other benefits of being a guest at the Cosmopolitan while also driving up revenue of other businesses within the building.
Giving Rose a rich backstory helped design these interactions in a more human-centered way and avoided the interactions seeming like unwanted advertisements to the guests. Developers found guests interacting with Rose at a much higher rate than traditional chatbots. Keeping the conversation going had an additional benefit of helping users be more aware of the capabilities of the application. This will be elaborated in the next part.
3. Using trigger words
As we know, conversations are very unlike user actions on GUI. A single message can be delivered to a chatbot in many different ways. While chatbots are designed with this in mind, it can also be helpful using trigger words. Trigger words can help a word or phrase register in a users mind to perform a specific task. For example, a system can say,
“Tell me what you’re in the mood for, take outs, light bites, sit down, buffet, or price fix.”
In this sentence, notice that the bot is giving suggestions about trigger words that would register in a user’s mind. Let’s say a user is in a mood for some “light snacks,” the user would know to use the word “light bite” while talking to Rose. Helping Rose have the ability to have a continued conversation engagingly can help slip in these trigger words in the mix of conversations, helping guests learn how to use the application and learn new things the chatbots have to offer
4. Having fallback dialogues
Given the flirtatious nature of the chatbot, it was understood that some users might have some inappropriate things to say to the chatbot. Though these are unwanted, having no response would be even worse for user experience. Also, simply saying, “I’m not sure about this” would be an unimpressive way to end a conversation. Rose was designed to have fallbacks for specific events that could steer the conversation in a different direction.
“[Rose’s] playful personality has wooed hotel guests who booked directly into spending 37% more than guests who do not engage with her “— Mamie Peers | Vice President, Digital Marketing, Cosmopolitan of Las Vegas
Rose was an interesting case study of a conversational architecture, giving HCI practitioners a unique way of looking at chatbots and voice UIs. While personas like Alexa and Siri have a specific voice and some interesting answers for unconventional questions, Rose’s rich backstory and customer engagement helped drive user experience beyond what a user might expect.
[1]“Rose, the Hotel Chatbot | R/GA.” R/GA Transformation at Speed, www.rga.com/work/case-studies/rose-the-hotel-chatbot.