Why a Hotel’s Chatbot Concierge Named Rose Created More Problems Than It Solved

If a chatbot named Rose is supposed to provide anything a guest might need during their stay at the Cosmopolitan Las Vegas, why is it that Rose was unable to deliver on a simple request—to make a dinner reservation?

While traveling to Las Vegas to attend an IBM Watson conference, I arrived at the Cosmopolitan hotel in the early evening feeling tired and hungry. When I checked into the hotel, I noticed there was a long line of people waiting to see the concierge, so I asked the attendant checking me in at the help desk if I could schedule dinner reservations. 

“You can either talk to the concierge,” he nodded toward the line which was at least twenty people deep, “or you can text Rose.” 

He handed me a black card that read, “Know my secrets. Text me.” The name ‘Rose’ and a phone number were in bold white type below the tagline. 

“You can either talk to the concierge,” he nodded toward the line which was at least twenty people deep, “or you can text Rose.” 

I went to my room and messaged Rose. “Hi Rose, I’d like dinner reservations at STK for 8 pm tonight.” 

Several minutes lapsed before Rose responded. 

“Hi, I’m Rose. I can help fulfill any need you may have. We have five world-class restaurants here in the hotel. What can I help you with?” Rose listed five restaurants, including STK which I had originally requested. 

“Hey, Rose. I want reservations at STK at 8 pm tonight.” 

I expected Rose’s response to be either, “Hi, I can help with that and confirming reservations,” or, “there’s no availability for that time frame. Let me provide you other options.” 

Instead, Rose introduced other aspects of the hotel while adding suggestive comments about how she could provide anything I wanted. Fifteen minutes passed before I stopped texting back. I called the restaurant, checked for availability, and made my own reservations. 

I didn’t contact Rose again. Rose did, however, message me the following day to introduce herself again saying, “Hi, I’m Rose. I can help you with any need you may have.” 

Story Playback and Action Steps

I expected Rose to act and replicate human behavior, especially since it was being represented as a persona named Rose. Think Siri or Alexa. 

Rose created more problems than she solved. Both for me and the Cosmopolitan. For me, because I wasted time making dinner reservations. For the Cosmopolitan, because I equated it with bad customer service. Even for Rose, because she was set up to fail. 

Let’s playback this story to see what could have been done to create a better user experience. 

Align Instructions with Expectations 

“Text Rose.” These were the instructions I was provided when I asked the staff how to make dinner reservations. 

I followed instructions, so why was it that Rose couldn’t understand and continue the conversation I started? Why couldn’t Rose execute my direct and to-the-point request? 

Simple. Rose was a chatbot that was following a pre-programmed script. Go off script, and Rose will have trouble following the conversation. Instead of following my lead, Rose was attempting to get me on her script. 

My expectation was that Rose could help accomplish my request, just like any other concierge would. The reality was that the AI/Machine Learning behind Rose didn’t have the ability to pick up on subtle cues needed to provide this level of authentic dialogue. 

With that in mind, if Rose is a question-and-answer program that will be thrown off by direct questions, then Rose should have initiated a script from the beginning. I would have followed her lead. 

My expectation was that Rose could help accomplish my request, just like any other concierge would. The reality was that the AI/Machine Learning behind Rose didn’t have the ability to pick up on subtle cues needed to provide this level of authentic dialogue.

For example, Rose could have texted: “Hi, I’m Rose. I can help you make a reservation, provide a list of tonight’s entertainment or give a tour of the hotel. 

I would have immediately said, “Reservation.” 

Rose would then have asked, “Which of our five world-class restaurants would you like reservations?” 

I’d say, “STK.” 

It could have taken a few minutes. That’s better than the twenty-minute line downstairs in the lobby. But, the experience was no different than calling an automated customer service line. 

Use Empathetic Design to Problem Solve

I understand why Rose’s development team focused on building personality that portrayed the mystique of the Cosmopolitan’s brand. It can be a strong branding strategy that also helps engage an audience that’s accustomed to speaking with human staff. 

I argue there was too much focus on developing personality rather than solving the needs of all hotel guests. Empathy was an afterthought. 

I would have started Rose’s development with exercises focused on adding value to the user’s overall experience. What are we trying to understand? How are we empathetic to a traveler who’s tired and hungry? How do we get Rose to skip all the sassy fluff and get straight to the point, especially if someone is direct?

I argue there was too much focus on developing personality rather than solving the needs of all hotel guests. Empathy was an afterthought.

Personality can always be added and is essential to an engaging experience, but it goes hand in hand with focusing on solving the problem first. It’s the whole logic and emotion balance. 

Start with a user experience (UX) audit. This allows you to get feedback on the challenges your customers face and how a UX strategy will fill those gaps and accommodate their needs. 

Use Technology as an Enabler

Remember that a technology platform alone will not improve your customer experience. You have to plan, strategize, and implement a platform that will execute based on your strategic and empathetic input. This is what creates an authentic and valuable experience. 

Conclusion

There are going to be some people staying at the hotel who will want a straightforward response. They may not want to endure ten flirty text messages to be able to make a dinner reservation. On the other hand, other people will enjoy interacting with Rose for her sassy and suggestive comments, with no particular end goal in mind than to have fun. Rose needed to accommodate both scenarios, just as she was marketed to do.  


About Defakto

Defakto crafts authentic experiences that connect with today’s evolving consumer. We do this by offering an unmatched combination of strategic consulting, design and user experience, technology enablement and data optimization. Headquartered in Dallas, Defakto’s clients include Lexus, Abilene Christian University, Spectrum and iRhythm, as well as Fortune 500 companies in Automotive, Health Care, Technology & Media Industries.

About This Series

How is technology design affecting the way people interact with cities, companies, devices, and each other? This series shares stories and insights about how design is shaping our connections, innovations, and virtually every other aspect of our lives. From artificial intelligence to medical device portals, read the stories and commentary to learn how brands are combining design thinking and technology to create authentic experiences for today’s evolving consumer.

Get on the list.

Sign up to keep your eye on what’s new and next in Dallas-Fort Worth, every day. 
And, you’ll be the first to get the digital edition of our new Dallas Innovates magazine: 
The annual edition publishes in January.

One quick signup, and you’re done.   

Related Posts

Comments are closed.