Chatbot Increases Hotels’ Room Service Sales 10-50% and Gets Wooed By Guests
In order to do anything for guests, the Edwardian team needs easy access to data from multiple systems. With Oracle as its PMS provider, Mrini and his team leverage the company’s open APIs to access and consolidate data from across platforms. The PMS gathers the initial guest contact info and how stays were booked. PMS data combines with guest information from the CRS, guest surveys and information from the F&B outlets to consolidate in one central database.
“The ability for two-way communications to pull and push data through systems made it possible to do what we wanted to do,” Mrini says.
Edward started as a mobile app for housekeepers that enabled them to see the queue of rooms that needed to be cleaned and track progress with live information. AI also offered added data for housekeepers to know about guest behaviors, such as if a guest always requested extra shampoo or liked a particular type of chocolate that a housekeeper could then leave in the room as a surprise and delight element.
The mobile app also connected to the HR system so the AI in Edward can allocate tasks. It knows who is on duty and who is where with geolocation. If a particular associate is too far away to deliver towels to a room in an efficient manner, Edward will assign the task to someone closer.
3 Pillars of Functionality
Service to Conference Organizers: Event managers can text Edward for any services – from needing to delay lunch to help with a projector and Edward will immediately deploy an alert to the proper parties. Event managers can also ask Edward for help with locations and times by accessing the event schedule.
Service to Employees: Common questions that go to HR can now be addressed by Edward. From shift inquiries to vacation balances and how to apply for paternity leave. “There are more than 90 topics Edward handles for employees,” Mrini says. “The most popular, however, is ‘what’s for lunch in the cantina?’”
Service to Guests: In addition to the aforementioned services Edward can provide for guests, he also can help book a table at a restaurant. More importantly – and more lucratively for Edwardian Hotels – guests can order room service via Edward. The Edward room service order is between 10% and 50% higher than phone orders.
With the success of the system for housekeeping, it was decided to create a mobile app for every department and then for guests as well.
“We started with online check-in and developed an app so guests could check-in online and select a specific room from a floorplan like the airlines – all in real-time,” Mrini says. “Every interaction with the app generates a live interaction with Oracle Opera booking platform via open APIs, so no employee involvement is required. Guests are accessing the system, selecting their room, making payment and are checked-in. Guests loved it.”
This functionality was delivered three days prior to arrival via SMS text with a link. Guests could then click the link to check-in. But something unexpected started to happen. Guests would respond to that original text and ask questions or for service. Since Mrini had done the coding for the text, his cell phone was associated with the number used to deliver the text message.
“I personally started receiving messages 24 hours a day from guests asking for anything from towels to spaghetti Bolognese,” Mrini chuckles. “Being in hospitality for 22 years, I couldn’t ignore them, so I was picking up and calling the hotels to report the requests. It got to the point where I was sleeping with my phone on my pillow. My wife kicked me out of our bedroom.”
To save further sleep interruptions – and potentially his marriage – Mrini decided to create an intelligent system to do what he had been doing – reading messages, understanding them and sending a reply. Thus Edward was born.
Edward uses both AI and natural language processing for understanding what a specific request is. Then the Oracle open API will sync the mobile number from which the request was received to booking system information to find out who the request is from. The AI comes in to mine the database and see what else is known about the specific person and it all happens in a split second.
“One of the most popular questions Edward gets is, ‘Is breakfast included in my booking,’” Mrini explains. “Edward will check, and if it’s not included he will respond with the cost, but then offer a promo and to add to the reservation for the guest. Edward can do this much faster than a human needing to look up a reservation.”
Mrini explains that the benefit of Edward is that he takes tasks that can be done by a machine. What a machine, Mrini notes, cannot do is face-to-face interaction that hosts should be able to do more of, rather than being behind a desktop.
Edward was born in May 2016 and guests quickly and seamlessly adopted engaging with the bot.
“It was immediate adoption because this was a signal our guests were sending that they wanted,” Mrini says. “They didn’t expect an automated system to be replying to them and to date, most guests don’t know, even though he introduces himself as a virtual host. They go on talking to him like he’s a person.”
This includes the confused guest who perhaps would have felt slightly jilted when Edward didn’t continue conversing with her after her stay had she not been informed that her attentive staff member was in fact a bot. She didn’t harbor too much ill will as she told the staff to keep the envelope with Edward’s gift and “buy him cables or whatever he needs.”
- An Edward Timeline of Efficiency & Effectiveness
Edward has helped 28,525 guests from 99 countries in 59 languages. Over the past three years they have seen some remarkable results:
2017 – chatbot handled 49.4% of all requests at an average of 2 minutes per call and 8 hours in a working day (this equates to 95 working days)
2018 – chatbot handled 58.3% of all requests at an average of 2 minutes per call and 8 hours in a working day, (this equates to 139 working days)
2019 (to-date) – chatbot handled 69% of all requests at an average of 2 minutes per call and 8 hours in a working day, (this equates to 84 working days)