Siegel Sez

June 14, 2019

Definitely Doug


by: Douglas Rice

Artificial Intelligence and Chatbots:  Are They for Real?
 
We constantly hear the marketing hype for artificial intelligence (AI), often suggesting that a computer can answer most any guest question or request. we see the surveys suggesting that many travelers, especially younger ones, prefer to interact with us via text or voice. But does texting and AI have a place in today’s hotels?
 
Today I will focus on what I will call digital concierges. These are chatbots, infused with a certain amount of artificial intelligence, natural language processing, and in many cases language translation, that have been customized to answer the questions a guest might pose to a real hotel concierge or the person answering the guest service line. I’ve looked at several in the past few weeks, and they’ve come a long way from some of the earlier attempts. Some have started to become quite useful. To be sure, most of them still can’t handle open questions like “what should I do tonight” or a complex room-service order. But they can handle simpler requests and questions, and in doing so they can reduce labor costs for the hotel, improve responsiveness to guests, overcome language barriers, and improve guest satisfaction. Today’s better solutions are hybrid, with a bot that tries to answer the question or request first, but then escalates it efficiently to the best available human staff member when it can’t.
 
Today I will focus specifically on systems that provide native text capabilities from the guest’s personal device, using generic platforms like SMS, WhatsApp, Skype, Facebook Messenger, WeChat, and similar. While a messaging function within a hotel’s mobile app may also be useful (and some platforms support this as well), I take as a given that guests mostly want to use their own device, and we know that the average guest doesn’t have (or want) the hotel’s app. With today’s voice-to-text capabilities in many of the platforms, it is increasingly easy to use them with voice as well, and using the user’s preferred platform means greater accuracy with voice.
 
I’m going to ignore (for today) the many chatbots that are only focused solely on the website/mobile app/reservation process. Most of these are sales-focused and can’t be easily extended to cover the rest of the guest journey. In contrast, the ones I am describing can generally deliver a consistent chat experience from discovery and booking through post-stay.
 
The benefits of digital concierges will depend on the nature of your operation, as well as the capabilities of the solution. But speaking generally, the volume of calls to hotel staff should drop significantly (30% is typical), even as the number of inquiries goes up (because there is less friction to submitting a question or request). This may mean cost savings from fewer staff in a large hotel, or improved service in a smaller hotel where the person tasked to answer the calls is the front desk clerk. If done well, guest satisfaction should also increase. Michael Marino, Chief Experience Officer at Caesars Entertainment, has cited a 10-point increase in satisfaction ratings from the introduction of their digital concierge. The best digital concierges can also increase upsell spend, by using CRM data and even prior conversations to target specific offers in-stay, or by making it easier to order room service or book a spa appointment.
 
Language translation can also be a major benefit to hotels with a polylingual clientele and/or staff. Many platforms will translate from the guest’s language to the bot’s native language and/or to the language preferred by the staff member who handles an escalated request. But as much as Google Translate has improved in recent years, used in isolation it cannot correctly translate many of the requests that guests actually make – there is too much hotel-specific lingo or context-dependent meaning, expressed in different languages in ways that defy direct translation. While most of these platforms incorporate Google Translate, hotel-specific overlays have proven necessary in order to both understand requests properly in different languages, and to translate messages for staff.
 
What are some of the other things you should consider in evaluating digital concierge platforms? You should carefully consider what you are trying to achieve and what use cases are most important to you; depending on these, some considerations will become more or less important.  But some important questions follow.
 
First, does it support all of the chat clients likely to be used by your customers? Can it be embedded in your hotel app (if you have one), or on the TV or guestroom tablet? What about the hotel or brand website?
 
Second, if the bot can’t answer a question and needs to hand it off to a human, can it be directed to the right human based on the question or request? Can the bot detect the tone of frustration that indicates that its response didn’t meet the guest’s expectations, and escalate the query to a human? Can the human view all messages in a single user interface, regardless of the originating text platform? Can they see a transcript of the conversation prior to handoff, preferably in their own language? Is the guest told how promptly they can expect a reply? And if a human doesn’t respond in a reasonable period of time, how Is the request escalated to ensure a timely answer?  I saw one service that even provides a call center for the first line of response to remove even more calls from the hotel’s own queue and make the handoff from bot to human nearly instantaneous.
 
Third, can the digital concierge be integrated with fulfillment systems, such as work-order management or point-of-sale? If so, does it need to clarify requests, such as how many/what type of towels do you need, or how do you want your hamburger cooked? How does it handle these clarifications? And when a request is closed out (e.g. in the work-order management system), can the bot advise the guest, such as “Your towels should have arrived, please let us know if you need anything else” (just in case something went wrong)? Integration with the CRM or PMS system can also be useful, especially if you plan to use the digital concierge to push targeted marketing offers. The best digital concierges can blend customer history or preferences with actual conversations they have had with the guest, to suggest the most relevant options. They can also integrate with surveys to capture guest satisfaction in-the-moment.
 
Fourth, what kind of analytics are provided? You should be able to get statistics on the specific types of questions being asked and requests being made. You should see what percentage of queries are addressed without human intervention, and what percentage need to be escalated to a human. You should be able to correlate introduction of the digital concierge with guest satisfaction ratings and folio spend. You should be able to learn what questions aren’t being answered adequately, so you or your vendor can refine the response algorithms.
 
Fifth, can the digital concierge support the entire guest journey, from booking to pre-arrival to in-house to post departure? And if you are operating a brand, will it scale appropriately, allowing brand consistency while still enabling the hotels to manage their own specific information? Can it differentiate between brand-level questions vs. hotel-specific questions, such as might occur during a booking process on the brand web site?
 
Sixth, how much setup and maintenance does the hotel have to do? Do they have the right staff to do this, or do they need a digital marketing agency involved to get the correct branding? The best bots can be designed with specific personalities, and you might want yours to have one that’s aligned with your brand. Who customizes the specific responses? (Hint: AI doesn’t do this!) If it’s the hotel, is the process sufficiently easy that you can support it over time? If the vendor, will the costs be manageable?
 
Seventh, think carefully about voice activation. Some of the platforms support voice input, but they generally support voice response only for limited queries, if at all. If voice input is supported, it’s important to evaluate the handoff of questions that can’t be answered by voice. Most of today’s platforms rely on the voice-to-text capabilities of messaging platforms for input, but respond via text on the same device. This approach may not work, however, with an in-room Amazon Echo or similar device that lacks a display; most implementations where these can actually respond verbally are limited to canned input phrases rather than natural language.
 
Overall, I was impressed with the amount of learning and sophistication that some of the digital concierge platform vendors have incorporated into their products; they have come a long way in a short few years. They are still many years away from eliminating the need for humans (if they ever do), but they can reduce labor costs and/or a deliver a better guest experience for many hotels.
 
A few of the more interesting companies I have looked at (with the usual disclaimer that these are not recommendations, just interesting places to explore) include Go Moment, Hotelway, Reply.ai, and SABA Hospitality. If you haven’t looked at these technologies recently, you may find a tour to be eye-opening.
 
Next week is HITEC. If you’re there, and see a technology you think I should cover in a future article, please let me know! I’ll spend a lot of time on the floor, but there’s never time to see it all!
 
 
Douglas Rice
douglas.rice@hosptech.net
twitter: @dougrice





want to read more articles like this?

want to read more articles like this?

Sign up to receive our twice-a-month Watercooler and Siegel Sez Newsletters and never miss another article or news story.