About three years ago, Oracle Hospitality acquired Nor1 to deliver an AI (artificial intelligence) solution for real-time pricing and inventory management—specifically to create ancillary offers to guests—and embedded that tool into Oracle’s property management system, OPERA Cloud. Despite the availability of sophisticated AI tools, Jason Bryant, vice president of Nor1, considers the hotel industry to be in the early stages of adopting them, but expects the usage to ramp up exponentially in the next few years.
In the following interview with LODGING, Bryant focused on the use of AI in guest personalization and communication, sharing his view that the best AI products in hospitality leverage machine learning to yield demonstrable business value. While AI in general is simply the replication of intelligent human behavior by some technique, machine learning is a type of AI that uses statistical techniques to enable the machine to progressively improve its decision-making.
How would you describe AI’s overall application to the hotel industry?
It can play a role anywhere that human decision-making is happening, especially tactical decisions. For example, what do we offer this guest, and at what price? AI can generate those offers. We associate hospitality with the guest journey, from acquiring a guest to how they book online to their pre-arrival experience. And once they’ve checked in, the guest journey includes the communication systems that are made available and how they get questions answered. If AI is not already touching all parts of that journey, it certainly will touch all of it. By and large, we’re trying to empower the humans on the property by automating things that are tedious, and thereby free staff to do things like strategize and directly engage guests for service delivery.
“AI” has become a buzzword that is often used in marketing software products. What do you consider genuine AI?
There is not an official definition, and if you talk to professors in data science programs, they are all going to have slightly different definitions. So, there is a lot of noise around the term “AI,” and vendors and different stakeholders use it in different ways. For us, it’s not just AI, it’s machine learning, which is a specific type of AI. One of its hallmarks is that it makes decisions and collects data in real time. So, let’s say you are a guest booking at a Fairmont hotel, and you’ve already got your dates and your credit cards getting processed. They pass this information into Nor1, and it decides what ancillary offer to present and at what price, in a matter of milliseconds. Then, whether you select the offer or decline it, that data point is fed back in to fortify the intelligence system. That’s the machine learning part, and it helps our models get stronger over time.
What would you advise hoteliers looking to get started in exploring AI products?
For any product, what matters is creating value. That would be my biggest message to any business, but certainly to hoteliers. Does it create more revenue? Does it create greater guest satisfaction? And can you measure that value? For example, for every single transaction, Nor1 is measuring how much guests like the offer by how much they’re willing to pay. That’s what’s important, showing the value.
When should hotel staff in lieu of an AI system communicate with a guest?
The hotelier needs to understand that the more penalties there are for giving the guests a wrong answer, the greater likelihood they will need a human engaged. For example, in an emergency scenario, you very much want a human engaged; you need to ensure that the guest will get taken care of.
In general, it’s not about deciding between human and computer communication to guests. The question is, are there touchpoints for our guests where an AI can possibly be serviceable? So, it’s not a complete replacement for staff interaction, it’s a support.
Do you feel AI communication will be increasingly accepted by guests in the long term?
My opinion is that as long as the AI understands the inquiry of the guest and is able to successfully answer and deliver on that request, over time, people will become more and more used to dealing with AI.
You’ve discussed how machine learning can enable ancillary offers to be tailored to guests. Could you share more detail on how AI supports personalization?
“Personalization” is another one of those terms that people use loosely. What we’ve done at Oracle Hospitality is to be super clear on the definition of it. Our definition is specifically related to AI. In the very beginning, I might not know anything about you as a guest; for instance, you’re not part of a loyalty program. The AI will use what information it has and start to make intelligent offers, but they may be solely contextual. The context would just be that you booked on a Friday for two days, for instance. Now let’s say I’ve engaged you and made you an offer. Even if you said no to the offer, what that helps me understand is that you do not want this upsell or this upgrade for $32, for instance. It’s important for the AI to know that the next time we engage you. The next step would be segmentation, whether you are booking for business or leisure, whether you are a loyalty guest, etc. That’s still not what we consider personalization, but it’s an additional step. Then we start getting into micro-segmentation of guest types.
Humans can’t do this level of segmenting: we’re looking at hundreds of micro-segments and correlating all this data. That’s what we’re able to do with machine learning. The final step is what we call true personalization, where the AI is providing an intelligent offer based on all that behavioral data to the front desk agent, who then presents the offer to the guest.
Do you foresee AI becoming an integral part of all property management systems?
I would call it table stakes; to sell a PMS, you will need to have these different levels of AI built in. It will be commoditized, and everybody will have access to it. They’ll expect it to be part of these core software products. And I cannot imagine that there is any CEO overseeing a PMS company right now who doesn’t have people already trying to figure that out. We need to be a little bit patient because to go from the prospect of what this technology can do to actually embedding it, and then leveraging and applying it, will take some time. But I would say that even in 2024, you’ll start to hear every PMS company talking about machine learning and Generative AI. And honestly, if they don’t, they will die. There will be no way for any PMS company that is not leveraging AI to keep up with companies that are.
