It’s fair to say many are curious about the latest developments in artificial intelligence (AI) regarding large language models (LLMs) such as ChatGPT. Who hasn’t asked ChatGPT to formulate an email, sales pitch, or social media message, or just had fun asking it to write something in the tone of voice of Mickey Mouse?
What’s important to contemplate is the role natural language can and will have on enterprise-level technology solutions, and how they can shape the hotel revenue strategy of the future.
It’s exciting to consider a more conversational-based way to interrogate the data and figure out actionable insights. It’s more human, and it’s more scientific.
One of the biggest challenges with the customer set that hotels work with is access to data. The interrogation of that data, the insights that come from the interrogation, and the skills and method of running that flow are all specialized items right now. And because of that, there’s a lot of technical complexity and a lot of labor complexity that’s involved.
Boiling Down the Data
The challenge the industry faces is boiling down the data to get to the core business problems and decisions underneath the questions.
Here’s an example:
As a hotelier, would you prefer a booking today of $150 or a 50 percent chance the room remains empty and a 50 percent chance that it sells for $300?
Both have the same expected value. The math would say that you should be indifferent between them. However, nobody is indifferent between them. There’s a point where your strategy, your operating landscape, your owner expectations, your faith in the error bands of the system, and your viewpoint as a business leader come in to set that dial.
How can AI measure your risk tolerance?
The notion of setting the dial and the rules of engagement aren’t going to change. However, once you set those rules, you can use tooling that allows the automation to calculate the downstream implications of that. Scenario planning and understanding are going to move up the hierarchy thanks to AI.
The current landscape makes it challenging for a hotel owner or revenue leader to interrogate the details of the business and understand what’s going on. What will evolve is a deeper understanding of that throughout the organization, thanks to this enhanced accessibility to the data. This will help everyone focus on what the computers won’t be able to articulate around, like risk tolerance, debt covenant coverage, or proprietary items that the systems won’t know about at any point in the business evolution.
Putting Value Before Perception
Often, asset managers and hotel owners have a perception of what’s the value of their assets on a per-night basis, and what the market will pay. And sometimes, those perceptions are vastly wrong, and often, they’re wrong on the high side.
For example, an owner or asset manager may think that there’s no way anyone should stay at their property for less than $1,000 a night. And they’re putting these constraints on their revenue teams.
There’s a lot of information asymmetry. They don’t know why that expectation isn’t resulting in the business performance they want. And the easy assumption is the hotel team is just not executing the right strategy.
However, if they were to dive into the data, it might be telling them that they need to renovate. The reason they’re not getting $1,000 a night is not front desk service, for which they’re getting great guest review scores and comments, but is ultimately the quality of their guestrooms, which are performing less well in the reviews.
All that text-based data has value. You’re capturing signals of customer satisfaction and intent. These analytics models can help you build business models, such as an ROI for a renovation, tied to these signals.
Understanding the Longer-Term Impacts of Taylor Swift
This relationship between quality and price exists in the RMS today. We’ve all read about the amazing impact the recent Taylor Swift tour has had on room rates. But how might that play out for your property in the future?
On a high-demand day, you can have two-star properties charging five-star prices. It’s a quick win. But what are the longer-term impacts?
What happens is the review scores that come from that $500 a night at a two-star property are substantially more negative because they don’t align with the customer’s perception of what $500 should get them. They’re rage reviewing. However, two years later, the person reading that review won’t know that person paid $500 a night for the weekend Taylor Swift was in town.
The LLM models start to stitch these data points together, and as the vendors build the data sets that enable the insights to surface, the value unlock will happen. The longer-term implications of today’s decisions will become clearer.
As these systems evolve, a lot of that information asymmetry will go away. The system—the interaction of the data, the technology, and the people—will become more rational. There’s a lot of emotion behind decision-making today, and these advancements in AI will bring more facts into the conversation.