Guest Smarts

23 January 2020



Assessment of the potential impact AI will have in the hospitality industry has mainly been confined to the back office, with comparatively little attention on how it can improve the guest experience. Greg Noone talks to Chris Silcock, chief commercial officer at Hilton, about how the operator is using machine-learning algorithms to become more responsive to its customers before, during and after their stay.


Chris Silcock is explaining how Hilton has used AI to rationalise the operator’s back office processes. To do this, however, Hilton’s chief commercial officer needs to take a conversational detour. He needs to talk about whiteboards. “I started in revenue management,” explains Silcock. Back then, his department could rely on relatively few computers. As such, Silcock and his colleagues were forced to rely on a combination of mathematical nous and the expedient method of showing their working to one another. “We were using primitive data to write selling strategies on a board in the back of the hotel, to determine what we should sell rooms for, for how many times, [and] at what price.” Gradually, as the revenue management department graduated from using marker pens to Excel spreadsheets, these forecasts grew sharper. The adoption of machine-learning algorithms in the past five years, meanwhile, has allowed Hilton to hone these predictions even further – which is essential given the great volume of data the operator handles, originating from online bookings and interactions with its mobile app.

“We have started to see what can be possible using data in a much more rigorous way throughout our platforms,” says Silcock. Nevertheless, “we’re still at the beginning of that journey”.

This is usually where the conversation ends when it comes to the role of AI in hospitality. Machine-learning algorithms function at their best, after all, when they divine simple patterns from huge volumes of existing data and perform a function informed by what they have learned. Plotting trend lines on a graph is one of the simplest ways that AI can prove its mettle. Making decisions based off patterns learned from human behaviour or speech, meanwhile, is significantly harder.

Even so, Hilton has persisted in weaving machinelearning algorithms that perform variations on these functions into its core service offering, using AI to do everything from aggregating customer comments poststay, to dealing with basic enquiries and taking the initiative to inform hotel managers about the wants and needs of frequent patrons. “Bringing it to life, that’s quite a difficult feat,” says Silcock, to which end the operator has hired over 150 data specialists and built a cloud-based platform capable of aggregating up to a billion data points a day. Hilton’s chief commercial officer is quietly confident that it is up to the job.

“We have a scaled platform, and we have an ability to feed data in and out of our properties, and to our team members, in order that they can use it effectively to serve customers,” says Silcock.

AI on the horizon

There was never a light bulb moment for Silcock when it came to the impact AI could have on hospitality. Rather it was a slow burn, as the Hilton veteran slowly realised how machine-learning algorithms could further rationalise existing lines of communications with guests.

One of these areas was in customer support. “We built a natural language processing capability that helps us out in several ways to support the customer experience,” says Silcock. Hilton quickly found that it could use this engine to solve up to 20% of the most basic enquiries made by guests. “If that engine cannot solve the request, then it will be escalated to one of our customer teams.”

One particular concern was highlighted that, were it not for AI, might have been lost in the noise of the thousands of questions put to Hilton every year. “Our honours members get an upgrade if space is available,” explains Silcock. Some of these prized guests were concerned that using Hilton’s digital app as a room key would stop them being eligible for that better room. Flagged by the AI, “we were able to listen, learn and then address that [issue] through communications to our customers”.

Another way Hilton has harnessed AI is sifting through thousands of online reviews and comments – not traditionally a place where companies might expect succour. “We can now ingest over 20 million comments per day,” explains Silcock. “This engine can pass through that data, understand the context, and start to throw out things that we should pay attention to, both positive and [in] opportunities where customers are telling us [about areas] where we need to focus.”

This can function on a large scale – like feedback received through more traditional market research – or help in identifying comparatively minor issues. The engine, says Silcock, can “identify to us a specific hotel that may be doing something exceptionally good that customers are starting to comment on that we should pay attention to, that we should drill into and work out how we can spread that best practice across our estate”.

Hilton is also starting to use it to personalise its communications with customers, and connect all their interactions with the operator across multiple channels. Looking forward, Silcock envisions going even further – using AI to leverage the thousands of data points generated by guests to inform hotels on not only the room they are likely to reserve, but the specific features inside it.

“We built a natural language processing capability that helps us out in several ways to support the customer experience.”

“At the simplest level, that could be if it has a desk, or where it’s located in the hotel, or if it has a bath or a shower, to the size of the TV,” he clarifies. “Then, rather than a customer choosing a generic room type, we will start to be able to connect those specific features and benefits with the preferences of the customer.”

Close connections

A path forward for this strategy was recently revealed with the introduction of Hilton’s ‘connected room’ concept, wherein all aspects of the in-room environment – including choice of entertainment, lighting levels and temperature – can be tweaked.

What evidence is there, though, that guests desire this level of customisability in their rooms? According to Silcock, the reasoning behind this goal is not only sound – it’s an extension of the process by which hotel staff have always been getting to know regular guests in person. “If you travel frequently to the same hotel, you will learn the room that you will decide is your favourite room,” he says. “That happens all the time, all over our estate.”

For the moment, this is primarily achieved through guest-staff relationships that begin at the front desk. “How I see us starting to use machinelearning in that respect is, as we start to inventory all those granular features of our rooms, and customers start to tell us their preferred room – which they do already – we will be able to use [AI] to connect the preferences in previous hotels the customer stays in, and seek out what would be the most likely accommodation the customer would prefer in a way that, manually, we would be unable to do,” says Silcock. “That’s all to be tested and proved, but our history in hospitality and working directly with customers certainly suggests that would be of huge value to our customers.”

Conversely, Hilton’s chief commercial officer is reluctant to introduce more conspicuous manifestations of AI, like voice assistants. “Our research, so far, has certainly shown that guests consider their hotel room to be a private space,” says Silcock. “Although we have conducted limited trials in some rooms with digital assistants, we don’t yet see widespread guest acceptance of that being placed in their hotel bedroom.”

This approach is emblematic of a broader philosophy that AI is a technological tool, and not a unique selling point for Hilton in its own right; guests, in short, shouldn’t need to glimpse how a tangle of data points and mathematical equations succeeded in divining which room they’d like to stay in. Rather, machine-learning algorithms should operate as any valued member of hotel staff should: discretely, efficiently and with a preternatural sense of what is required by the customer. 


Connected Room

Hilton began beta testing Connected Room in 2017 and it will evolve throughout its roll-out. The operator can push new features, apps and functionality to existing devices over time without the need to deploy additional hardware. For guests, this means that in the longer term they will be able to do things such as upload photos and art to display digitally in the room, or connect to new devices and use voice commands to control their room’s features or to access their content.

Hilton cites a benefit to franchise-owners too, who sometimes face challenges implementing new technology due to high costs and operational challenges. Connected Room should make it easier for them to keep pace with the rapid change in technology and easily integrate innovations into the guest experience.

The Hilton Honors app is the only way to access Connected Room controls and other technology conveniences from one’s mobile device. For example, Hilton Honors members are able to use the app to check in digitally before arrival, select their room from a digital map and open their hotel room’s door with their mobile phone thanks to the app’s digital key function.

Source: Hilton 

Facade of the Hilton Pilar in Buenos Aires, Argentina.
Chris Silcock, chief commercial officer, Hilton
Lobby of the Hilton in Guadalajara, Mexico.
The Hilton Connected Room is the first step in how AI can impact the guest experience for the better.


Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.