Much has been said about the role that AI and data can play in customer experience. Much less has been discussed about the implications behind those changes. Consider this scenario: An application aimed at one customer at a time, providing an increased level of sophistication and satisfaction with each interaction, constantly improving experiences as it learns from the previous one. Each level calls for increased support in communications, data, and AI. Let’s explore those levels and how they evolve.

Checking into a hotel, even with "priority" status, can be tedious and often result in nothing more than validating your identity and method of payment and being given a room. 

Instead, leading providers are exploring what's possible with CX when you know what customers want, combine data from multiple sources, use instantaneous inferencing, reach them on multiple channels, and generally make life more pleasant for them—without being creepy or intrusive.

For example, a guest walks into the lobby and a beam picks up her phone and checks her in, puts her room key on the phone, and displays a map to the elevator and the room. A recommendation engine for this straightforward set of experiences is simple and easily scalable to hundreds of simultaneous check-ins across a hotel chain. 

This experience relies on data that the hotel chain already has available, and works jointly with the many transactional systems: room reservations, food and beverage, pay per view, and others to determine the mix of offers and appropriate offers given the customer profile, loyalty program, and stated preferences for privacy and connectivity. This would work on mobile, via native apps and browsers, and connect to any other systems or IoT solution to sense location and needs. 

Now let’s level up. The application sets the room temperature she likes, the TV news channel she watches, adjusts lighting and shade, and pops up room service if she usually uses it. 

"There is a vegan restaurant 0.3 miles away, and they deliver. Would you like to order?” 

-cue a pop-up on the flatscreen TV displaying the menu-

The data requirements here become more varied and distributed: streaming video services, streaming music services, smart thermostats, food delivery apps, expense reporting apps, and more. 

Communication requirements become more complex as more point-to-point channels are involved. Beyond checking in and in-room convenience, the application can be more intelligent and be able to branch to different interactions. Did the guest arrive late in the evening and will stay only one night? In that case, making suggestions for a horseback ride or a spa appointment the following afternoon is annoying. If the guest is staying a few days and is part of a group or conference, text analysis of the conference agenda can make several useful suggestions:

• The opening session is in the Grand Ballroom. You will need fourteen minutes to walk there. Here is a map.
• There is free time between 4 and 6:30 p.m. before cocktails. Would you like to take a guided trail hike for 90 minutes? (And note: Do not suggest a trail hike in Scottsdale in July).

Advancing the language capabilities in the communication yields more precise, granular interactions focused on the specifics the customer needs to know.

• "I'm not interested in a 90-minute hike, is there a 60-minute one?”
• "Has my CEO checked in yet, and is she planning on an activity tomorrow?"
• "How many people have arrived for the conference, and how many of them are analysts?"
• What kind of analysts?
• "Oh, sorry, I meant industry analysts."

Data requirements for these more intelligent and personalized interactions are even more vast and complex: weather, Nielsen, travel sites, tourist guides, travel bloggers and influencers, permissions about arrivals, and more.

As you can see, many complications and implications follow the scenarios and escalate as the personalization and complexity of transactions increases. This is in line with all research on CX that states the complexity increases the deeper the connection with the customer grows. Creating a satisfying customer experience requires data from many sources, provisioned in near-real-time, but there is more to consider. The data sources are not typically tailored for this application. It takes intelligence to provision the data and integrate it for the application. To create accurate recommendations for mappings of data sources, and related objects, one must leverage methodologies including machine learning, natural language queries, intelligent search, and graph-based relationship patterns.

if the application has to query data from multiple sources, many of which are not under the control of the application, a useful technique is to create a knowledge graph of the data sources to guide queries quickly to sources without moving data around the network. A knowledge graph powerfully distills complexity to something understandable. Connections between nodes form a directed graph, which is the bedrock for background processing. A suite of AI tools employed for constructing and using the knowledge graph is: 

• Recurrent Convolutional Neural Networks
• Semi-structured data parsing: Hidden Markov Model and gene sequencing algorithms

If these terms seem esoteric, they are, but it takes complexity to make things simple. There is so much more that can be done with the customer experience using communications, data, and AI… but integrating the pieces is not simple. The controlling point of customer satisfaction is reaching people when and where it matters, how they want to be reached, and delivering to their expectations.  Not wasting their time with trivial and irrelevant interruptions, and enhancing their experience with useful and timely resources, is vital for success in CX and engagement.