It's All About The Experience


A blog about managing and improving customer experience and improving profits.

AI In CX Improvement? Five Great Examples
AI In CX Improvement? Five Great Examples

As a lifelong tech enthusiast and internet start-up cofounder, I have embraced technology solutions in my daily life and business career. I have been here for it, from using primitive neural networking software in financial planning to deploying robot-cleaning devices at home and asking Google Max to use its collaborative filtering capabilities to play the music that Google thinks my dinner guests would like. 

However, as a professional with a lifetime of work in the brass-tacks of customer experience (CX) management, I was skeptical about artificial intelligence's effectiveness in addressing problems in the customer realm. 

My recent research on AI applications in customer experience improvement has been enlightening and, to be honest, sobering! Below are five powerful examples of companies using AI to revolutionize customer experience management at scale. Several examples are likely disruptive and industry-changing, but all are worth understanding. Let's dig in!

Disney's Magic RFID Data Creates Magical Experiences

Disney's RFID Magic Band technology provides my first example. The MagicBand is a bracelet Disney gives customers at its theme parks. The technology "magically" provides users with a room key, theme park ticket, and credit card all in one device, allowing guests to enter parks, unlock hotel room doors, pay for purchases, check in for rides, and sync photos to their account with a simple tap of the wrist (Hollander, 2022). The technology eliminates the need for guests to carry multiple items and reduces waiting times at ticket booths. Additionally, MagicBand enables staff to provide personalized experiences by quickly identifying guests and anticipating their needs based on data collected from the bands (Hollander, 2022). The appeal of this technology for frictionless purchasing and upsells is intuitively obvious on the revenue side. 

But there is more. There is the data. Disney uses the information from the MagicBand to optimize customer experience. It collects this information with RFID sensors installed throughout its parks and resorts. The data allows Disney to connect each tap of the MagicBand to a specific guest's profile, enabling data scientists to analyze customer behavior. Predictive analytics allow Disney to open and close attractions to maximize customer satisfaction, optimize staffing, and better manage restaurants, attractions, and park queues. (For more on Predictive Analytics and its applications, please see my previous blog post on Big Data.)

In terms of personalization, these bands help curate a visit personal to the last detail. The technology can orchestrate a timely appearance of one's favorite Disney character or a surprise birthday message during fireworks. The MagicBand technology has been a significant investment for Disney, with CEO Bob Iger approving approximately a billion dollars for its implementation (Sweeney, 2013). Based on what I have learned, it was a worthy investment.

Uber's COTA - Improving the Handling of Customer Tickets With AI

Imagine the customer service challenge of handling hundreds of thousands of customer messages in thousands of cities worldwide every day. Uber's Natural Language Processing (NLP) System, COTA, enhances customer experiences. COTA, short for Customer Obsession Ticket Assistant, uses machine learning and NLP models to assist agents in providing better user support. NLP breaks words and phrases into components the computer can understand, then buckets them into response categories. COTA empowers agents to resolve issues more accurately and swiftly, reducing ticket handling time by about 10% (COTA: Improving Uber Customer Care With NLP & Machine Learning | Uber Blog, 2018).

COTA's success led to the development of COTA version 2, which integrated deep learning models into Uber's machine learning platform. Uber eponymously named the deep learning system Michelangelo. This second-generation system significantly outperformed its predecessor regarding accuracy, ticket handling time, and customer satisfaction. Additionally, COTA v2 introduced a model management pipeline to ensure continuous model updates so that Uber's Data Science team could continuously improve the model.  

The second version also focused on enhancing the extensibility of future NLP models within Uber. According to Uber, adopting deep learning in COTA v2 significantly improved the breadth of support resolution solutions for Uber's diverse services and simplified service delivery across multiple languages and regions, thus making the system much more adaptable to future use cases as Uber's business grows. (Scaling Uber's Customer Support Ticket Assistant (COTA) System With Deep Learning | Uber Blog, 2018).

In addition to CODA, he has been working with Converseon, a company focused on the AI specialty of sentiment analysis. Customer sentiment analysis involves identifying and categorizing the text's positive, negative, and neutral emotions. It assists brands in learning what consumers are saying about them in online testimonials and posts. However, Customer sentiment analysis has an added benefit in offering marketers qualitative and quantitative data that goes beyond the "what" to clarify the "why." Uber used sentiment analysis to help manage its leadership transition and address customer satisfaction in 2017 (Dupre, 2017). More recently, Uber has used the technique to analyze and apply changes to Uber's rideshare maps. It uses the technique to identify customer dissatisfaction with its routing and identify and make needed changes based on customer feedback (Uber Engineering, 2018b).

Robot Placing Customer Rating 5-Star Ranking

Airbnb Solving Big Platform Problems During Hypergrowth

Airbnb faces the challenge of managing a vast global platform of users and hosts on its platform and developing scaleable ways of creating optimal experiences for both. Airbnb has used Artificial Intelligence and Machine Learning in a variety of ways. To improve conversions on listings, the company used image classification computer models to label millions of listing photos accurately, improving the display of beautiful assets in listings (Standley, 2023). This approach allowed for more personalized experiences by enhancing user search, discovery, and personalization through machine learning ranking models based on user engagement data like clicks and booking rates.

To solve a platform-wide problem using booked properties as party houses, Airbnb intensified its background check procedures by implementing advanced artificial intelligence. This initiative scrutinized wide-ranging online public data — including social media, blogs, and search results — to detect signals of potentially harmful behavior by users seeking accommodations. This patented process aimed to assess the authenticity and dependability of guests by contrasting online profiles with those processed by AI, thereby eliminating discrepancies and highlighting genuine identities (Owen, 2021).

The company also harnesses an algorithm to evaluate activities across external online platforms, formulating a "trustworthiness score" for each guest. In concert with Trooly, a firm acquired by Airbnb in 2017, the breadth of data assessed encompasses online posts and comments and extends to database entries and professional memberships. Reservations flagged as high-risk based on this metric are then treated to a meticulous manual review by Airbnb's team, underscoring their commitment to safeguarding the community (Owen, 2021).  

With these enhanced measures and investments, Airbnb has invested in a secure environment for its global user base. It affirms that it will execute background verifications for guests from the United States and India within ten days before their stay (Owen, 2021).

Airbnb has utilized machine learning in various ways to combat fraudulent activities on its platform. One approach involves using network analysis to identify potential scam listings. By analyzing data on Airbnb listings, including reviews and host connections, researchers have uncovered networks of connected hosts with fake reviews, indicating potential fraudulent behavior (McAleenan, 2022).

Airbnb also employs machine learning techniques to fight financial fraud. They use "targeted friction" to prevent fraudsters from using stolen credit cards on the platform while minimizing negative impacts on genuine users. (Friction is ideally something that blocks a fraudster, yet it is easy for a good user to satisfy.) This proactive approach involves leveraging machine learning, experimentation, and analytics to identify and block fraudsters effectively (Fighting Financial Fraud With Machine Learning at Airbnb, 2018).

Airbnb employs image recognition technology to process and understand visual evidence through user photographs. When combined with customer comments, this AI guides the complaint-handling process, ensuring it takes the right actions to address the issue swiftly and accurately. Airbnb's AI has been trained to identify a cleanliness issue and analyze its severity. The automation and level of precision ensure that complaints are categorized relatively and consistently, something only possible with AI (TensorFlow, 2019).

MacDonalds - Making Fast Food Dynamic

McDonald's Restaurants has harnessed predictive analytics and artificial intelligence to enhance their menu options and drive increased sales. MacDonalds is leveraging technology from Dynamic Yield, a Tel Aviv-based firm specializing in personalization and decision logic. Using this technology, McDonald's has begun to offer personalized suggestions on digital displays based on customer preferences and external factors such as time of day, weather, and previous menu choices. A snowy day might trigger the placement of hot chocolate as a menu option, while a hot day might cause the system to suggest lemonade (Kamal, 2023). 

The technology considers factors like time of day, current order selections, restaurant traffic, the popularity of items, and weather conditions and will update digital restaurant and drive-thru menu displays (Owen, 2022). MacDonald has tested the technology with kiosks, websites, drive-throughs, and restaurant menu displays. 

To optimize its supply chain order, McDonald's also plans to connect the AI and menu data recommendations with its supply chain network, improving the management of stock levels (McDonald's Is Using AI and Data to Optimize Its Supply Chain, 2023). By linking predictive customer demand with stock levels at individual stores, McDonald's hopes to optimize its inventory management and reduce waste throughout the entire chain. 

Dynamic Pricing at Wendy's

A use case I am less sure about is Wendy's plan to use AI to develop dynamic pricing for its restaurants. Wendy's is using AI to develop dynamic pricing for its restaurants. The chain hopes to adjust prices based on demand and other factors like time of day and traffic levels. This strategy involves fluctuating prices similar to surge pricing used by ride-sharing apps like Uber or Airbnb. The fast-food chain plans to follow MacDonald's lead and invest in digital menu boards to enable these changes, with a similar focus on enhancing customer experience and increasing sales through suggestive selling and menu adjustments (Yeo, 2024)

CEO Kirk Tanner mentioned that Wendy's will trial dynamic pricing, potentially raising prices during busy times and lowering them during quieter periods. This move aligns with Wendy's broader investment in technology to enhance its digital business, including implementing AI-enabled menu changes and suggestive selling tests based on factors like weather. While dynamic pricing is typical in various industries, its application in fast food is relatively untested. For the future, Wendy's aims to modernize its operations by leveraging technology such as AI chatbots for drive-through orders. (Rogelberg, 2024).

How Can AI Help Your Business?

So, how can you determine whether AI can help your business? As we have seen from the examples above, Artificial intelligence and Machine Learning (ML) can significantly benefit businesses that use them. AI models are good at completing repetitive tasks quickly and efficiently at scale but must be appropriately trained. They are rarely infallible. 

Here are some ways of thinking about how these tools can apply to your business. Applications of AI/ML include:

Driving Revenue Growth: AI can uncover new opportunities for products and services. As we saw with several examples above, AI tools can identify pricing, pairing, and upsell opportunities, improve service speed, and, through Natural Language Processing, provide customer insights for informed decision-making (Martech, 2022).

Boosting Efficiency: AI can automate processes, optimize supply chain operations, and, as we saw with Uber's example, help in decision-making by analyzing data faster than humans (COTA: Improving Uber Customer Care With NLP & Machine Learning | Uber Blog, 2018). AI Automation can improve customer experience by providing faster decisions in consumer finance, insurance underwriting, and claims processing, for example, by replacing processes that took weeks or months with those that take seconds. Faster decisions mean higher transactional conversion rates, raising revenue (Finance Magnate Contributors, 2023).

Improving Customer Service: AI tools like chatbots can enhance customer service by providing prompt responses and reducing wait times. A chatbot, kiosk, or online food ordering process always gives you exactly what you requested in your preferred language. (How AI and Machine Learning Can Impact Your Business and Small Businesses, n.d.)

Enhancing Product Quality Assurance: Machine learning can quickly identify patterns in large datasets to ensure product quality and detect fraud or cybersecurity threats (How AI and Machine Learning Can Positively Impact Your Business and Small Businesses, n.d.)

Where to Start?

When considering implementation, I looked back at a Harvard Business Review article by Wharton AI Professor Kartik Hosagnagar and Apoorv Saxena. Saxena is the former head of AI at JP Morgan Chase and cut his teeth as the head of the AI Vertical at Google before that. Hosnegnagar and Saxena's advice on building an AI program in large organizations has aged remarkably well. The two AI leaders suggested a slow simmer rather than a rapid boil to implement AI in a large organization (Hosanagar, 2018). 

Here it is in a nutshell.

1. Start with good, clean data. Start by developing data gathering and management capabilities throughout the organization. Early data projects help organizations gain experience with large-scale data gathering, processing (cleaning), and labeling—skills that companies must have before embarking on more ambitious AI projects. Establishing organizational learning on developing good data hygiene in a new AI platform is far more important than seeing a significant impact in the short run (Gruttadauria, 2024). 

2. Develop a portfolio of short—and long-term projects rather than betting your organization's AI future on a single, massive "moonshot" project (Hosanagar, 2018).

3. Focus on early wins by improving processes, but focus on more than individual touchpoints. Don't use AI to pave the cow paths. Instead, look at the entire process and explore how AI can improve it. Leaders can consolidate early wins by creating value for multiple parts of the organization, then press on to more significant challenges using a portfolio approach (Why Portfolio Management Is Essential for AI Projects, 2024).

4. Don't Bet the Ranch! Remember, your business does not have to invest all its AI project-related money upfront. Managers can control costs by hiring slowly but steadily and using outside suppliers for AI and machine learning infrastructure early (Hosanagar, 2018). As the AI field has developed, outside resources have become increasingly available in various AI-related specialties. Leverage early wins using outsourced resources, then gradually build your in-house capabilities (Roy, 2023)

5. Hit Lots of Singles, Then Look For Home Runs. Your last step will be to swing for the fences by identifying new, game-changing customer experiences only your AI solutions team can create! 


AI, Data & Analytics Network. (2024, February 16). How Disney World collects customer data. AI, Data & Analytics Network.

Airbnb uses artificial intelligence to transform its business. (2023, August 28). BDO.

COTA: Improving Uber Customer Care with NLP & Machine Learning | Uber Blog. (2018, January 3). Uber Blog.

Cruz, M. D. (2022, January 6). Analyzing Disneyland Reviews with NLP - Towards Data Science. Medium.

David Koenig, Associated Press. (2023, September 20). Airbnb says it has removed 59,000 fake listings from the platform in effort to crack down fraudsters. NBC 5 Dallas-Fort Worth.

Dupre, E. (2017, July 10). Can sentiment and emotional analysis help Uber steer its brand around? - DMNEWS. DMNews.

Editor. (2023, May 5). AI in Short-Term Rentals: How Machine Learning Shapes STR. AltexSoft.

Fighting Financial Fraud with Machine Learning at Airbnb. (2018, March 20). InfoQ.

Finance Magnate Contributors. (2023, June 13). The role of AI in insurance: From underwriting to claims processing. Financial and Business News | Finance Magnates.

Gruttadauria, B. (2024, February 9). Unlocking the Power of AI with an Effective Data Hygiene Strategy.

Hollander, J. (2022, October 28). Disney's MagicBand: Breaking Down One of Hospitality's Greatest Innovations. Hotel Tech Report.

Hosanagar, K. (2018, July 24). The first wave of corporate AI is doomed to fail. Harvard Business Review.

How AI and machine learning can positively impact your business and small businesses. (n.d.).

Https:// (n.d.). Paraspot Software Website.

Hurler, K. (2023, February). I'm hating it: McDonald's AI-Powered Drive-Thru sucks.

Inc, A. (2023, November 28). AI in Short-Term Rentals: How Machine Learning Shapes STR. Medium.

Kamal, N. (2023, March 7). How McDonald's utilized big Data - Nuha Kamal - Medium. Medium.

Marr, B. (2017, August 24). Disney uses big data, IoT, and machine learning to boost customer experience. Forbes.

Martech, A. (2022, January 19). How do businesses use artificial intelligence? Wharton Online.

McAleenan, J. (2022, January 7). Identifying potential scam listings on Airbnb - Towards Data Science. Medium.

McDonald's is using AI and data to optimize its supply chain. (2023, September 25). ProcureCon Supply Chain 2024.

McDonald's: Loyalty program continues to beat inflation - The Anchor. (n.d.).

Mehta, I. (2023, November 8). Airbnb leans on reviews to make listings more reliable as it tests review summaries using generative AI. Tech Crunch.

Owen, R. (2021, October 4). Artificial intelligence at Airbnb – Two Unique Use-Cases. Emerj Artificial Intelligence Research.

Owen, R. (2022, January 26). Artificial intelligence at McDonald's – two current use cases. Emerj Artificial Intelligence Research.

Payments. (2022, January 27). McDonald's leverages loyalty personalization to combat aggregators.

Rogelberg, S. (2024, February 27). Wendy's will implement Uber-style surge pricing for your Baconator—with the help of AI. Fortune.

Roy, K. (2023, January 17). 6 Business benefits of outsourcing your AI projects. Medium.

Scaling Uber's Customer Support Ticket Assistant (COTA) System with Deep Learning | Uber Blog. (2018, August 23). Uber Blog.

Schaal, D. (2023, November 9). Airbnb's latest tools: 'Guest Favorites,' AI photo tours, and more. Skift.

Standley, E. (2023, May 11). How AI is Revolutionizing the Future of Airbnb.

Sthapit, E., & Björk, P. (2019). Sources of distrust: Airbnb guests' perspectives. Tourism Management Perspectives, 31, 245–253.

Sweeney, D. (2013, January 9). 4 Benefits that MagicBands bring to the wonderful world of Disney Parks. Forbes.

TensorFlow. (2019, March 6). Powered by TensorFlow: Airbnb uses machine learning to help categorize its listing photos [Video]. YouTube.

Uber Engineering. (2018a, June 5). Uber Tech Day: COTA -- Improving Uber Customer Care with NLP, ML, & DL [Video]. YouTube.

Uber Engineering. (2018b, October 22). Applying Customer Feedback: How NLP & Deep Learning Improve Uber's Maps.

Vaičiulaitytė, G. (2018, January 2). Twenty-five funny tweets about Amazon Alexa that prove there's nothing artificial about her intelligence. Bored Panda.

Why portfolio management is essential for AI projects. (2024, January 17).,value%20and%20ensuring%20ethical%20compliance.

Yeo, A. (2024, February 28). Wendy's is trying out dynamic pricing. It's not as Uber as it sounds. Mashable.

b i u quote

Save Comment
Showing 0 Comment