Is “Personalization” Still the Word of the Year in 2021?

At the TravelTech Show, Ed Silver, CIO of iSeatZ and myself, VP, Travel, Transportation and Hospitality, at DataArt shared our views and experience on AI and personalization in travel in a panel called "Are we finally entering the age of AI-based Personalization in travel?"

Personalization has been popular in marketing for many years. In 2019, Personalization was named the Marketing Word of the Year by U.S. Association of National Advertisers.

A win-win nature of personalization

It’s not just a trend. In fact, personalization offers long-term benefits to both customers and companies. A recent Salesforce report on customers’ expectations found that 52% of respondents always expect offers to be personalized. According to PWC, 63% of American consumers are willing to share their personal data in exchange for more valuable service.

Personalization helps transform relationships with clients from a simple sale to a long-term partnership. It can provide true choice for clients, which can improve their experience. Furthermore, personalization helps extend the market, putting your company in a more stable position and making your brand more valuable.

It’s a win-win for customers and companies. Customers can receive promotions and discounts in exchange for their information. Companies, on the other hand, can see revenue increase by up to 15%, according to McKinsey.

AI vs. ML: What’s the difference?

Let’s dive into the basics and look closely at core purposes of AI and ML, two related topics that are often confused. Artificial intelligence is a broad category that refers to computer systems that perform tasks based on human intelligence. Both Artificial Intelligence and Machine Learning are considered scalable tools that can be used to build systems to analyze behavioral patterns of customers, link them to personal data and geographical locations, and create clusters of users. Based on cluster analysis, companies can apply predefined rules to diverse groups of customers.

Salesforce found that 60% of customers are open to using AI in customer engagement. Additionally, service providers may implement cross-sale and up-sale practices to better understand the needs of customers who are unlikely to book a service.

Machine Learning, on the other hand, is a type of artificial intelligence that can review large volumes of data and discover specific trends and patterns that would otherwise be invisible to humans. Its main purpose is to create structures without human intervention and form conclusions.

What’s in it for the travel industry?

The travel industry has started adopting these technologies, diversifying operations with AI and using machine learning algorithms to monitor browsing behaviors and purchase history. That data can be used to match products, packages, deals, and reminders to relevant travelers. It uses the results to extend appealing offers or to deliver personalized contextual help.

Swiss Deluxe Hotels uses an AI-based Smart Negotiation Solution to produce a personalized and instantaneous price proposal according to the guest’s history, profile, online interaction, and behavior. With this system in place, Swiss Deluxe Hotel guests receive the best service for the best price. If a prospective guest does not complete the booking process, Swiss Deluxe Hotels still benefits by collecting valuable guest data.

Uber offers a Smart Reply system that allows riders and drivers to communicate using in-app messaging. The system uses machine learning and natural language processing (or NLP) to anticipate and personalize responses to frequently asked questions. Drivers can reply with a single tap. Uber also offers personalized destination suggestions based on ride history and frequently traveled destinations.

Possible pitfalls

Although machine learning has many advantages, it is NOT for everyone. Here’s why.

• Machine Learning requires data sets to train on. These should be inclusive, unbiased, and sufficient in size.
• Another major challenge is the ability to accurately interpret results generated by the algorithms. Algorithms need to be chosen carefully if they are to work properly.
• It takes time and computational resources to train your model and make sound predictions.
• Machine learning has a high rate of error susceptibility. Let’s say you train an algorithm with data sets that are too small to be inclusive. You’ll end up with biased predictions coming from a biased training set. This leads to irrelevant offers being displayed to customers. In this case, these blunders can set off a chain of errors that can go undetected for quite a while. When these errors are noticed, it can take time to recognize the source of the issue, and even longer to fix it.

At AI’s current level of sophistication, the bottleneck for many applications is finding the right amount data to “feed” the software. Big Data can be useful, but for many applications, it can be more fruitful to ensure you have a strong data set that clearly illustrates the concepts we need AI to learn. This means the data should be comprehensive in its coverage of important cases and should be labeled consistently.

Shifting your focus from software to data offers a significant advantage: it relies on the people you already have on staff. There’s currently a huge shortage of AI talent, which makes hiring difficult. A data-centric approach lets subject matter experts who have a vast knowledge of their respective industries contribute to AI system development.

Ed Silver pointed out that since iSeatz is in the digital travel and loyalty space, they also see and use the power of AI-based personalization every day. iSeatz collects information about their customers’ behavior and analyzes it to target them in a natural way, offering highly contextualized options. As a result, loyalty members will likely be the first to return. According to iSeatz’s research “the most engaged members of a loyalty program have a 5X higher 10-year average customer value, are 35X more likely to have a travel rewards credit card, book 8X more frequently (with 50% fewer cancellations) and spend 10X more on core products.”

Along the same lines, DataArt’s AI/ML Center of Excellence recently produced a solution for predictions based on multi-dimensional data sets that scored in the top 0.6% out of seven thousand participants on Kaggle open competition. After being trained on enough data, the data sets could help busy travelers solve planning problems on their next trip.

To sum up, a large amount of processed, well-prepared data can help AI and ML mechanisms achieve success. You don’t have to spend fortune on AI to get there. You can start with native AI tools that allow you to test and learn. If you need a custom solution, one of the most efficient ways to start working with AI/ML is to get services from an expert-level provider.

Connect with DataArt and find out about the AI/ML service options available for your business.