You can’t have artificial intelligence without data. Lots of it. But as we grapple with how to take best advantage of the promise AI holds for managing business travel, we have to ask what kinds of data should be captured and utilized?
That’s a question that may bring some new answers in a post-pandemic world. When travel went pear-shaped in 2020, so did most of the historic data around it, and now that we’re on the rebound, new data points may require different tools and a different data strategy.
“Travel management is no longer solely about measuring spend versus savings,” says Jamie Pherous, managing director at CTM. “While helping customers optimize their budgets remains crucial, the scope of travel management has become far broader and more complex.” Beyond tracking standard metrics like fare type usage, advanced booking days and spend by supplier category, there are now many extra factors to consider in the travel program’s overall performance.
Without doubt AI thrives on Big Data, since without significant data sources AI can’t learn and deliver results, identify trends and patterns and share outcomes to help drive program change, says Julian Russell, executive director information technology and supplier relations, GlobalStar Travel Management. He maintains that the potential of Big Data and AI is clear but the delicate balance between input and output must be considered, along with a continued definition of the ‘why’ as well as the ‘how.’
“Global reporting tools matured some time ago to a point where they were able to provide significant bespoke dashboards which were often rich in data, but poor in context and meaning,” Russell says. “At a time when corporate travel managers are often stretched beyond breaking point, it’s important that the seemingly endless possibilities provided by Big Data and AI are managed carefully without leading to a case of quantity over quality with no real quantifiable output.”
Russell notes that the potential to introduce new data points is critical, but there must be careful consideration in understanding what those data points add to the overall story. He points out that traditionally, corporate travel programs have focused on trip data, routes, carriers, properties, rates, class, accounting references and spend.
Analysis of such information has supported travel managers in identifying KPI adherence, savings opportunities, adherence to policy and risk management. But other data can also be revealing.
“Information from additional data points can potentially start to consider soft data to provide a clearer picture,” he says. This can include incorporating traveler feedback to support well-being, comments directly from the supplier to inform decisions and richer environmental data to support sustainability programs.
“This doesn’t necessarily require new tools or data strategies,” Russell adds. It just means a change in the way existing tools source content, analyze and consolidate that content, and ultimately how the output clearly interprets the additional data sources.
“Historical data is great, but we need to append and complement with new insights and adjust models to accurately reflect the current and future state of travel,” says Shital Sabne, director of data products at ARC. “The tools and strategies employed should be flexible and responsive to the ongoing changes in the travel landscape.”
He notes that new travel patterns and traveler preferences have emerged since 2020, upending the models previously used for year-over-year comparisons in our data.
The benchmarks can indeed pose a challenge, asserts Steve Reynolds, general manager for Emburse Travel. You can’t compare today’s hotel rates and air fares to those during COVID, he says, but must pull out the outlying years of 2020-2022. “This makes trending difficult but at least we can now compare to 2023 and determine which way rates and fares are trending.” For a more historical perspective, that can mean looking at 2019 and earlier, skipping 2020-2022, and then including 2023 and 2024.
Other kinds of data also come into play, along with the need for speed when necessary. Today’s travel programs need real-time data to respond to the more dynamic environment and make informed decisions quickly, says Chloe Carver, head of corporate travel for Acquis Consulting. “New data sources, such as health data, economic indicators and traveler sentiment, can also be integrated into the data strategy to provide more nuanced and timely insights.”
Big Data, Big Driver
Old songs may tout love and marriage, but when you talk about things that go together today, you’d have to cite the combo of Big Data and AI.
“Big Data plays a huge role in making AI models successful and accurate,” Sabne says, noting that ARC employs AI and machine learning in several of its advanced data analytics solutions that are trained on roughly 2.6 billion annual passenger trips. “Without large amounts of data, AI models would not perform efficiently.” Too, the availability of Big Data drives innovation by uncovering hidden patterns and correlations that would be impossible to detect with smaller datasets.
“AI clearly thrives on a large set of data sources and the potentially endless points of content available through the world wide web,” Russell says. “The challenge comes with moderating and verifying that content and agreeing with the travel manager where additional content will deliver value.” He adds that fragmented content is already adding a layer of complexity to the corporate travel eco-system, demanding additional data points and better consolidation of that data to provide clear trends. “If AI is able to simplify the ability to source and consolidate this information without the need for endless API connectors, this will deliver clear value to the end user,” he notes.
“Knowledge is power, and that’s especially true for AI,” Pherous says. “Data is the key to making AI efficient and effective.” He notes that just like us, the more ‘knowledge’ AI has, the better and faster it can respond. Big Data allows us to train AI, sharpen its decision-making, and ultimately increase relevancy, accuracy and trust for its users.
One challenge, according to Pherous, is that some smaller AI startups in the managed travel space have limited access to the data required to optimize their AI models and need to rely on third-party data sets and tools. Risks in this approach include the concern that the AI is pre-trained on someone else’s data, and that when TMCs rely on third-party data, they risk violating data governance and security protocols by sharing sensitive information such as traveler booking data, profiles, payments or search patterns. CTM takes a different approach, he says, using open-source AI models that integrate with its proprietary tech and leverage its own data so that no client or traveler data are shared with third parties.
At the same time, it’s important to realize that while more data is generally better, quantity is only part of the conversation.
“Feeding AI with garbage in leads to garbage out,” says Michael Duffy, VP of product and innovation at Grasp Technologies. “However, that statement isn’t black and white when it comes to AI.” Artificial intelligence may not know how to qualify the data, he notes. What if a data row is ‘partially good’–will it use it or use the elements that are satisfactory? Will the bad data cause the AI to make false outputs? Having quality data is crucial for the success of AI as well.
Duffy agrees that it’s important to track against both internal and external benchmarks, but there needs to be a recognition that the comparisons will not always be apples to apples.
“A travel manager managing travel for both a sales department and engineering team would likely have different policies and different benchmarks for each,” he says. “But comparing both could still be helpful. And externally, it’s still helpful to understand what is happening in the market.”
Smart Moves
So how can travel executives make best use of data and analytics in a world where AI will be playing a bigger part? “Spend time learning about AI and thinking about how you want AI to play a role in your overall business and data strategy,” Carver advises. She notes that education in this area is crucial to making the right investments, and that AI and data literacy across the organization can help employees understand how AI is being used and enable the organization to fully leverage its capabilities.
Sabne notes that AI can help uncover new insights for travel executives, but only when it’s implemented in the right way. “Business leaders need to think about how an AI solution will help them solve a current challenge or take advantage of an opportunity, not view it as an end goal,” he says. “There are many different data sets already available to many businesses without the use of AI.” These include product, geographic, sales, customer and other internal data that can be mined using non-AI tools available.
In fact, determining what resources make the best choices is a major part of the current AI challenge. “AI is crazy expensive to implement and can take years,” Reynolds warns. “Companies currently claiming a quick application for AI are buying into the recent hype, but it’s really their current rules-based system with a few tweaks to look like AI.”
In designing AI data strategy, Carver suggests focusing on data governance and quality. “As AI increasingly influences decision-making in business travel, data governance becomes more and more important,” she says. “Executives should consider investing in tools that can cleanse and standardize data across multiple sources to enable more accurate insights.
