Every airport wants to move from firefighting disruption to predicting it, yet most are held back by fragmented systems and incomplete data foundations. International Airport Review explores what is possible once a robust foundation is built.
While the aviation industry is eager to adopt artificial intelligence (AI) and predictive analytics, these advanced tools are functionally useless without a robust data architecture. Across the globe, airport leaders are realising that rigorous data preparation is the critical first step before advanced agentic AI can be deployed.
As Fernando De Castro, Head of Business Development, Innovation and Operations Excellence at Zurich Airport Brazil, emphasises, the principle of “garbage in, garbage out” applies heavily to airport operations. When building its AI capabilities, the airport group is prioritising data structure first, noting, “we are structuring the data to make it more comprehensive, enabling more structured projects for the future.”
Capturing the baseline by digitising the physical environment
You cannot predict what you do not comprehensively measure. Airports must first digitise physical processes to capture baseline historical data. Hong Kong International Airport (HKIA) has pioneered this with a “digital apron” initiative developed in collaboration with its civil aviation department.
By combining visual apron images with automatic flight information, such as call signs and destinations, the airport captures precise data points, notably “the timestamp for the whole aircraft turnaround process,” says Steven Yiu, Airport Operations Executive Director of Airport Authority Hong Kong. The airport can then use this information to analyse delays or time performance. The historical digitalisation acts as the essential foundation for their technology ecosystem, providing “the platform for our future AI application once we digitalise the process.”
Breaking down silos with centralised dashboards
Data is only actionable if it is centralised and visible to decision-makers in real-time. Hartsfield-Jackson Atlanta International Airport manages its massive throughput by relying on localised data aggregation. “A lot of decisions are made based on data,” explains Rebecca Francosky, Director of Air Service Development. “We have several Power Bis on the operational and managerial side. From an operations perspective, we track how many people are going through TSA, what the seats look like, what the TSA is projecting versus how many seats there are.” This cross-referencing allows operations teams to anticipate bottlenecks and ask.
Zurich Airport Brazil applies this centralisation across a wider geographic footprint via a Remote Operation Centre based in Florianópolis. This centralised hub allows them to manage resource allocation and operational planning for multiple regional airports simultaneously. By aggregating “weather conditions from the airports and also from which region, the queue times with analytics, with push notifications,” they can instantly notify response teams when a queue forms.
Moving from analytics to predictive simulation
Once data is structured and centralised, airports can move beyond real-time analytics to simulate operations and prepare for future states. HKIA has advanced into this territory using a Baggage Handling System (BHS) simulator. Before operations start each day, they load the flight schedule and expected passenger loads into the simulator to forecast “the demand of each baggage conveyor belt, so we can do load balancing.” This allows management to strategically align staff and operations because they can identify the peak time of the system.
The human element in a data-driven world
Despite advanced simulations, data forecasting will always face the hurdle of unpredictable human behaviour and rapidly changing airline schedules. De Castro points out the difficulty in trusting initial airline slot requests: “When the airlines start the allocation of these slots, we know that the number that is not going to happen is that one.” To create accurate forecasts, airport professionals must combine raw capacity numbers with “market economics, understanding how the demand is flowing so we can trust and understand that these numbers all make sense.”
For Charlotte Douglas International Airport (CLT), they prefer a hybrid model which combines human intelligence with digital tools and AI. As Martha Edge, Innovation and Experience Director, explains: “The hybrid model for me is my favourite, having empowered people on the ground armed with information on what is coming and able to make those decisions themselves.” This leaves humans “to handle the emotional side” of customer service, and AI is left to “do all the boring, repetitive tasks.” This sentiment is echoed at Fredericton International Airport, where Johanne Gallant notes that integrating AI simply makes the team more efficient: “It gives you time to do other things, to think outside the box and strategise more. Before, you’d spend a lot of time doing some of the intelligence work that’s now being done by AI.”
Unfortunately, according to De Castro, airports do not have all the information they would wish to have in many scenarios. To solve this, we need industry collaboration and a willingness to share data, as it will benefit all parties.
We will be exploring this topic in even greater depth at the International Airport Summit taking place on 10-12 November in Rome. In our fireside chat ‘From firefighting to foresight: building the data foundation for prediction,’ panellist Giovanni Russo, Chief Operating Officer of Geneva Airport, will examine how to connect core operational systems, build reliable real-time data feeds and tackle longstanding interoperability challenges between airport, airline and ANSP platforms.
Make sure you are in the room when Giovanni takes to the stage.
No comments yet