VINCI Airports manages 70 airports in 14 countries, welcomes 320 million passengers per year, and generates €4.5 billion in revenue. Faced with increasing operational and environmental constraints, the operator is turning to artificial intelligence to optimize its performance. A look back at a successful transformation with Benoît Forest, Director of Operations at VINCI Airports.
Who is Benoît Forest?
Benoît Forest was appointed Operations Director of VINCI Airports in January 2025, after having held the position of Head of Data for nearly three years. With over nine years of experience within the Group, he steered the company’s data transformation, including the deployment of the Smart Data Hub project across 15 countries, the creation of 10 use cases covering all strategic areas of VINCI Airports (traffic, commercial, operations, finance, environment), and the integration of predictive analytics.
He has a degree in Management Control from CNAM and combines financial and operational expertise acquired in the airport and highway sectors at VINCI Autoroutes.
The challenge: Unified management of a complex data network
VINCI Airports operates a diverse network stretching from Japan to Chile. Each platform has its own IT systems, local constraints, and operational specificities. While this diversity is a source of strength, it also complicates decision-making at the group level.
Their business model is based on two complementary levers:
- Aeronautical fees (50% of revenue) charged to airlines for the use of infrastructure
- Commercial revenue (30%–40%) generated by passenger spending, particularly in duty-free shops, restaurants, and car parks
This dual model imposes a key requirement: knowing, understanding, and anticipating passenger traffic. Everything else follows from this: team sizing, calibration of commercial offerings, financial forecasts, service quality, investment trajectories, and decarbonization strategy.
The challenge: Collecting and processing scattered, non-standardized, siloed data across the Group’s various airports to build a unified view of traffic flows. This is the purpose of the transformation launched with Artefact and Google Cloud. The result is the construction of a global Data Factory capable of federating network data, structuring it, and making it available for decision-making.
A data infrastructure designed as a performance lever
To meet this challenge of group-wide management, VINCI Airports chose a unified, scalable, and secure data architecture built on Google Cloud Platform, chosen to provide every business unit in every airport with the means to better understand, anticipate, and act, based on consolidated, contextualized, and analyzable data at different levels.
Working with Artefact, the group set up a Data Factory in 2023, capable of:
- Collecting data from heterogeneous systems across 14 countries
- Controlling data quality and integrity via automated safeguards
- Harmonizing data structure to enable reliable comparisons between platforms
This approach avoids the burdens of traditional IT harmonization. Thanks to the cloud, airports retain their local tools but expose their data in a common base, where it is cleaned, structured, and then used to feed dashboards, AI models, performance analyses, etc.
Google Cloud Platform: A robust architecture
To facilitate scaling up, the Google Cloud Platform architecture relies on proven components, designed for volume, speed of analysis, and security:
- BigQuery for storing and analyzing massive data
- Vertex AI for training and deploying predictive models
- Cloud Run + Streamlit to provide business teams with a simple, fast, and interactive interface
- Cloud Storage for centralized management of models and versions
- Cloud Build for continuous integration and fluid deployment of workflows
Three key use cases to boost airport performance
1) At the source: Passenger traffic prediction
Passenger traffic is the company’s key data. Predicting it makes it possible to:
- Guide long-term strategy (investments, opening new routes, financial projections)
- Scale day-to-day operations (security, reception, baggage handling)
- Manage commercial offerings based on passenger profiles
VINCI Airports and Artefact thus designed multi-scale predictive models tailored to the needs of each management level, providing:
- A consolidated view for group management, with annual flow projections
- Weekly or daily operational views to anticipate activity peaks
- A local view per airport, cross-referenced with behavioral data
These models are based on traffic history, exogenous variables, and weak signals. They enable trajectories to be simulated, human forecasts to be compared with algorithmic projections, and trade-offs to be optimized.
2) Improving airport operational efficiency
Thanks to data, VINCI Airports can now anticipate congestion points and adapt its resources in advance. For example, by scanning boarding passes, it is possible to predict when passengers will arrive at security checkpoints. By cross-referencing this data with line processing capacities, teams can adjust staffing levels in real time to ensure waiting times of less than 10 minutes. In terms of security, passengers at baggage security checkpoints pass through an X-ray machine to detect any potential threats.
3) Optimizing sales strategies for travel products
By analyzing traffic flows and purchases, VINCI Airports gains a better understanding of who consumes what, when, where, and in what context.
A British passenger in transit does not behave in the same way as a French passenger on a domestic flight. A 6:15 a.m. flight to Lisbon does not generate the same average basket size as a vacation departure to Punta Cana.
By analyzing boarding passes scanned at checkouts and cross-referencing them with traffic data, the group identifies consumption patterns and can recommend that airport retailers adjust their product offerings based on passenger profiles and destinations.
Success Factors: From implementation to business adoption
Engaging teams from the start of the project
Artefact supported VINCI Airports from the earliest stages of the project, taking a collaborative approach with operational staff right from the project launch. Enhanced business support is absolutely essential to strike the right balance between human intelligence and predictive models.
“The timing of onboarding operational teams is very important. It must start on day one. This ensures that data scientists understand the business concerns and incorporate highly operational needs into the design of solutions.” – Benoît Forest, VINCI Airports Operations Director
This makes it easier to develop concrete use cases that are aligned with real business challenges (traffic, security, retail, finance, etc.) and ensure their long-term adoption.
Data quality, the key to business trust
Initially, VINCI Airports teams viewed data governance as secondary. Very quickly, with Benoît Forest, they realized that it was actually fundamental. Because when models malfunction due to poor data quality, business trust quickly erodes.
VINCI Airports and Artefact have implemented a series of technical safeguards to ensure data reliability upstream:
- Detection of missing files (e.g., an alert such as “traffic not received from the previous day”)
- Structure control to ensure that schemas remain stable over time
- Integrity tests (an abnormal deviation at an airport triggers an automatic alert)
These automated checks, operated entirely within Google Cloud Platform, prevent silent drifts and ensure the stability of models in production.
“I think the success of this project lies in understanding the strategic business challenge, defining the scope, deploying the solution, providing training, and now the tool is used by business teams on a daily basis.” – Benoît Forest, VINCI Airports Operations Director
Continuously measuring predictive performance
VINCI Airports has implemented a post-operational evaluation process. Teams compare what the models had predicted with what actually happened on the ground. The objective is to identify discrepancies, understand their causes, and adjust the models to improve their accuracy over time. This efficiency is made possible by the deployment of AI.
Going further with AI: Systematizing prediction
With the stabilization of the data environment, VINCI Airports has embarked on a second phase: the transition to systematized prediction. Teams have trained models to refine traffic trajectories, anticipate activity peaks, and enrich their understanding of passenger behavior.
The approach is based on differentiated granularity:
- Annual strategic forecasting, useful for the financial department.
- Weekly or daily projections for operational management.
- Post-operational analysis, to compare what the model predicted with what actually occurred
“AI here allows us to move from local intuition to shared knowledge, without replacing teams, but by giving them the means to save time and focus on decision-making.” – Benoît Forest, VINCI Airports Operations Director
The rise of Generative AI for comprehensive data analysis
As part of the ongoing project, VINCI Airports is now exploring the potential of generative AI through three use cases:
- Secure GPT: An assistant integrated into dashboards
- Talk to my data: Conversational database queries
- Document intelligence: Automatic extraction and synthesis of complex content (procedures, audits, reports)
These developments are necessary and represent the 2025 challenges for Benoît Forest’s team. This wealth of information is still mainly materialized in real-time visual dashboards (Power BI). Although effective, these tools remain rigid and require constant development to evolve.
The ambition is therefore to allow every employee to interact directly with all data via autonomous AI agents, capable of answering complex questions and pushing the analysis far beyond the current capabilities of traditional dashboards.
