The majority of footballs CO2 emissions are generated off the pitch – they happen in the stands, or rather, on the way there. Fan travel accounts for the largest share of emissions during a matchday, yet understanding this environmental impact has remained elusive due to scarce data on transportation habits. Together with Bayer 04 Leverkusen, Vodafone has developed an AI-powered solution that taps into mobile network data to estimate and ultimately reduce the climate footprint of football games and similar major events – making real progress toward greener goals.
A new Kick-Off for Sustainability in German Football
With the German Football Leagues (DFL) new sustainability guidelines in place, clubs are now expected to measure and report their CO₂ emissions – and rethink how fans get to the game. The challenge? Around 80% of a football match’s total emissions come from fan travel, yet there’s been no reliable method to measure it due to a lack of data on transportation modes, travel patterns, and fan behavior.
This pressing need sparked a pioneering collaboration between Vodafone’s Innovation & Tech Center in Dresden and Bundesliga club Bayer 04 Leverkusen, with Jörg Knoop, Head of Center of Excellence AI at Vodafone, presenting the proof of concept: a system that estimates CO₂ emissions from fan travel using mobile network data.
The Hidden Power of Mobile Data
With nearly 90% of people carrying their phones at all times within an arms reach, Vodafone’s mobile network – covering around one-third of the German market – holds an incredible, anonymized data source that reflects real-world human motion. By transforming this data into actionable environmental insights, Vodafone only supported Bayer Leverkusen in advancing its ESG initiatives but also created and developed a scalable, AI-powered data product with real-world impact and broad potential for use across various applications.
Tracking Emissions of Fan Travel with AI: A 3-Step Playbook
- Motion Data Collection: Vodafone collected anonymized mobile motion data from devices in the greater vicinity of the stadium, focusing on activity five hours before and after the match. Using triangulation between cell towers and timestamps, fan journeys were reconstructed, including movement speed, distance, and acceleration – forming the key features for a AI-driven identification of travel behavior.
- Transport Mode Prediction: An ensemble of AI models was trained to predict how fans traveled – by car, train, bike, or foot. A hybrid model combining classical machine learning (random forests) with deep learning encoder-decoder networks was used. These models were trained on a mix of publicly available datasets and semi-labeled data generated by Vodafone’s development team.
- CO2 Emissions Calculation: The actual travel distances were calculated using a routing engine, and emissions were estimated by Partnering with ESG data specialist Envoria through applying average emission factors per transport mode. Vodafone upsampled the results to reflect the full fanbase, to account for those not using Vodafone services. While individual emissions can vary (due to carpooling, vehicle types, etc.), this method offers a reliable, reproducible approximation.
Turning Insights into Action
The proof-of-concept game in Leverkusen revealed powerful insights:
- 64% of fans arrived by car
- This group accounted for 88% of total CO₂ emissions
- Resulting in an estimated 277 tons of CO₂ for a single game
These insights give clubs concrete starting points to optimize transportation strategies, improve fan communication and incentivize more sustainable travel.
Key Takeaways: Embracing Imperfect Data and Turning It Into a Valuable Product
Data Availability Over Accuracy
Sure, the model and data for measuring the CO₂ emissions of fan travel doesn’t capture everything – carpooling, exact vehicle types, upsampling or edge cases. The results are, by design, approximations. But they’re accurate enough to uncover meaningful patterns, make matches comparable, and support smarter data-informed decisions. Waiting for the “perfect” dataset or an ultra-precise model often leads to inaction – and missed opportunities. Vodafone proves that even imperfect data can still be a very powerful driver of real-world impact.
Turning Data into a Product
Vodafone’s primary business is telecommunications, but the value of its network data goes far beyond just improving its own operations. It isn’t just about calls and texts anymore it’s a massive, underused asset that can generate new business value. By productizing this data – ethically and responsibly – Vodafone steps into a new role: an enabler of sustainable transformation across industries.
Collaboration is Everything
This project didn’t happen in isolation. It required deep collaboration between a telecommunications company, a football club, and a sustainability expert. It’s a reminder that when technology, data, and domain expertise come together, we can solve challenges that once felt impossible.
Kicking Off a Greaner Future
Vodafone’s AI-driven solution demonstrates how mobile data can be repurposed for environmental impact – not just connecting people, but helping to connect sustainability goals with real behavior patterns. The implications go beyond football. Concerts, festivals, and any large public event could benefit from similar approaches.
