From Rio to Roskilde: The real-life experiment for big data and sustainable cities

Back in August last year, I delivered a presentation on how the Chief Data Officer is fast becoming the next corporate rockstar.  “Rockstar?” They questioned.  Yes, total rockstar – except it’s not just the CDO that fans of business value are flocking to, it’s the data scientists and analysts that support them, playing to the tune of tapping data for insight to support more informed decisions and better business outcomes.

So it came as no surprise to hear about the fascinating story of Roskilde festival, who are using data and insight to better support their flock of rockstars and deliver the ultimate experience to thousands of music fans.

The story of Rio to Roskilde, as told by Per Ostergaard Jacobsen at IBM Analytics Agenda in Sydney, the Co-director for the Centre of Business Data Analytics at Copenhagen Business School (CBS), started in back in 2012 in Rio at the Rio+20 sustainability summit which Per Ostergaard describes as “lots of talking, but no action“.  The team wanted to create a working model to test some of the great ideas generated from the event, and proposed the Roskilde festival as the perfect real-life experiment to test ideas and train sustainability models, bringing the knowledge and lessons learned back to Rio and to the world.

As the largest festival in Northern Europe, Roskilde very much reflects the workings of a city with 130,000 citizens, but for just eight days!  With 175 acts and a community of citizens speaking the language of music, the 100% non-for-profit event consumes 200 tonnes of food and produces 300 tonnes of garbage – as such, the festival provides the perfect city lab to explore concepts around creating sustainable business and smarter cities.

The challenge:  how can we build sustainable business models using big data to create mutual value for citizens and community, and for customers and business.

Challenge accepted, Per Ostegaard and his team looked to the data for insights – and there certainly was plenty of it!  With the 15 data sources including over 91 million observations captured by the festival mobile app, 61 million observations captured on social media, 12,000 interviews, plus weather data and sales of tickets, merchandise, food and beverage – it was absolutely critical the team build a platform that could support both structured and unstructured data, captured in real-time, with the capacity to scale as data volumes increased.  Per Ostegaard reflects the team was somewhat naïve when they started about the volume of data that would be generated during the course of the 8-day festival, but fortunately they made sound architectural decisions up front about the use of cloud computing and so could scale to accommodate as needed.

Common to big data best practice, the team created a landing zone where the data feeds were first fed for preliminary analysis.  Of key importance was that only two people in the team had access to this landing zone, as the team took a strong stance on ensuring privacy and ethical standards in the use of customer data.  As you can imagine, a wealth of customer data is captured during the event – not just the purchases they’re making, but where they are located and moving throughout the event – so the critical first step was to de-personalise the data to remove any attributes that could be use to identify any one particular individual.

The data was then moved into a development zone for analysis, and later to a demonstration zone so key stakeholders could see the insights derived around the event.

From a technology perspective, the team leveraged a range of IBM and home-grown capabilities to ensure superior scalability, security, and to speed up the installation and configuration process – taking just one day to provision the environment.  This included a combination of Watson Analytics, SPSS, Bluemix, DashDB, R/Excel and Sodato (created by CBS).

One of the fascinating insights captured included a time-lapse video of mobile activity during the course of the festival – as demonstrated by Per Ostergaard here:

Analysis of the event data uncovered key insights about how the food and beverage sales changed throughout the day, for example, strawberry smoothies were the recovery drink of choice by music fans and sales of roasted pork sandwiches peaked at 6pm, as well as the direct impact that weather had on purchases.

It also uncovered important associations across concert attendance that provides valuable insight when making decisions around community safety, customer service and location of merchandise stands, particularly around the movement of crowds both during- and after- each concert.

Understandably, the team expected festival attendees to go and listen to live music, but the data showed 10-15% of attendees never actually went to a concert!  Instead, they attended the festival to socialise in the camp area and party with their friends.

One of the most fascinating perspectives Per Ostegaard shared was the concept of tribal movements – when people go to a concert they don’t move as one person, they move with their tribe, or group of friends.  When it comes to marketing, you have to make decisions on how to attract and serve the tribe, not the individual.

These are just a few of the insights uncovered by Per Ostegaard and his team that will be used by the festival organisers to plan and execute an even better event in the years following.  The Rio to Roskilde story is a great example of how data can provide incredible insight into customer behaviour, to understand how to market and serve them better.  But more than that, it creates opportunity to make smarter decisions around safety and environmental impact – lessons that can be taken from Rio to Roskilde, from Roskilde to the world.

Now, time to put together the business case for a site visit next year! 🙂