Data Spikes: The Real Games Behind the Tokyo Olympics
Context:
At MullenLowe Group Japan, we had a relatively unique offering where - within a relatively small team - we had a strong cultural insights output, as well as world-class data science and analytics. Combining our strengths, we looked at how we could proactively innovate an offering that paired these dual strengths: qualititative research and cultural intuition, alongside data science and analytics.
Solution:
With the Tokyo 2020 Olympics on the horizon, we looked at first-party data from one of our clients, Eurosport: a TV network that had access to viewership data for the preceding Pyeongchang 2018 winter games. First, we used sophisticated data science techniques to cluster audiences using unsupervised machine learning.
By mapping these cultural “tribes” holistically, starting with the data, we were able to demonstrate proof that a lot of our initial biases as marketers are probably wrong: For example, while one might expect that all skiing-related sports would appeal to a similar group, in the data, we proved high crossover among fans of alpine skiing, freestyle skiing, and snowboarding. But other competitions, such as ski jumping and the nordic-combined event, did not appeal to this group. Taking the analysis a step further, we then layered on other publicly available data about people in this tribe that we identified as "thrill-seekers". This revealed that judo, equestrian, squash and surfing are the summer events the same group would be likely to enjoy.
Rooted in this understanding, it was then possible to begin constructing a cultural narrative and strategy around these “tribes”, using tools such as Facebook Business Manager to pry further into adjacent cultural consumption and attitudes (e.g. in fashion, music and film), giving us the ability to construct a well-rounded picture of each tribe. We were also able to return to the original data set to tease out further granular insights into content consumption.
The thrill-seeking cohort, for instance, showed more willingness to replay visually intense events after the competition (as opposed to those who favoured less visual sports like curling), which indicated that brands wouldl be able to re-engage with these viewers even months after the official games end.
Although a purely speculative exercise in this instance, by finding a compelling ‘story’ for cultural analytics, and clearly proving its potential value to clients, we went on to use a similar process and methodology for many client projects that produced results that far exceeded conventional single-discipline, siloed research methods.
Overall, our strong belief was that combining data science with the study of cultures is not only effective but also necessary in a world where we may share more in common with people on another continent than our neighbors. In an industry where we are still briefed with unhelpful demographic markers such as "Asian millennial" - an identifier that could be applied equally well to a 20-something Muslim in Indonesia and a 30-something salaryman in Tokyo - brands need to understand that such over-simplifications are effectively useless. Given that culture is the best vessel for transmitting and amplfying a brand's message, traditional segmentation doesn't reflect the modern world.