It’s a challenge to find out what society thinks about a company or brand, especially when you have 40 million followers spread across 10 countries and 190 social media profiles, like BBVA. Even so, the winners of the BBVA Hackathon 2020 accomplished it in only three days, developing a tool to measure users’ feelings about the bank on social media, using machine learning techniques.
The hackathon was held completely online, making it more inclusive as more people from interdisciplinary backgrounds were able to participate. In total, 792 young people from Mexico, Peru, Colombia and Spain took part. Split into teams, they used their talents to solve the challenges presented to them.
Among all the competitors, Data Team, with members from several countries, won the hackathon. Their solution to analyse comments about BBVA on social networks, which they launched in response to the Social Listening challenge, led them to victory. Thanks to their innovative idea, the team won an 80,000 peso prize (around 3,500 dollars). “I am still excited, happy and grateful” says Mayra Uribe, a mathematics student and member of Data Team, a few days after the winners were announced.
A challenge about listening to and understanding society
Identifying how society feels about a company or brand is easier said than done. The hackathon’s Social Listening challenge was specifically designed to push teams to develop a tool to monitor society’s opinion of BBVA. “We wanted to better understand users on social media, gather their feedback and identify the impact of the BBVA Group’s activities at critical times such as the COVID-19 crisis”, explains Mónica Couret, one of the people responsible for the challenge, which was co-led by various data science teams at the bank.
More than 50 hours of work in a multi-disciplinary team
The members of Data Team, three of whom are from Peru, two from Colombia and one from Mexico, all with different professional backgrounds – from data scientists to product developers – didn’t know each other before the hackathon. Nonetheless, they achieved perfect coordination as a team, steering their project ‘Client perception in a click’ to success.
“Each one of us brought different skills and knowledge to the group. The fact that we have multi-disciplinary backgrounds was a crucial factor for the success of the project”, explain the team members.
To start, they focused on defining the tool they wanted to create, before exploring the technologies they would need to achieve this goal. “The idea was to make a dashboard that didn’t just display diverse indicators, but that also told a story,” outlines Guillermo Bastian, SAP consultant and member of Data Team.
Data Team members
Variety of data sources
The challenge required participants to use their creativity to determine the best sources of data, and their technical skills to analyse these sources. By not providing them with a specific data set, participants were given greater freedom to make the final solution even more innovative
The Data Team members analysed comments not only on social networks such as Twitter and Instagram, but also on the banking product comparison site HelpMyCash and the Google Play Store, to benefit from a diverse range of data sources. In order to make the study even more thorough, they employed scraping techniques to extract both the messages written by users and the emojis they used.
The team used natural language processing (NLP) to carry out a primary analysis of the information. After that, they labelled a sample of data in order to train machine learning algorithms and build a predictive model that classified comments as positive or negative. By doing so, in addition to providing a snapshot, they succeeded in training the tool to extract, process and analyse data automatically over time.
A successful solution and an enriching experience
During the three days of work, participants had the support of coaches who stayed in constant contact with them in order to provide them with assistance. The result of this collaboration and support is a tool that offers the possibility of visually analysing users’ feelings, drawing comparisons with other banks and understanding the perceptions of opinion leaders. All of these data allow marketing experts to make clear-cut decisions, a point that was highly valued by the hackathon judges.
Data Team’s project, which rose to the top spot of the challenge and the global competition, beating out 140 other teams, sparked the interest of different departments at the bank, which are going to look into the possibility of implementing it.
The six members of Data Team agree that the competition was valuable and a great learning experience. Its multi-disciplinary nature, the rapport built by working as a team and the creativity needed to apply the technology were three of the key ingredients to success.