Data science is a rapidly growing field that companies are utilizing to enhance the design of their digital products. Think of data scientists as mad scientists, experimenting with data insights and advanced analytical techniques to concoct the perfect recipe for an amazing user experience and business success. It's a constantly evolving field full of potential, like a never-ending buffet of possibilities. But remember, the ultimate goal is to make sure the dish we serve packs a punch and leaves a lasting impact on the company.
In this article, based on the latest episode of the Product Builders podcast, we will look at what data science is and provide examples of how it's applied to produce positive results.
Leveraging data science is important for companies looking to stay competitive in the digital landscape. It can help deliver better experiences to users, generate valuable insights about your products, improve efficiency throughout your organization and much more.
Data science aims to make sense of complex data sets and use that information to inform decisions and solve problems. It involves using various tools and techniques to extract insights and knowledge from data. This can include collecting and analyzing large amounts of data, creating predictive models, and visualizing results.
On the surface, data science can be seen as a multidisciplinary approach to turn data into actionable insights and make better decisions. However, depending on the data science professional or organization you talk to, the role and definition of data science will vary.
According to Florent Blachot, VP of Data Science and Engineering at Fandom:
"In my many years of working in data science and building data science teams, I've never had any of my colleagues or potential hires define data science the same way. Though there are some common threads in the definitions I've heard, it's difficult to come up with an all-encompassing definition due to the many layers and areas of focus within the field. Data science exists on a continuum and I've found it helpful to form a framework for what data science looks like in action, which has five phases."
Florent says a large number of data science projects don't succeed because some teams fail to define their objective or they lose their "why" along the way. For a data science project to succeed, teams must understand their "why." They need to consider what business value they are trying to deliver, the impact, and why the greater organization should care.
The analytical phase is about understanding the data points you want to collect, how to collect them and deriving actionable insights from the process.
Once you have your insights, it's time to build a prototype or process for how to apply those insights in a way that helps you reach specific business objectives.
The industrialization phase consists of automation and optimization. Data scientists should strive to automate their data collection and analysis as much as possible. They should also focus on optimizing the precision and scalability of their methods.
The final phase involves ensuring your findings and process for achieving a specific goal through data science are usable and available to other stakeholders within your organization. Maintenance also involves keeping your code and process running smoothly while retraining your models at regular intervals if needed.
In our discussion, Florent shares an example of his five-phase data science process in action. While working at Ubisoft, his team was tasked with tracking a competitor's video game's churn score and figuring out how to identify players who were at risk of churning. Doing so would allow Ubisoft to form contact points with players at risk of churning while they were still in the gaming environment instead of trying to reach them once they'd already churned.
To start, they had to define what it meant to be a churner in this specific video game. During their analysis, they found that users who played once or twice in a 14-day period were much more likely to churn than users who played three or more times. They used this data to form their initial modeling code, which would send potential churn info directly to the CRM team so that they could contact relevant players. They also ran tests on different time horizons and built a churn score using standard machine-learning libraries.
Once the first three phases of the data science process were complete, the team worked through the last two phases to automate and optimize the process, update it as needed and make it user-friendly for the CRM team.
Florent shares two key data collection methods: surveys and taggers. For those unfamiliar, the term "tagger" can be understood as tagged information or actions. When a user performs a specific "tagged" action on a website or within a game, that info gets automatically collected and analyzed to highlight user behavior. Surveys provide useful qualitative data, while taggers provide concrete quantitative data about what users are actually doing.
This combination of both qualitative and quantitative data is helpful when trying to produce specific business objectives.
While analyzing user behavior for the Assassin's Creed III video game, Florent found that many players weren't playing the game as intended. Instead of playing the game stealthily, many of these users played with a more "run and gun" mentality. Due to this finding through quantitative data, Florent and his team decided to run surveys in parallel with their user behavior tracking methods. Doing so allowed them to understand their users' interests better when playing the game. With that data in hand, they were able to bring their findings to the developers and other teams so that they could market and design the game to meet the needs of different users better.
Data science will play an increasingly important role for companies looking to stay competitive in the digital landscape and beyond. It can help deliver better experiences to users, generate valuable insights about your products, improve efficiency throughout your organization and much more. However, any data science initiative needs to be tied to defined business objectives to be helpful. As Florent stresses, all projects must start with a value-driven "why" behind your actions. The in-depth technical aspects of this field are all guided by your end goal.
If you have any questions or comments about this topic, please reach out and get in touch!
Stay in the loop! Join our newsletter for monthly updates on industry news, company updates, recent work and job openings.
In this episode, Sunil Gunderia, Chief Innovation Officer at Age of Learning, sits down with us to discuss the importance of personalized education & how technology is increasing equity in education.
iOS or Android? 😅 An age-old question. Trying to choose between the platforms for your app development doesn’t need to be complicated if you know what to look for.