Empowering a data-driven culture with a modern data stack at OneFit
Oct 12, 2020
As head of business intelligence and data analytics at OneFit, I'm committed to helping our parent and sister companies — Urban Sports Club and Fitogram — transform into data-driven, digital-first businesses. As we’ve modernized our data platform, I’ve concluded that a data culture is worth every minute you invest into building one. As you read our story, you’ll see why.
The OneFit ‘sweat and joy’ lifestyle
OneFit is a fitness platform that allows people to work out in many locations with a single membership. In Amsterdam for example, members can take classes or train at more than 500 gyms and studios. They can also use their membership to participate in a variety of group activities, including yoga, boot camps, dancing, and swimming.
With the OneFit platform, members get a vast number of options to choose from when deciding where, how, and with whom to workout. Once someone has joined, they can find their preferred location on a map and book a class in an instant, and can even invite friends.
Why pursue a modern data technology stack?
When I first joined OneFit, I quickly realized I had my work cut out for me. Management’s goal was for us to achieve operational excellence, but getting there required jumping some big hurdles. We had no single version of trusted data — there were several key performance indicators (KPIs) for the business, but the associated metrics were poorly defined. Because of this, people reported different results for the same KPIs, making it difficult to translate business goals into actionable and unique metrics.
At that time, all our data lived in spreadsheets, CSVs, text documents, and various databases and services. Because we lacked a centralised reporting system, servicing an ad hoc request sometimes took as long as two weeks. Many of my colleagues were even creating hand-drawn charts and geographical heat maps to show sales and demand in certain areas of the cities we service.
Setting business priorities
Considering the challenges we faced with efficiently using data, everyone sensed that leveraging a business intelligence (BI) platform would contribute to positive business outcomes. However, not everyone could fully grasp the value of what data and analytics could do for them in their role, so naturally there were doubts and questions when we started on our data journey.
As a leader during this transition, my first action item was to assess the critical concepts that would have a noticeable and immediate impact on business, so I prioritised defining our company KPIs and metrics. My second action item was to get our basic business needs reflected in the data, which required getting input from each department, manager, and our CEO. My third action was to look into the ways we could automate reporting so that time-intensive tasks like getting our executives reports every Monday could be done in less time and with just as much impact. I was determined to find a way to make the process simpler and faster, knowing that doing so would be a great way to show the impact and value of a first-class, modern data stack.
Building a data stack from the ground up with Google BigQuery + Looker
Becoming a data-driven organization started with choosing a data warehouse. At OneFit, we have data coming from \~20 different sources, including customer tickets, our production database, our CRM system, and ad data from Google, Facebook, and Instagram. As we weighed our options, we knew we needed a data warehouse that could consolidate all our data and make it easily accessible to everyone, which is one of the reasons we chose Google BigQuery.
Another thing we liked about BigQuery is that it let’s you define all the various data relationships from right within the warehouse and then feed that directly into a data platform, Big Query BI Engine, Salesforce, or machine learning application. Plus, because BigQuery is part of the Google Cloud ecosystem, it fit perfectly with the other services we were already using — like Google Maps Platform in Google Cloud and the Kubeflow machine learning management system.
For our BI layer, we evaluated roughly 10 different tools before deciding on Looker. Our primary reason for choosing Looker was because of its agile modeling layer, LookML. We were also impressed at how secure and simple to manage the platform was, and that it allows full version control and can connect to any database — providing a single, centralised location for our core business logic. The Looker web interface was another big selling point for us, as it makes self-service data exploration, data visualisation, and exporting results easy for any user, regardless of technical expertise.
Driving buy-in and adoption
Once we had our data stack in place, we set out to establish use cases that would help justify the value of our new data-driven approach to everyone within the organization.
We started with the low-hanging fruit: investor reporting. Thanks to Looker, instead of taking days to pull insights and assemble these reports, the finance department can now generate these quickly through dashboards. We also set out to show the correlation between sales and demand in each city with data. With BigQuery and Looker, we can check the inventory demand in different cities based on the number of members in each location. This then empowers our sales team to determine where there are untapped opportunities and eliminates the need to manually draw heat maps to track sales and demand.
We also wanted to prove the impact of the new data stack by creating efficiencies within our community success department. In the past, every time a member called in with an issue, our team would have to go through a process that required accessing multiple systems, and took at least one minute per call. By connecting Looker directly to Zendesk, we automated this process so that now, anytime a member calls, a team member receives an embedded pop-up with all the information they need to help service the caller.
These data-driven proof points not only had a huge impact on the business, but spread the buy-in of our new data stack and adoption of these new tools across the organization in just a few weeks' time.
A flourishing data culture
With our technical foundation laid and buy-in acquired, we shifted our focus towards building a data-driven culture that could serve the needs of the whole organization. To not overwhelm the data team while still ensuring that the privacy and security of our data was in line with the GDPR, we developed a multi-pronged strategy to create a data culture.
We began by determining the various personas of people within our organization — analysts, data scientists, product managers, marketers, and others — and then trained data ambassadors in each group. During these trainings, we defined the different metrics that were pertinent for each team, and created a self-service model for every department. The data ambassadors then passed on their knowledge to others within their groups, helping us plant seeds that would later blossom into a data-driven culture.
By first taking the time to understand how everyone was getting the data they needed when they needed it, we were able to democratize access to data across the entire organization. In less than 18 months, we created a thriving data-driven culture that enables everyone to make business decisions with data. Even with our small, six-person data team, we’re able to support this company-wide culture where today, 83% of our Looker dashboards are being used weekly across all of our teams.
Democratizing data to drive success
For any company looking to build a modern BI infrastructure, I recommend defining your data sources, understanding what business metrics are core to your success, and building a data stack that will support a single source of truth. And when it comes to developing a data culture, I encourage you to democratize data access in a way that builds trust and stems from understanding what every individual needs to find their own success with using data.