Crystal Widjaja - Pemanfaatan Big Data di dalam Bisnis GO-JEK | BukaTalks

Summary

Crystal, a data expert at Gojek, discusses the challenges and evolution of data management at the company as it rapidly expanded its product offerings. Initially dealing with a monolithic database, Gojek's growing microservices led to a decentralized and unstructured data landscape. Crystal explains how the team adapted by organizing the data to ensure it could be easily accessed and used by product owners, developers, and business units. She highlights the importance of aligning data with Gojek's North Star metric—completed transactions—and using innovative data strategies to drive growth and improve decision-making across the company.

Takeaways

• 🚀 Gojek experienced rapid growth, expanding from 4 to multiple services in a short time, creating new data challenges.
• 📊 Initially, Gojek’s data was managed in a single MySQL database, but as services expanded, data began being stored in various formats like MongoDB and Postgres.
• 🗄️ The company faced issues with unstructured data, often stored in raw JSON format, leading to difficulties for product owners and developers to access useful insights.
• 💡 Gojek began focusing on how to better organize data so that it could be easily explored and utilized for decision-making, aiming to enable creativity and discovery of hidden insights.
• 📈 The ‘North Star’ metric of Gojek became ‘Completed Transactions,’ which aligned all departments to focus on increasing this key performance indicator.
• 🔍 Product owners and business units began to focus on relevant metrics that directly impacted completed transactions, such as total bookings, driver availability, and cancellation rates.
• 🧑‍💻 Gojek used tools like Tableau for data visualization, aligning reports with business needs like customer experience, driver incentives, and cancellation reduction.
• 🛠️ Developers were also aligned with service uptime and driver availability, contributing indirectly to the North Star metric of completed transactions.
• 🔗 Gojek emphasized connecting different data points, even across unrelated services, to discover deeper insights, such as linking Go-Ride and Go-Food data with customer profiles.
• 🌏 The company utilized standardized geographical data (S2 IDs) to analyze event-level data, such as driver location, helping them to assess the impact of specific events or regions, like traffic in Sudirman.

Q & A

What motivated Crystal to return to Indonesia and work for Gojek?

Crystal returned to Indonesia due to Gojek's potential for growth, particularly in mobilizing the informal labor economy and addressing Southeast Asia's infrastructure opportunities.

What was the initial scope of services offered by Gojek when Crystal joined?

When Crystal first joined, Gojek offered four main services: GoRide, GoSend, GoMart, and GoFood.

How did the data management challenges evolve as Gojek expanded its services?

As Gojek expanded its services, the data management challenges grew. Initially, all data was stored in a single MySQL database, but with new services, the data became fragmented across multiple databases like MongoDB, Postgres, and Logstash, leading to disorganized and unstructured data storage.

What philosophy did Gojek initially adopt when dealing with their growing data?

Gojek's initial philosophy was to prioritize storing data, even if it was unstructured or messy, under the belief that bad data is better than no data. This allowed teams to be data-driven and make decisions, albeit with some challenges in data organization.

How did Gojek's approach to organizing data change as the company grew?

Gojek shifted its approach to organizing data by categorizing it in ways that would make it easier for different teams to find relevant insights. They began organizing data by product lines, business units, and data types, aligning the data structure to meet the needs of product owners, developers, and PMs.

What is the 'North Star Metric' at Gojek, and why is it important?

The North Star Metric at Gojek is 'completed transactions.' It is important because it aligns the entire organization around a single goal, ensuring that every team understands how their efforts contribute to increasing completed transactions.

How does Gojek align product teams and PMs around the North Star Metric?

Product teams and PMs are aligned by breaking down the North Star Metric into more specific metrics, such as total bookings, allocation rates, and cancellation rates. These metrics help each team understand how their features and efforts contribute to the overall goal.

What challenges did product owners face when using Gojek’s data initially, and how were they addressed?

Product owners initially struggled with using Gojek’s disorganized data, as it was stored across different microservices and databases. This issue was addressed by reorganizing the data to make it more accessible and aligning it with key business metrics, allowing product owners to find the insights they needed more easily.

What role do visualization tools like Tableau play in Gojek's data management strategy?

Visualization tools like Tableau are central to Gojek's data strategy, as they provide an interface where teams can explore and analyze data in relation to key metrics like total bookings, driver allocation, and cancellation rates, helping them make data-driven decisions more efficiently.

How does Gojek’s data linking improve its ability to derive insights from different services?

By linking data across different services, such as GoFood bookings with driver statistics and economic data, Gojek enables teams to find unexpected relationships and derive insights that wouldn't be obvious if they looked at each service's data in isolation.