JW Player Rewrites the Script to Put Teams in Control of Their Data

This customer story was contributed by Emily Hopper, Senior Data Scientist at JW Player. JW Player provides video hosting and streaming, advertising services, and analytics for a billion monthly unique viewers in 200 countries.

Introduction

As a senior data scientist at JW Player, I work alongside a team of data professionals, including five data engineers, three customer-facing analysts, three data scientists, and three experts in product analytics, who focus on providing insights for internal data requests.

At JW Player, data plays a crucial role in helping us understand how our customers interact with our product and how we can improve it. By closely tracking how our clients use our products, we’re able to provide more effective customer support and make more informed recommendations. This allows us to better address any issues and ensure that our clients get the most out of their investment.

In 2022, we partnered with CastorDoc (now Coalesce Catalog) to help us improve the data discovery experience across the organization.

Catalog has been instrumental in helping JW Player with two essential business aspects: data discovery and team productivity. Our aim is to make data available to all members of the company, regardless of their technical background. We envision a scenario where every employee can use Catalog to harness the full potential of our data.

Challenges: The need for a central data repository

When I joined JW Player in 2021, the company held a summit where all employees involved with data gathered to discuss data issues and ways we could enhance our data strategy. During the summit, colleagues highlighted the pain of not having a central repository for documentation in the company.

To address this, we conducted interviews with stakeholders, including customer support, account managers, support engineers, data engineers, product managers, marketing, and anyone who might benefit from a data catalog. One of the most pressing issues that emerged was data discovery, particularly for non-technical people who found it difficult to navigate the data warehouse.

“One of the most pressing issues that emerged was data discovery, particularly for non-technical people who found it difficult to navigate the data warehouse.” —Emily Hopper, Senior Data Scientist, JW Player

As a new employee, I personally struggled with this problem. Even after six weeks of onboarding, I had no idea how to find new information other than asking, and different people often provided different answers to my questions.

Given requests like “I need this information split out by account key,” I would have to coordinate with colleagues across several teams to find all the right tables to join backend-relevant segmentation to performance monitoring data to customer-support-relevant segments.

We were primarily looking for lineage tools so that data producers would understand what downstream processes could be impacted by any changes made upstream.

We thought lineage would also be useful for building a single source of truth. People in the organization use different names and terms for the same things, which can be hard to follow. For example, metric X may be known by two different names across two different teams. We thought that if we could easily trace the source of metrics, it would be easier to know when metrics were the same (or different).

We were also looking for a central repository for documentation since our company had dispersed documentation across various platforms, such as GitHub markdown pages, Google Docs, Confluence, custom knowledge websites, and Slack. The scattered documentation made it difficult to find the answers we needed when we had questions.

Solution: Finding the right fit for all users

To find the most suitable data catalog solution, we followed a simple method. We began by compiling a comprehensive list of features that we deemed necessary, based on insights gathered from surveys and discussions.

Next, we evaluated a range of tools against these criteria, including open source options. In total, we assessed five different solutions to ensure we made an informed decision.

We then shortlisted two potential options and conducted a trial where participants dedicated some time to test each platform and provide feedback via a survey. Catalog did the best job at ticking all the boxes for what we wanted.

One factor that differentiated Catalog from the other solutions we evaluated was its user interface. Not only was it visually appealing, but it was also user-friendly, which aligned with our goal of making the tool accessible to business users. Specifically, we were seeking a solution that was approachable and less intimidating, and Catalog offered just this.

We ultimately chose Catalog because of its strong, easy-to-use lineage feature, which helped us gain a better understanding of our data flow. For example, I found the “Show SQL Source” button to be very practical and easy to locate, setting Catalog apart from other tools we evaluated. While some of the other tools may have had similar capabilities, they were not as user-friendly as Catalog.

“We ultimately chose Catalog because of its strong, easy-to-use lineage feature, which helped us gain a better understanding of our data flow…. While some of the other tools may have had similar capabilities, they were not as user-friendly as Catalog.” —Emily Hopper, Senior Data Scientist, JW Player

Stakeholders don’t have the time to watch lengthy training videos or go through extensive training sessions, so we needed the tool to be easy to use from the start.

The implementation of Catalog was both rapid and straightforward, and the tool itself is highly secure and never accesses our data. With other vendors, we had some security concerns.