Emma - The Sleep Company is a leading sleep technology company that has revolutionized the mattress industry across multiple markets. With complex data operations supporting critical business decisions across their global operations, Emma depends on reliable data infrastructure to maintain their competitive edge and deliver exceptional customer experiences.

Results

• Cut pipeline runtime in half (from 7 to 3 hours) while adding more models and improving overall system performance

• Reduced bug ticket workload by 55% enabling the team to focus on strategic feature work instead of constant firefighting

• 10% to 100% test coverage on all new models with comprehensive documentation requirements

• Restored stakeholder trust in data quality shifting stakeholder preference from legacy to quality-assured models

• Eliminated manual monitoring replacing time-consuming manual job checks with automated alerting and proactive incident management

The Challenge

When Celine Geelan joined Emma's data team as Staff Analytics Engineer, she encountered a data infrastructure crisis that was severely limiting the team's ability to deliver reliable business insights. The existing setup lacked fundamental monitoring capabilities, creating a cascade of operational problems that impacted the entire organization.

"At the time we didn't really have any automatic monitoring in place. The morning checks of our jobs were very manual and we often missed things," explains Celine. "There wasn't much visibility into how our dbt™ project was running, and it was something that dbt Cloud™ didn't offer with our team's integration."

The impact on business operations was severe. "We often didn't even know we had a bug until a stakeholder told us. We would miss when jobs didn't run, when something broke, or when tests failed. We had basically no visibility," Celine recalls. This reactive approach had devastating consequences for data trust across the organization.

Adding to these challenges was a massive legacy project with complex Tableau dependencies. "We had a large legacy project that depended on many Tableau dashboards, but we had no visibility into the connections between them," explains Celine. "When we tried to refactor and simplify, we couldn't safely clean up because we didn't know what we might break downstream."

The Solution

Emma's evaluation process focused on finding a solution that could provide the monitoring and visibility capabilities that dbt Cloud™ couldn't offer with their existing integration. The decision ultimately came down to a critical capability gap that Paradime could fill.

"There was one thing that made it just super easy and it was, we have no monitoring and Paradime can offer us monitoring and dbt Cloud™ can't," Celine explains. This clear differentiator made the business case straightforward and the implementation decision easy for leadership.

Immediate Impact Through Bolt Monitoring

Emma's first step was implementing Paradime's Bolt monitoring and alerting system, connecting directly to their Microsoft Teams channels. This seemingly simple change had profound immediate effects on their operations.

"The first thing we did was set up a job that would send test alerts to a channel," Celine notes. "The monitoring setup was excellent because you get the alerts, someone checks the channel in the morning, and you can copy and paste all the failure information directly. It allowed us to can take action right away."

The team organized around this new capability, with their five-person team rotating daily monitoring responsibilities. "We had exactly five people, so each person could take one day of the week, which made it easier," Celine shares.

Radar for Strategic Performance Optimization

Beyond monitoring, Emma leveraged Paradime's Radar product to identify and systematically address performance bottlenecks across their dbt™ project.

"Radar was very useful because we could see a dashboard with all the problematic models - the long-running ones and those that were constantly failing," explains Celine. "That helped us prioritize which areas of the dbt™ project we needed to focus on and clean up."

The results were dramatic. "We had an ongoing cleanup project, and after four or five months we eliminated the worst-performing models and improved our main job execution time from six or seven hours down to under three hours. And we actually included more models in the pipeline, which makes it even more impressive," Celine shares.

Lineage Visibility for Safe Refactoring

One of Emma's most valued features has been Paradime's comprehensive lineage visibility, particularly the connection between dbt™ models and Tableau dashboards. This capability enabled safe refactoring decisions that were previously impossible.

"The second valuable feature that everyone in analytics engineering uses regularly is the lineage to Tableau," Celine notes. "That helped significantly with refactoring because we could identify models with so many dependencies that they would crash our browser, and we knew not to touch those."

The lineage visibility had an unexpected benefit for their BI team as well. "It really guided our BI team too. We could show them how many dashboards were connected to certain models and ask if we really needed hundreds of dashboards on the same model when there might be overlap. The BI team ended up cleaning up an enormous number of dashboards, and Paradime helped by visualizing these dependencies."

The Impact

The transformation at Emma has been comprehensive, affecting every aspect of how the data team operates and how stakeholders perceive data quality and reliability.

From Firefighting to Strategic Work

With automated monitoring in place, Emma's data team experienced a dramatic shift in focus. "We can actually focus on feature work. So now we are actually building new things and we're actively improving on things. We're no longer only working on bug tickets," Celine explains. "Before, probably 60 to 70% of our time was spent on bug tickets, and that's now down to a more reasonable 20%."

Quality and Testing Revolution

The monitoring capabilities have fundamentally changed how Emma approaches testing and documentation. "We went from maybe 0-10% test coverage to now having tests on everything we build," Celine shares.

Restored Stakeholder Trust

Emma shifted from reactive to proactive communication about data issues. "Even when failures happen, the fact that we're the ones informing stakeholders about issues has improved trust significantly." Celine explains.

This led to concrete business outcomes. "People start to trust our data more. We've seen a massive shift from people using legacy models to our new models. That's because we can now demonstrate that our new models are better documented, tested, faster, and more reliable." Celine notes. "Before we didn't have that trust. So really the trust in the data has increased massively."

What's Next

With their monitoring and alerting foundation solidified, Emma is focused on further optimization and comprehensive cleanup of their legacy systems.

"What's next is to further optimize our orchestration and also move it all to Paradime," Celine explains. "We've been focused on building new features. Starting next year, we hope to begin a more comprehensive cleanup of our legacy systems."

For Emma, Paradime has transformed data operations from a source of organizational friction into a competitive advantage, enabling the team to deliver reliable insights that drive business success across their global sleep technology platform.