How Aurora Built A Golden Source of Truth for 1000 Users

Aurora Innovation, a leader in autonomous vehicle technology, evaluated Holistics alongside 40+ other BI tools.

Aurora selected Holistics for its ease of use and powerful semantic modeling layer, which enables the data team to define consistent, reusable data definitions and manage them centrally, creating a golden source of truth for 1000 end users.

The feedback has been very positive, especially in terms of the ability to build and iterate quickly and respond to data requests. Once a dashboard and dataset are set up, making changes is fast and seamless, which has been incredibly useful.

Our leaders are also experimenting with dashboarding, which is great to see—something that wouldn’t happen if everything had to be done manually with SQL.

– Juan Argote, Director of Aurora Data Science

Holistics arrived just when Aurora’s previous SQL-based BI tool could no longer keep pace with its rapid growth. The tool did not provide the necessary level of support required by a fast-paced, data-oriented organization.

Reliability was a struggle. Dashboards might break whenever data definitions change.

Without a central data model, different teams wrote their own SQL queries, leading to disparate data definitions and conflicting reports. As the number of dashboards grew, maintaining them became more labor-intensive and error-prone, with every minor adjustment requiring major interventions from engineering or data teams.

“Sometimes dashboards wouldn’t be available, and we knew we needed to find an alternative solution,”Juan recalled.

Beyond reliability, performance was another issue. Query performance was dependent on the vendor’s infrastructure, leaving Aurora with little control.

As the volume of data and complexity of queries increased, the SQL-based tool began to buckle under the pressure of handling large datasets and complex integrations, leaving teams struggling to access the insights they needed. The engineering team even considered building their own dashboard solution.

At one point, we even had an internal engineering team working on ad hoc dashboarding solutions. We needed a tool that could grow with us, adapt to our needs, and empower our teams.

– Juan Argote, Director of Aurora Data Science

And so, the search for a new BI tool began.

While Aurora’s Engineering teams were more familiar with SQL-based BI tools, Aurora Data Science started exploring tools with a modeling approach to mitigate the limitations of relying solely on SQL.

They strongly prefer the data modeling approach for its consistency and accuracy as it lets data teams define business logic in one central place, creating a single source of truth that can be reused across reports and dashboards. This consistency is especially critical for a company like Aurora, where data accuracy is often mission-critical. With this insight, the team evaluated both SQL and modeling BI tools.

Your Analytics Guidebook really resonated with us, especially the part about data modeling. It was clear that Holistics was thinking about BI in a future-facing way, which was exactly what we needed.

We were looking for tools that weren’t just good today but would continue to evolve. We needed something that was not only functional but also thoughtfully designed for our future needs.

We are dealing with vehicle data that impact drivers, and it’s critical that we onboard a tool that our product engineering team can trust for data accuracy and consistency.

– Juan Argote, Director of Aurora Data Science

The evaluation process that followed was thorough and demanding. Aurora conducted a comprehensive market scan including all the market leaders, before narrowing it down to 6 for an in-depth Proof of Concept. The shortlist was based on a detailed set of criteria, spanning 3 major categories and over 30 features.

Aurora’s first priority was seeking reassurance in the vendor’s stability, innovation, and community support—qualities that would indicate both staying power and a commitment to continuous improvement. With these factors in mind, they assessed each vendor on several key criteria:

• Community and Support System: Leaned towards vendors with a large, engaged customer base—a good sign that the product is continuously improving and responsive to user needs.
• Velocity of Product Development: Reviewed product roadmaps and release notes, to ensure that the tool was forward-looking, actively maintained, and keeping up with the latest industry demands.
• Business Maturity: Sought a vendor with staying power and reliability—established in the industry, profitable, and sustainable—to avoid risks associated with early-stage companies.

We wanted a tool that was not only powerful but also easy to learn and use. Our goal was to enable more self-service across the team.

– Juan Argote, Director of Aurora Data Science

Aurora wanted a BI tool that everyone could use - from business users to data analysts.

• Learning Curve: Ease of learning was key. Aurora’s team evaluated how quickly both tech-savvy and non-technical users could get comfortable with the tool, focusing on core functions like data exploration, report building, and dashboard setup.
• End-User Self-Service Experience: Aurora looked for a BI solution that would let end-users access insights on their own, reducing dependency on technical support and democratizing access to data-driven insights.

When evaluating self-service BI tools, Aurora prioritized the following features:

• Data and visualization discovery: A search function to easily locate existing visualizations and datasets.
• Dashboard interactivity: Options to filter and drill down into data for deeper insights.
• Dashboard export: The ability to export dashboards in CSV and PDF formats without data size limitations.

Aurora also looked into features that would enable the data team to work more efficiently, making it easier to build and share dashboards while minimizing errors. The team identified the following essential capabilities:

• Semantic Modeling layer: The ability to define reusable data definitions, manage them centrally, and expose curated datasets for self-service.
• Data Source Connectivity: Ability to connect seamlessly to databases like Athena, Redshift, Snowflake, and Postgres.
• Reusable Objects: Options to create reusable elements such as views, charts, and SQL snippets.
• Meaningful Error Reporting: Clear and informative error messages for syntax issues to aid troubleshooting.
• Data Materialization:Capability to materialize logical views and store them in a “cache” for faster access.

Performance was a key criterion in Aurora’s evaluation. They prefer vendors that use the intelligent caching systems of BI tools like Looker and Holistics, which effectively reduce load times and cut query costs, were highly desirable.

Additionally, Holistics’ Job Monitoring feature lets Aurora track detailed query performance statistics directly within the app was seen as a very relevant feature.

Since adopting Holistics, the Aurora team has been pleased to see the platform continuously proving its value as the right choice by introducing additional performance-enhancing features, including:

• Aggregate Awareness: Automatically selects the most optimal aggregated tables for each query, maximizing query performance.
• Holistics Canal: A high-speed query streaming engine that offers enhanced flexibility and scalability for querying operations.

Rolling out a BI tool to ~1,000 users meant Aurora needed strong administrative features to maintain data integrity and security at scale, ensuring that data is accessed and used as intended. Role-based access control was a must, along with a few key capabilities:

• Version Control & Git Integration: With so many users, Aurora needed solid version control to track changes, collaborate, and keep things organized. Git integration was a must, fitting right into their engineering-driven workflows.
• Okta Integration for SSO: Okta integration for SSO made it easy to onboard people quickly and securely.
• User Activity & Usage Insights: This helps the team have a bird-eye view of dashboard usage & user activity, making it a lot easier to understand how your dashboards are being used.

A range of different chart types like maps, charts, trendlines, and reference lines were non-negotiables for Aurora’s diverse needs. Given the nature of Aurora Innovation’s business whose mission is to deliver the benefits of self-driving safely, quickly, and broadly, maps are extremely important.

The geographical mapping capabilities of Holistics provided the right mix of detail and variety to meet those needs.

Aurora wanted users to set up alerts so that whenever key metrics hit certain thresholds, a notification goes straight to the right platform. This ensures the team catches issues early and responds quickly.

The evaluation period lasted 3 months, and Holistics emerged as the top choice across all evaluation criteria, with the team especially impressed by how quickly they could get up and running with the platform. The transparent, scalable pricing model further reinforced their confidence, making Holistics the clear choice moving forward.

We built something in Holistics in half a day, without going through a sales channel or setting up meetings. The documentation was brilliant—we could just read the documents and get started.

Pricing was a key component. We needed something that was clear and scalable as we grow and Holistics’ transparent pricing made it easier to plan for the future without worrying about blowing our budget.

– Juan Argote, Director of Aurora Data Science

One of the standout features of Holistics that benefited the Aurora team was its reusability capabilities.

As an engineering-driven organization, Aurora’s data team wants to apply software best practices in their analytics workflows. With Holistics AML Extend, they efficiently manage and reuse datasets, calculations, and visualizations, saving time and ensuring consistency across the organization.

Our users needed more insights and different levels of data aggregation, which Holistics was able to deliver.

With this (Holistics AML Extend), we can define constants and functions that could be reused across different models and reports, significantly reducing redundancy and making our codebase much more maintainable.

– Stephen Lee, Staff Data Scientist at Aurora

After adopting Holistics, Aurora quickly saw the value in cross-filtering and drill-through features—tools they hadn’t prioritized but now find super useful.

Cross-filtering gave the product team more control over behavior analysis across thousands of tickets and various attributes. This feature lets them use any dashboard element to act as a filter across all reports, updating connected data points in real-time. This made spotting trends and understanding data relationships easier at a glance.

Drill-through functionality also proved invaluable, enabling the team to dive from high-level overviews into detailed insights, making data exploration more dynamic and flexible.

The drill-through functionality has been very powerful. It takes a bit to get used to, but once you start using it, it provides a ton of flexibility.

– Juan Argote, Director of Aurora Data Science

When Holistics was first implemented, many hesitated as SQL had been the go-to method for handling data queries and dashboard creation. The immediate benefits of a data-modeling BI tool, where metrics are defined centrally, weren’t always clear at first.

Adopting a data-modeling tool like Holistics required a shift in mindset.

While building foundational data models takes effort upfront, the benefits are clear: it streamlines workflows, ensures consistency, and makes data more manageable over time.

– Juan Argote, Director of Aurora Data Science

As time passed and more teams became familiar with the platform, the true value of Holistics became clearer. The ability to centrally define and maintain metrics created a trusted, golden source of truth - something that wasn’t easily achievable with SQL-based tools.

Instead of individual teams building ad-hoc solutions that risked inconsistencies, Holistics provided a centralized, consistent framework for metrics.

For folks who had experience with data modeling and BI tools, the usability and features Holistics provided were well-received. Internally, it took us some time to learn the tool, but it became clear it was an improvement over what we had before.

I love that the software design of Holistics forces us to think about data logic upfront, creating that single source of truth rather than piecing things together on the fly with ad-hoc dashboards that break easily over time.

– Juan Argote, Director of Aurora Data Science

At Aurora, their Data Science and Data Platforms teams played a championing role in promoting a self-service strategy. Recognizing that reliance on engineering for every data request would slow down decision-making, they spearheaded a shift toward empowering teams to handle their own data needs.

“What we’ve learned is that you need two things: team members who are interested and motivated to take on self-service, and top-down leadership to guide and support the initiative,” Juan explained.

By securing leadership buy-in and working closely with different departments, Aurora Data Science and Data Platforms helped lay the groundwork for the widespread adoption of a self-service model using Holistics.

As a company, we’re moving from a more development-oriented approach to a product-oriented one, and that requires seamless ways to present performance metrics. That shift drove us toward democratizing data access and adopting a self-service model.

On the project side, there were initiatives like understanding our on-road performance. We pulled data from different sources and surfaced it through visualizations in Holistics, which helped show the value.

– Juan Argote, Director of Aurora Data Science

As the strategy took root, more teams across Aurora began to embrace Holistics as the one-stop shop for well-vetted metrics with the gold standard for data.

Starting with engineering and finance, all departments quickly saw the value in being able to create, modify, and explore dashboards without waiting for ad-hoc engineering support.

Our finance team is fully onboarded and has been able to do a lot of self-service. They’re now able to do a lot of their own data analysis themselves. They can easily combine metrics and pull relevant data from different sources to build what they need.

Other teams are starting, and I think it’s just a matter of time before it becomes more widely adopted. It’s been great to see that kind of progress.

– Juan Argote, Director of Aurora Data Science

Now, more team members at Aurora have embraced the platform.

For instance, Alyssa T., a Product Manager focused on metrics and performance improvements, started by exploring existing dashboards, and then curated and filtered data to suit her needs. As her experience grew, she began creating reports directly from datasets.

This shift reflects a broader trend at Aurora, where more team members feel confident to explore and build dashboards themselves. This creates a flywheel effect, encouraging a culture of accessible insights and data-driven decision-making that continues to gain momentum across the organization.

There’s a bit of a flywheel effect happening. As more people use Holistics and see the benefits, it’s encouraging others to get on board.

We’re now better at understanding our on-road performance and capturing financial metrics dynamically. There’s been a lot of data projects around getting better at understanding how our product performs from a customer perspective.

– Juan Argote, Director of Aurora Data Science