Dashboards as Products: Spotify’s Framework for Quality at Scale

Dashboards are a core part of how Spotify stays data-informed. In 2023 alone, more than 4,900 dashboards were created in Tableau or Looker Studio, and over 6,000 Spotifiers used them. There is no central team building these; instead, a free market of producers and consumers operates on scaffolding provided by the Analytics Platform team.

The platform strategy relies on two complementary tools. Tableau is used for highly customized, detailed dashboards that function as full-fledged internal products with distinct user bases. Looker Studio serves as the quick option, offering seamless integration with Google products like BigQuery, letting users move from SQL to a finished dashboard quickly. Guided by internal “Golden Path” documentation, individual teams choose the tool that fits their specific needs.

Defining Dashboard Quality: Vital Signs and Design Checks

To ensure widely shared dashboards are interpreted correctly, Spotify developed a two-part Dashboard Quality Framework.

The first part, Vital Signs, consists of automatic checks that confirm a dashboard is “alive and well.” Leveraging API and log data, these checks monitor owner status, descriptions, use of LDAP groups for sharing, the lifecycle of underlying data endpoints, and the recency of updates and extract refreshes.

The second part is the manual Spicy Dashboard Design Checklist. This evaluates visual design, usability, insights, and trust—elements that require a human eye. It is implemented as a self-evaluating checklist that flags a dashboard once successfully completed.

Combining these measurements results in one of three quality labels:

  • Low — fails Vital Signs (regardless of the design checklist outcome)
  • High — passes Vital Signs
  • Golden — passes Vital Signs and the Spicy Design Checklist

Currently, 10% of eligible dashboards have achieved Golden status. The platform team also aims to keep more than 80% of all dashboards at High or above, notifying owners when their dashboards fall short of the criteria.

Supporting Producers and Managing Extracts

For Tableau users, Spotify built a service to streamline extract management. A SQL scheduling tool, already used for batch workflows in BigQuery, was enhanced to produce and publish Tableau .hyper files for Tableau Cloud. These workflows run when upstream dependencies are ready, are often faster, and handle larger data loads. They are also easier to edit since the SQL is stored in .yaml files rather than embedded in Tableau.

Reinforcing the idea that dashboards are products, owners receive usage statistics for their dashboards—active users, weekly retention, and user breakdowns by org or job family. This data allows owners to email users about updates or deprecations, consider marketing to new audiences, or retire dashboards with very low usage.

Centralizing Consumption Through Dashboard Portal

Discovery is handled by Dashboard Portal, an internal, searchable catalog of all published dashboards from both Tableau and Looker Studio. Its primary features include:

  • Search — by title, description, and field names.
  • Curation — authors can build hub collections for their teams, such as an “Ads Dashboard Hub.”
  • Trust and Quality — quality labels appear as badges on thumbnails and pages. Users are more likely to view dashboards with a Golden label, allowing the platform to spotlight the highest-quality work.
  • Ubiquity — a single portal removes the need to know which tool hosts a dashboard.
  • Enhanced Context — dashboards are embedded directly with owner info, last refresh date, usage stats, and underlying SQL queries.

This combined approach—distributed creation, clear quality standards, and centralized discovery—treats dashboards not as static reports but as managed products delivering actionable insights across the organization.