From fragmented metric definitions to a governed dbt layer the Clever team owns.
Astrodata audited and rebuilt Clever’s dbt transformation layer, standardized metric definitions across Finance, Sales, and Marketing, and delivered a fully documented reporting suite in Sigma with a structured handoff to the internal team.
The situation
Clever provides schools, students, and educators with a unified identity for secure, seamless access to learning tools. Their data team had a modern stack in place: dbt, Sigma, and Snowflake. The issue was not tooling. The issue was that company-wide metrics were built by operations managers in Salesforce and Google Sheets instead of the data warehouse, because of the analytics team’s limited capacity. Requests for more specific metrics and reporting grains couldn’t be met in Salesforce.
Clever’s analytics team needed to supplement their capacity with experienced analytics engineers to work with their team, inside the existing stack, to migrate Salesforce reports, create inter-operable analytics tables, and establish a definition layer that all reporting surfaces could rely on.
Why this engagement mattered
Inflexible reporting and reporting limitations are a common byproduct of Salesforce dashboards and dbt projects that grow without centralized ownership. Calculations accumulate across marts, dashboards, and ad hoc queries without a unifying standard. When different teams work in silos, the problem is rarely the data itself. It is the absence of a canonical definition layer that all consumers reference.
What a governed metrics layer does beyond resolving technical inconsistency is create a structured surface for business alignment. When definitions are anchored in a single upstream layer that is versioned, tested, and documented, it becomes possible for teams such as Finance, Sales, and Marketing to have a concrete artifact to align around. For metrics tied to company health reporting and individual performance, that kind of shared foundation is often what makes alignment possible in the first place.
Resolving this requires more than refactoring SQL. It requires stakeholder alignment on what metrics mean, disciplined upstream implementation in dbt, and documentation thorough enough that the logic survives personnel changes. Skipping any of those steps produces a cleaner codebase that still generates disagreement at the dashboard level.
Astrodata’s approach prioritizes definition alignment before model development, so that what gets built reflects what the business has actually agreed on.
Our approach
We worked with Clever’s data and business teams across three workstreams.
dbt model governance and metric standardization
Working directly inside Clever’s existing dbt project, the Astrodata team worked with analysts and operations managers to find Salesforce and Sigma reports, identified sources of metric inconsistency, and rebuilt the transformation logic around agreed-upon business definitions.
- Medallion architecture enforced end-to-end. Raw, staging, and production layers clearly separated, with consistent naming conventions across all models.
- Metric definitions anchored in dbt. Calculations previously distributed across dashboards, spreadsheets, or individual analysts’ queries were moved upstream, versioned, and documented.
- CI/CD implemented in dbt Cloud. Automated testing and lineage visualization so the team could identify dependencies and catch regressions before they reached downstream consumers.
- Community packages and standard patterns applied where appropriate, with Clever-specific business logic layered on top to keep the codebase both idiomatic and precise.
Sigma dashboard buildout
With a governed dbt layer in place, the Astrodata team built out Clever’s reporting suite in Sigma, covering Finance, Sales, and Marketing stakeholders. Complex business logic was moved to dbt. Sigma was used for last-mile calculations, filters, and interactivity. Metric definitions were not replicated at the dashboard layer.
Documentation and knowledge transfer
Documentation was treated as a first-class deliverable throughout the engagement, not a final step.
- Inline code comments throughout the dbt project explaining logic, edge cases, and intent.
- Standalone documentation covering the metrics catalog, data model structure, and dashboard suite.
- Co-development sessions with Clever’s technical team to review the codebase, validate consistency with Clever’s existing conventions, and answer questions prior to handoff.
The outcome
After this project, Clever’s metric definitions were unified in a single governed dbt layer, with Finance, Sales, and Marketing pulling from the same source. The Sigma dashboards were built on top of that layer, so reporting surfaces reflected definitions the business had agreed on.
At handoff, the Clever team took full ownership of the codebase. The project was documented, tested, and structured consistently with Clever’s existing conventions, making it maintainable without ongoing Astrodata involvement.
| Metric trust & ownership | Before | After |
|---|---|---|
| Metric definitions | Distributed across dashboards, models, and spreadsheets | Unified in governed dbt models |
| Cross-team consistency | Finance, Sales, and Marketing often produced different answers | Single source of truth across all reporting surfaces |
| Documentation | Sparse, with logic residing in individuals' institutional knowledge | Inline and standalone docs covering the full metrics catalog |
| Team ownership | High dependency on original model authors | Fully transferable; Clever team operates independently |
| CI/CD | Manual | Automated testing and lineage tracking via dbt Cloud |
Why Astrodata
Three values in action.
- Curiosity
The Astrodata team audited the existing codebase before proposing changes, taking time to understand the original intent behind models before deciding what to modify.
- Integrity
Where metric definitions were ambiguous, Astrodata surfaced the ambiguity and worked with Clever's business stakeholders to reach a documented resolution rather than making unilateral decisions.
- Sustainability
Every implementation decision accounted for long-term maintainability. The goal was a team that could operate the system independently, and the handoff reflected that.
“When I joined Clever to lead a newly centralized business analytics function, we had a tiny team and a lot of work to do. I quickly realized that we’d need to augment our staff with seasoned analytics engineers to achieve our ambitious annual goals on time. We couldn’t have done it without the Astrodata team. They ramped quickly, operated mostly independently, and delivered quality work!”
Industries, technologies, and capabilities
Ready for a metrics layer your team actually owns?
If your metric definitions are scattered across dashboards and spreadsheets, or your dbt project has outgrown its governance — we’d like to talk.
Request a consultation