Platform Architecture · dbt

dbt makes your models testable. It does not decide what "revenue" means.

dbt consulting, at the signal layer, is about definitions and guarantees. dbt lets you assert a model's shape and content: contracts enforce column names and data types at build time, and tests return the rows that disprove an assertion. What it cannot do is agree on which revenue, which user, or which session your business means. We govern that layer so the numbers in your warehouse match the CRM and finance.

Start with an Assessment →Not sure where to start?

01Where it breaks

Where dbt projects break.

These are the patterns that show up across dbt projects where the build is green but the business definitions underneath it were never agreed.

02How we engage

What we do on a dbt engagement.

Metric definitions first

Agree revenue, user, session, and attribution windows with finance and marketing, then encode them once.

Contracts and tests

Enforced contracts on the models other teams consume; generic tests for integrity and singular tests for the business rules generic tests cannot express.

Reconciliation

Warehouse revenue reconciled to the CRM and the system of record with a documented tolerance. See the BigQuery and Snowflake warehouse truth layer, and our article on dbt for marketing attribution.

Upstream signal

Where event collection feeds the models: server-side GTM or RudderStack.

What does dbt model governance involve?

Enforced contracts on the models other teams consume, generic tests for integrity, singular tests for the business rules generic tests cannot express, and a named owner for every model that feeds a reported number.

03Who works on this

The team behind this work.

Valentina Borisovna, Lead Experimentation and Data Visualization Analyst at Analytico

Valentina Chivikova

Analytics Data Engineer · Calgary, Alberta

Valentina builds the pipelines behind Analytico's measurement work, moving event and business data into the warehouse in a form teams can model, test, and trust.

04Questions

dbt — frequently asked questions

What is a dbt model contract?

A set of guarantees about a model's returned columns, data types, and optional constraints; dbt fails the build if the model violates it. (Source: dbt docs, model contracts.)

What is the difference between a contract and a test?

Contracts define shape at build time; tests validate content after the build. (Source: dbt docs, model contracts.)

Which generic tests does dbt include?

unique, not_null, accepted_values, and relationships. (Source: dbt docs, data tests.)

Do we need dbt for marketing attribution?

Not always. dbt is the tool many teams use to build attribution and revenue models in the warehouse, and its value is the tested, versioned definitions. The decision that matters more is which definition of revenue and which attribution window the business agrees on; we start there.

How does an engagement start?

Start with the Measurement Architecture Assessment: a scoped diagnostic conducted by a senior principal. It begins with a 30-minute conversation, runs three to five weeks, and ends with a document you own.

Send us the three models your finance team trusts least.

We will show you which definition each one is missing.

Start here

The Assessment is calibrated to your dbt project and warehouse — not a generic analytics engineering review.