Why sources define raw data contracts in dbt - Challenge Your Understanding

Challenge - 5 Problems

🧠 Conceptual

intermediate

Purpose of Raw Data Contracts in Sources

Why do data teams define raw data contracts when setting up sources in dbt?

A To replace the need for data testing in later stages

B To speed up the data loading process by skipping validations

C To automatically generate visualizations from raw data

D To ensure the incoming data meets expected formats and quality before transformations

💡 Hint

Think about how contracts help keep data reliable and consistent.

Raw data contracts define expectations on the source data's structure and quality. This helps catch issues early and ensures transformations work with clean, predictable data.

❓ Predict Output

intermediate

Output of dbt Source Freshness Check

Given this dbt source freshness configuration, what will be the output status if the source data was last updated 3 hours ago?

sources:
  - name: sales_data
    freshness:
      warn_after:
        count: 2
        period: hour
      error_after:
        count: 4
        period: hour

A Status: warn (data is older than 2 hours but less than 4 hours)

B Status: pass (data is fresh within 2 hours)

C Status: error (data is older than 4 hours)

D Status: unknown (freshness not configured properly)

💡 Hint

Compare the last update time with the warn and error thresholds.

The data was last updated 3 hours ago, which is more than the warn_after threshold of 2 hours but less than the error_after threshold of 4 hours, so the status is warn.

❓ data_output

advanced

Result of Source Schema Validation

Consider a source defined with a raw data contract expecting columns id (integer) and date (date). If the actual source data has an extra column status (string), what will be the result of the schema validation in dbt?

A Validation fails because <code>status</code> column type is incorrect

B Validation fails due to unexpected extra column

C Validation passes because extra columns are allowed by default

D Validation passes only if <code>status</code> column is nullable

💡 Hint

Think about whether dbt schema tests reject extra columns by default.

dbt source schema validations check for required columns and their types but do not fail if extra columns exist. Extra columns are ignored unless explicitly tested.

🔧 Debug

advanced

Debugging a Failed Source Contract Test

A dbt source test for a raw data contract fails with the error: Column 'user_id' contains null values. The contract expects user_id to be non-nullable. Which option correctly fixes the issue?

A Remove the <code>user_id</code> column from the contract

B Update the source data to remove or fill nulls in <code>user_id</code>

C Change the contract to allow <code>user_id</code> to be nullable

D Ignore the test failure and proceed with transformations

💡 Hint

Consider data quality versus contract expectations.

The contract expects user_id to never be null. The correct fix is to clean the source data to meet this expectation, not to weaken the contract or ignore the problem.

🚀 Application

expert

Designing a Raw Data Contract for a New Source

You are adding a new source table orders to your dbt project. The raw data contract must ensure the following:

• order_id is unique and non-null
• order_date is non-null and recent (within last 30 days)
• customer_id is non-null

Which dbt source test configurations correctly enforce these rules?

A Unique test on <code>order_id</code>, not_null tests on <code>order_id</code>, <code>order_date</code>, <code>customer_id</code>, and a custom freshness test on <code>order_date</code>

B Unique test on <code>order_id</code> and <code>customer_id</code>, no freshness test

C Not_null tests on all columns and a unique test on <code>customer_id</code>

D Only a freshness test on <code>order_date</code> and not_null on <code>order_id</code>

💡 Hint

Think about which tests enforce uniqueness, non-null, and freshness.

Uniqueness and non-null tests ensure data integrity for IDs. Freshness tests check if dates are recent. Only option A covers all requirements correctly.