dbt Analytics Engineering Certification Practice Test – 19 Free Exam Questions with Answers

Correct answer: D. The mock input data that replaces the actual ref() and source() inputs during testing

The given block defines mock data that replaces actual ref and source inputs for isolated testing. The expect block is the main distractor, but it defines the expected output rows rather than the inputs.

Correct answer: D. manifest.json

The manifest file provides a complete project representation needed for state comparison. Run results are the main distractor, but they only contain execution outcomes rather than the full project structure required for diffing.

Correct answer: A. dbt run –select fct_reviews –defer –state ./prod-artifacts

Combining defer with the state flag resolves upstream refs to production objects defined in the artifacts. Option D is a strong distractor, but the plus symbol forces local builds of those upstream models.

Correct answer: D. A YAML declaration of a downstream consumer of dbt models, such as a dashboard or ML model

Exposures are yaml declarations defining downstream consumers of your models, like dashboards. The main distractor refers to public models, which is a project governance feature rather than a lineage documentation tool.

Correct answer: B. A reusable Jinja code block defined in the macros/ directory, callable from models, tests, or other macros

Macros are reusable Jinja code blocks defined in the macros directory that accept arguments and can be called from models or tests. Do not confuse them with stored procedures; macros compile into your SQL rather than existing in the warehouse.

Correct answer: C. private, protected, public

The three access levels are private, protected, and public. Protected is the default, allowing references within the same project. Public allows cross-project references, while private restricts access exclusively to models in the same group.

Correct answer: A. Database Error

A Database Error occurs when Jinja compilation succeeds but the warehouse rejects the SQL. Use this to distinguish from Compilation Errors, which happen locally when Jinja fails to render before hitting the warehouse.

Correct answer: D. compiled/ contains the pure SELECT with Jinja resolved; run/ contains the full DDL-wrapped SQL sent to the database

The compiled directory contains pure SELECT statements with Jinja resolved, useful for debugging. The run directory contains the full wrapped SQL sent to the warehouse, including DDL commands like create table, representing actual execution.

Correct answer: C. 0 6 * * 1-5

The cron expression 0 6 star star 1-5 means minute zero, hour six, and days Monday through Friday. Remember the five field order for scheduling deployments in dbt Cloud jobs.

Correct answer: D. It tells dbt to use production database objects for any ref() not being built in the current run

The defer flag resolves ref calls for unselected models to their production objects. Skipping unchanged models is the main distractor, but that behavior is handled by the state modified selector.

Correct answer: C. The 1 failed model and the 15 skipped models (16 total), excluding the 30 successes

The retry command re-executes any nodes that did not succeed, including both failed and skipped models. The main distractor is option B, which incorrectly assumes skipped models are ignored.

Correct answer: B. env_var() reads system environment variables and works in dbt_project.yml; var() reads from vars and only works in Jinja

The env_var function reads system environment variables and works in dbt_project.yml and profiles.yml. Remember that var reads project variables but is restricted to Jinja contexts, unlike env_var which can securely configure environments outside of models.

Correct answer: A. To assign models to a logical grouping that restricts cross-group references based on the access config

Groups organize models into logical collections, often representing team ownership or business domains. Combined with the access config, groups restrict cross-group references, preventing private models from being referenced outside their group.

Correct answer: A. The table name listings is missing quotes — it should be {{ source('airbnb', 'listings') }} with both arguments quoted

The table name lacks quotes, causing Jinja to evaluate it as an undefined variable. Always pass string literals to functions like source, wrapping arguments in quotes to prevent compilation errors during execution.

Correct answer: A. dbt build runs models and tests in DAG order, testing each model immediately after it is built

The build command executes nodes in DAG order, interleaving models and tests so tests run immediately after their dependencies. It prevents downstream execution if tests fail, unlike running sequentially without fail-fast behavior.

Correct answer: A. Set severity: warn on the test configuration

Setting severity to warn allows the test to fail without returning a failure exit code. The run-level warn-only flag is a distractor because it applies globally to all tests rather than configuring a single test.

Correct answer: A. dbt test –select test_type:unit

The test_type:unit selector filters the dbt test command to execute only unit tests. Remember that dbt unit tests are integrated into the standard test command rather than having a dedicated standalone command.

Correct answer: B. Every upstream model: dim_listings_w_hosts, mart_fullmoon_reviews, and their ancestors

The plus prefix traverses the entire upstream dependency graph to build all ancestors. Option A is a strong distractor, but it only selects the direct parents and misses the indirect upstream dependencies.

Correct answer: C. No warehouse object is created — dbt injects the model's SQL as a CTE into the downstream model at compile time

Ephemeral models do not create physical tables or views in the warehouse. Instead, dbt compiles the model code into a Common Table Expression within downstream models. Use this materialization to avoid cluttering your schema with staging tables.