Why Seeds for static reference data in dbt? - Purpose & Use Cases

The Big Idea

What if you never had to copy-paste static data again and could trust it was always correct?

The Scenario

Imagine you have a list of country codes and names stored in a spreadsheet. Every time you build your data models, you manually copy and paste this list into your queries or scripts.

It feels like a small task, but it happens over and over, and you worry about typos or outdated info.

The Problem

Manually copying static reference data is slow and error-prone. You might forget to update the list, causing wrong results.

It also makes your data models messy and hard to maintain because the same data is repeated in many places.

The Solution

Using seeds in dbt lets you store static reference data as CSV files inside your project.

dbt automatically loads this data as tables you can join with your models, keeping everything clean, consistent, and easy to update.

Before vs After

✗ Before

SELECT * FROM sales JOIN (VALUES ('US', 'United States'), ('CA', 'Canada')) AS countries(code, name) ON sales.country_code = countries.code

✓ After

SELECT * FROM sales JOIN {{ ref('countries_seed') }} AS countries_seed ON sales.country_code = countries_seed.code

What It Enables

You can easily manage and reuse static reference data across your entire project without duplication or errors.

Real Life Example

A retail company uses seeds to store product categories and tax rates as static data, ensuring all sales reports use the same consistent info.

Key Takeaways

Manual copying of static data is slow and risky.

Seeds let you store static data as CSV files inside dbt projects.

This keeps your data models clean, consistent, and easy to maintain.

Practice

1. What is the main purpose of using seeds in dbt?

easy

A. To create dynamic tables based on SQL queries

B. To load static reference data from CSV files into your database

C. To schedule dbt runs automatically

D. To write Python scripts for data transformation

Solution

1. Step 1: Understand what seeds are in dbt

   Seeds are CSV files that contain static reference data you want to load into your database.

2. Step 2: Identify the main use of seeds

   Seeds let you easily add fixed data tables without writing SQL queries.

3. Final Answer:

   To load static reference data from CSV files into your database -> Option B

4. Quick Check:

   Seeds = static CSV data load [OK]

Hint: Seeds = fixed CSV data loaded as tables [OK]

Common Mistakes:

• Confusing seeds with models that run SQL
• Thinking seeds schedule dbt runs
• Assuming seeds are for dynamic data

2. Which command do you run to load or refresh seed data in your database?

easy

A. dbt test

B. dbt run

C. dbt seed

D. dbt compile

Solution

1. Step 1: Recall dbt commands related to seeds

   The command dbt seed loads CSV seed files into the database as tables.

2. Step 2: Differentiate from other commands

   dbt run runs models, dbt test runs tests, and dbt compile compiles SQL but does not load seeds.

3. Final Answer:

   dbt seed -> Option C

4. Quick Check:

   Load seeds = dbt seed [OK]

Hint: Use 'dbt seed' to load CSV data tables [OK]

Common Mistakes:

• Using 'dbt run' to load seeds
• Confusing 'dbt test' with loading data
• Thinking 'dbt compile' loads data

3. Given a seed CSV file countries.csv with columns id and name, what will be the output of this dbt model SQL?

select * from {{ ref('countries') }}

medium

A. A table with all rows and columns from countries.csv

B. Only the id column from countries.csv

C. An empty table because seeds are not loaded automatically

D. An error because seeds cannot be referenced

Solution

1. Step 1: Understand how seeds are referenced in dbt

   Seeds become tables in the database and can be referenced using ref() like models.

2. Step 2: Predict the query output

   The query selects all columns and rows from the seed table countries, so it returns the full CSV data.

3. Final Answer:

   A table with all rows and columns from countries.csv -> Option A

4. Quick Check:

   ref(seed) = full seed table [OK]

Hint: ref(seed_name) returns full seed table [OK]

Common Mistakes:

• Thinking seeds cannot be referenced
• Assuming seeds load empty tables
• Expecting partial columns only

4. You ran dbt seed but your seed table did not update. Which of these is the most likely cause?

medium

A. You forgot to add the seed CSV file in the seeds folder

B. You ran dbt run instead of dbt seed

C. Your seed CSV file has syntax errors

D. You did not configure the seed in dbt_project.yml

Solution

1. Step 1: Check seed discovery mechanism

   dbt automatically discovers and loads CSV files from the seeds/ folder with dbt seed.

2. Step 2: Identify why table doesn't update

   If the CSV file is missing from the seeds/ folder, dbt seed runs successfully but skips that seed, leaving the table unchanged.

3. Final Answer:

   You forgot to add the seed CSV file in the seeds folder -> Option A

4. Quick Check:

   Seeds folder missing CSV = no update [OK]

Hint: Place seed CSVs in seeds/ folder for dbt seed [OK]

Common Mistakes:

• Thinking seeds require config in dbt_project.yml
• Confusing dbt run with dbt seed
• CSV syntax errors (would cause explicit failure)

5. You want to use a seed file currencies.csv with columns code and symbol inside a model to join with a transactions table on currency_code. Which is the correct way to write the join in your model SQL?

hard

A. select t.*, c.symbol from transactions t join currencies c on t.currency_code = c.code

B. select t.*, c.symbol from transactions t join currencies.csv c on t.currency_code = c.code

C. select t.*, c.symbol from transactions t join seed('currencies') c on t.currency_code = c.code

D. select t.*, c.symbol from transactions t join {{ ref('currencies') }} c on t.currency_code = c.code

Solution

1. Step 1: Recall how to reference seeds in dbt models

   Seeds are referenced using {{ ref('seed_name') }} to get the table name in SQL.

2. Step 2: Identify the correct join syntax

   Joining transactions with {{ ref('currencies') }} correctly uses the seed table in the join.

3. Final Answer:

   select t.*, c.symbol from transactions t join {{ ref('currencies') }} c on t.currency_code = c.code -> Option D

4. Quick Check:

   Join seed with ref() = correct [OK]

Hint: Use ref('seed_name') to join seed tables in models [OK]

Common Mistakes:

• Using raw CSV filename in SQL
• Forgetting to use ref() for seeds
• Trying to use a non-existent seed() function