dbt · Extra High-Value

What are dbt seeds?

Easyseedscsvreflookup

Interview question

What are dbt seeds, and when should you use them?

How to answer

The short answer

Seeds are CSV files in your dbt project (usually under seeds/) that dbt loads into the warehouse as tables with dbt seed.

Example

seeds/country_codes.csv:

Output
country_code,country_name
US,United States
IN,India
DE,Germany
Shell
dbt seed
# then in a model:
select * from {{ ref('country_codes') }}

Yes: seeds are referenced with ref(), like models.

Good use cases

  • Small mapping / lookup tables maintained by analysts in git
  • Static status code descriptions
  • Tiny seed lists for accepted product categories in lower environments

Bad use cases

  • Large fact histories (use EL tools / warehouse loads)
  • Frequently changing high-volume data
  • Secrets or dumps in CSVs

Mental model

Configure seeds in dbt_project.yml (column types, schema) when defaults are wrong.

Interview tip: "Seeds are version-controlled micro-dimensions. If it needs an EL pipeline, it is not a seed."