Skip to content

django-data-seed

Point it at your Django project and get realistic, constraint-valid data at bulk-insert speed — with zero configuration.

pip install django-data-seed
python manage.py seeddata --count 20 --seed 42

What is database seeding?

Seeding is filling a database with data so you have something to work against — during development, in a demo, on a staging environment, or inside tests. The data isn't real, but it should look real and behave like real data: valid emails, believable prices, foreign keys that point at rows that exist, dates that come in a sensible order.

Why backend developers need it (and how it saves time)

Without a seeding tool, you end up doing one of these, over and over:

  • clicking through the Django admin to create a handful of rows by hand,
  • writing and maintaining fixture files or a factory per model,
  • copying a production dump (and dealing with the privacy headache that brings).

All of it is slow, and the result is usually junk — test test, a@a.com, every order in the same state — which makes demos look fake and lets bugs hide until real-shaped data shows up in production.

django-data-seed replaces all of that with one command. It reads your models, works out the safe insert order, generates realistic values, and bulk-inserts them — turning "an afternoon of fixtures" into a few seconds.

What is django-data-seed?

A Django app that seeds your whole project (or a single model) by introspection. You don't describe your data — it figures your schema out and generates data that fits it:

  • realistic values from Mimesis, refined by the field's name (city → a city) and constraints (max_length, choices, …);
  • foreign keys resolved through a dependency graph so parents exist before children, reusing rows instead of exploding a fresh tree per child;
  • rows that are internally coherent (ordered dates, one persona per row);
  • bulk_create speed, in-memory uniqueness, and a pre-flight check that refuses to run against a stale schema.

Why django-data-seed (vs the alternatives)

factory_boy model_bakery django-seed django-data-seed
Setup per model write a Factory none (junk values) none none
Realistic values only if you write them no Faker, no inference inference + locales
Whole-project seeding no no yes (unmaintained) yes
Bulk performance no no no bulk_create, batched
Maintained in 2026 yes yes no yes

factory_boy and model_bakery are great inside unit tests — but they make you write a factory per model. django-seed did whole-project seeding but is unmaintained. django-data-seed fills that gap: zero-config, realistic, whole-project, fast — and maintained.

Where to next