Hi Sarah!

While select_related handles foreign keys traversal, it doesn’t handle Many-to-Many relationships (M2M) or reverse lookups, which require prefetch_related.

If you’re debugging background tasks (like Celery workers for example), then you don’t have things like Django Debug Toolbar handy. I recommend logging query counts explicitly in your development environment or writing a test case that fails if the query count exceeds a threshold.

from django.test.utils import CaptureQueriesContext
from django.db import connection

def test_dashboard_query_count(self):
    # Set up complex data...

    with CaptureQueriesContext(connection) as ctx:
        response = self.client.get('/dashboard/')

    # Assert we don't exceed expected query count
    self.assertLess(len(ctx.captured_queries), 15)

from django.test.utils import CaptureQueriesContext
from django.db import connection

def test_dashboard_query_count(self):
    # Set up complex data...

    with CaptureQueriesContext(connection) as ctx:
        response = self.client.get('/dashboard/')

    # Assert we don't exceed expected query count
    self.assertLess(len(ctx.captured_queries), 15)

We do this so much we added a helper to django-test-plus specifically for it.

Another often overlooked tool is using the --print-sql flag if you are running scripts via management commands, or simply inspecting the connection.queries list in a shell session.

Key Takeaways

1. select_related for ForeignKey and OneToOneField (SQL JOINs)
2. prefetch_related for ManyToManyField and reverse ForeignKey (separate queries + Python joining)
3. Test your query counts - don’t guess, measure
4. Use django-debug-toolbar during development, but also have tests for production code paths

Hope this helps!