A collection is a folder of saved requests that share configuration. Once an endpoint works, saving it means you never rebuild the headers by hand again.
Group by service, not by ticket
The organising principle that survives contact with a real project is one collection per API, with folders inside it per resource:
Billing API
├── Customers
│ ├── List customers
│ ├── Create customer
│ └── Cancel subscription
└── Invoices
├── List invoices
└── Void invoice
Collections named after tickets rot within a sprint. Collections named after services stay useful for years.
Set the base URL once
Put the base URL and authentication on the collection rather than on each request. Then switching between staging and production is one edit instead of thirty, and a rotated token is updated in a single place.
Import from OpenAPI
If the API publishes an OpenAPI 3.x document, import it. You get the requests and the response schemas in one step, which is considerably faster than clicking through documentation.
Sync and share
Sign in and collections sync across your machines. Team workspaces extend that to other people, with roles controlling who can edit. Everything stays free — there is no seat pricing to run into.
If you have not saved anything yet, start with how to test a REST API and save the request at the end of it.