What is partial success in GraphQL, and why does it matter for testing?

Because a GraphQL query can request many fields, each backed by its own resolver, some fields can succeed while others fail. The response then carries both data (with the successful fields, and null for the failed ones) and errors (describing the failures).

{
  "data": { "user": { "name": "Alice", "paymentMethod": null } },
  "errors": [
    { "message": "Forbidden", "path": ["user", "paymentMethod"], "extensions": { "code": "FORBIDDEN" } }
  ]
}

Testing implications:

• Don't write if (errors) fail() or if (data) pass() as exclusive branches — check both.
• A field returning null is not automatically a bug; cross-reference the errors array's path to see whether that null is an expected error or genuine data.
• Partial success is the mechanism behind subtle authz leaks and resolver failures that a naive "did I get data back?" check will miss.