Let your data
keep its shape.

Choose structure for the receiving application

JSON is useful when the output will be read by software and records have nested properties or arrays. An export can carry metadata beside the records, making dataset identity, currency, and schema context explicit.

The downloadable sample contains a dataset label, a currency field, and an array of three customer records. Each customer has the same four documented fields as the CSV, XLSX, and PDF samples.

Agree on values, not just syntax

A valid JSON document can still be an ambiguous export. Decide how you represent dates, money, identifiers, unavailable values, and empty arrays. Document which properties may be absent. Use strings for identifiers when arithmetic is not meaningful, and agree on numeric handling with the receiving runtime.

The API patterns guide shows how to make these choices part of a stable contract. The validation guide explains how to check a completed payload.

Plan how the consumer reads it

A single array is straightforward for small datasets, but a large document can require significant memory if both ends load it entirely. Choose a reading and generation strategy that fits the workload. For longer exports, separate the job from the download and publish completion only after the file is ready.