Delivery and integration
Data feed
Also called: Feed
A recurring delivery of data from a source to a destination on a schedule.
In practice
A feed is a promise about freshness and format. Treat changes to it like API changes.
What makes a feed operational
A useful feed specifies more than a file format. Its contract describes fields and types, record identity, delivery cadence, destination and the meaning of missing or changed records. It should also distinguish a full snapshot from an incremental update. A consumer needs to know whether an absent row means deletion, collection failure or simply no update in this batch.
Example: a daily product feed
An illustrative daily feed might include product ID, source URL, price, currency, availability and observation time. A batch identifier allows a warehouse to load the same delivery again without duplicating records. A manifest can describe the expected record count and schema version. Keep collection time separate from delivery time: a file delivered this morning may still contain observations collected yesterday.
Checks before ingestion
Validate required fields, uniqueness, types and freshness before making a batch available to downstream users. Decide how to handle partial deliveries and retries, and retain a path for reprocessing corrected data. Monitor late or missing batches separately from valid empty results. These controls turn a recurring export into a dependable input for dashboards, applications and model workflows.
Import.io delivers CSV, JSON or Parquet to files, warehouses, SFTP, APIs and webhooks, with a manifest on every managed delivery.
Delivery options →