What are the two Dataset API endpoints?
The Marketplace Dataset API exposes two ways to pull records, depending on the job:- Search (sync) is Elasticsearch-powered and returns records inline in sub-second responses. Use Search for real-time lookups, lead enrichment and sampling.
- Filter (async) is snapshot-based and supports CSV/JSON file uploads and large payloads. Use Filter for bulk exports and value-list filtering.
Search dataset (sync)
Elasticsearch query, results inline, sub-second latency. Up to 1,000 records per call.
Filter dataset (async)
Snapshot-based job, CSV/JSON uploads, payloads up to 200 MiB. All 250+ datasets.
Which endpoint should I use?
Use Search for low-latency, single-call lookups and Filter for bulk or file-driven jobs.How do the two endpoints compare?
The Search and Filter endpoints accept the same filter object but differ in engine, latency and dataset coverage.
Results between the two engines can differ. Elasticsearch tokenizes text differently from the snapshot engine.
How do I authenticate?
Both endpoints use Bearer token authentication. Pass your API key in theAuthorization header:
How much does the Dataset API cost?
Both endpoints cost $2.5 CPM (per 1,000 records returned), the same rate as the Marketplace. There is no premium for real-time Search, and there is no charge when a filter returns 0 records.Where is the filter syntax documented?
Both endpoints accept the samefilter object, with the same operators and nesting rules. The filter syntax reference documents the operator list, filter groups, up to three levels of nesting and CSV/JSON file references.
Related
- Search dataset (sync)
- Filter dataset (async, file uploads)
- Filter syntax reference
- Get dataset metadata
- Get dataset list
- To build or refresh a dataset before querying, see Request a collection.