Each title resolves to a single canonical record that corresponds to one specific work. A remake is a distinct canonical record from the original. Regional variants (e.g., a title packaged differently for a specific market) are preserved as variants beneath the canonical record. Multi-part releases (e.g., a film split into parts for streaming) are modeled as separate episodes or separate titles, depending on how providers and industry standards classify them.

The model is trained specifically on edge cases like these, using ground-truth data from 100M+ consumer interactions to validate ambiguous matches before they enter the canonical catalog.

Median API response time is under 100ms for single-title lookups. Bulk queries (availability across multiple services for a set of titles) scale with the query size but remain well under 500ms for typical request patterns. The API is served from Cloudflare's global edge network, so latency varies by region; US and European partners see the lowest numbers.

The rate limit is 50 calls per second on standard plans. Higher limits are available for production use cases that require burst capacity.

S3 exports are full catalog rewrites delivered up to twice daily. Every file in the bucket is replaced with the latest version of the complete dataset. No partial or delta exports. This means your pipeline always works from a clean, authoritative snapshot rather than trying to reconcile a stream of changes against a prior state.

The full US subscription is approximately 1.8GB uncompressed. Files are delivered in line-delimited JSON by default, with CSV, Parquet, and ORC available depending on your contract. For teams that prefer event-driven updates, the REST API with webhook notifications is the right fit.

Reelgood maintains one canonical show-season-episode hierarchy per title. When providers disagree on season structure (e.g., a show packaged as two seasons in one market and three in another), Reelgood resolves to one canonical structure while preserving all market-specific variants as additional records beneath it.

This means your downstream systems always receive a consistent, deduplicated representation of any series, regardless of how it's packaged across services or regions. The canonical ID for any episode remains stable even when provider packaging changes.

Yes. Reelgood's onboarding process maps your existing catalog to our canonical ID framework using metadata signals. No shared external identifier required. The result is a lookup table that maps each of your internal IDs to the corresponding Reelgood canonical ID.

Going forward, every record returned by the API or included in S3 exports contains both the Reelgood canonical ID and any third-party IDs (EIDR, streaming service-specific IDs) that exist for that title. You can use whichever identifier your systems prefer.

The canonical record persists. The canonical ID never changes and the title remains in the metadata catalog indefinitely. The availability record is updated to reflect that the title is no longer streaming on any active service, with the window close date and the last known service recorded.

Historical availability data, including the full movement history across all previous windows, remains accessible. This is particularly useful for rights and licensing teams tracking windowing patterns over time.

The partner API uses API key authentication passed as a request header. Keys are issued per integration and can be scoped to specific data endpoints or rate limit tiers. Key rotation is supported without downtime.

All API traffic is encrypted in transit via TLS 1.2 or higher. IP allowlisting is available for enterprise accounts that require it.

Current streaming availability data is refreshed continuously and available within 5 minutes of ingestion. This applies to real-time availability queries via the API and to the current state of the catalog in S3 exports.

Historical availability data covers 7+ years of records with exact window open and close dates. Historical records are immutable once written. A title that left a service in 2019 retains that exact record indefinitely. New historical data (records from newly onboarded services or markets) can be backfilled on request.