FAST operations · Operational blueprint
A FAST Catalogue Readiness Checklist
Most catalogue failures are ordinary: an identifier changes, a territory is missing, an episode is attached to the wrong season, or a rating arrives without its rating system. Content review belongs in the same readiness process, not in a separate spreadsheet discovered at launch.
Executive brief
What the operating team needs to know
A catalogue is ready when its records survive the receiving partner's validation and still make sense in the viewer experience. Passing an XML or JSON schema is only one gate. Identity, rights, episode order, artwork, ratings and the actual media asset must agree.
The practical operating model is a staged hand-off. Commercial teams define the delivery; metadata teams assemble the record; standards teams review content-sensitive gaps; engineering validates the package; and one named owner resolves rejects. This prevents the familiar launch pattern where every team has a green spreadsheet but the partner still cannot publish the title.
Higher first-pass acceptance
Catch structural and editorial defects before the partner's ingest queue does.
Clearer launch ownership
Route each rejection to the team that can actually correct the source record.
Reusable partner mappings
Keep one governed catalogue record and translate it into partner-specific delivery shapes.
Working model
Principles before process
Use these principles to shape the record, then follow the blueprint below for the hand-off.
Treat identity and availability as foundations
Use stable internal identifiers and preserve the identifiers shared with each partner. Availability should identify the channel, territory, start and end times, and whether the event is live. Validate those relationships before checking presentation details.
Validate metadata as a connected record
Titles, descriptions, artwork, credits, runtime, release date, rating and series structure need to agree. A technically valid feed can still be unusable when an artwork language, episode number or territory value contradicts the rest of the record.
Add the distribution review before hand-off
Flag material that may affect warnings, parental controls, scheduling, ad suitability or a territory-specific deliverable. Record the exact asset reviewed. If the partner receives a different cut, the earlier result should not be assumed to carry over.
Define readiness as release gates
Replace one broad 'metadata complete' status with gates that have evidence: identity, rights, presentation, content controls and technical delivery. A title can pass one gate and fail another, which gives operations a useful queue instead of a vague red cell.
Each gate needs an owner, a failure reason and a retry path. When a partner rejects a title, store the original response and the corrected field. That history becomes a partner-specific test suite for the next delivery.
Validate relationships, not isolated fields
A valid episode number is still wrong when it points to the wrong season. A valid artwork URL is still wrong when it carries another language or aspect ratio. Run cross-field checks: series to season, episode to asset, availability to territory, rating to system, and artwork to locale.
Give stable internal identifiers priority over display names. Partner identifiers can then be mapped without letting a renamed title, recut episode or duplicate translation create a new untracked object.
- Compare runtime and episode order against the delivered media.
- Test that every required relationship resolves in both directions.
- Flag contradictory countries, languages and availability windows.
Make rejection data operational
Do not leave partner errors in email threads. Normalize the partner code, affected field, title identifier, owner and resolution. Review the top recurring causes each week. If one mapping creates dozens of rejects, fix the transformation rather than correcting titles one at a time.
Measure acceptance by delivery batch and error category. A useful dashboard distinguishes source-data defects from mapping defects and partner-policy decisions, because each needs a different fix.
Reference architecture
The operating blueprint
Each stage has an owner and a concrete output. That makes failed hand-offs visible before launch.
- 01Partner operations
Define the delivery contract
Record service, territories, windows, required fields and accepted values.
Output · Versioned partner profile - 02Catalogue operations
Resolve identity
Confirm title hierarchy, identifiers, runtimes, versions and language relationships.
Output · Connected title graph - 03Rights operations
Prove rights and availability
Validate territory, channel, start and end times, timezone and restrictions.
Output · Eligible delivery set - 04Metadata and standards
Complete presentation and policy
Check localized copy, artwork, ratings, warnings and access controls.
Output · Viewer-ready metadata - 05Distribution engineering
Contract-test the package
Validate schemas, enumerations, URLs, media references and a representative ingest.
Output · Partner-valid payload - 06Launch owner
Close the feedback loop
Triage rejects, correct the source or mapping, and preserve the test case.
Output · Accepted catalogue and lessons
Decision map
What to do when the record is not clean
Common situations, a practical next move, and the reason the shortcut fails.
| Situation | Recommended move | Why |
|---|---|---|
| A required field is unknown | Use the partner-approved null or hold the record | Invented defaults create harder downstream corrections. |
| Two sources disagree | Apply the documented source-of-truth rule and retain the conflict | Silent overwrites remove the audit trail. |
| A feed passes schema validation but fails ingest | Capture the partner error and test the semantic relationship | Valid syntax does not guarantee usable metadata. |
| One fix applies to many titles | Correct the transformation and replay the batch | Per-title patches preserve the systemic defect. |
Failure modes
Patterns worth stopping early
These are operating defects, not writing problems. Fix the record or workflow at the source.
Green spreadsheet, red launch
Teams mark fields complete without testing partner rules.
Add a representative partner ingest before catalogue sign-off.Duplicate identity
A rename or language variant becomes a new title.
Anchor mappings to stable internal and partner identifiers.Ownerless rejection
Errors circulate between commercial, metadata and engineering.
Assign each gate and error class to a named operating owner.Patch accumulation
Manual partner fixes never return to the source system.
Choose explicitly between source correction and partner-specific transformation.Evidence example
What a review record looks like
One published example from the catalogue—not a rule or template answer for another asset.
Practical questions
What teams ask during implementation
Short answers for the decisions most likely to slow a release or partner hand-off.
Does schema-valid mean catalogue-ready?
No. It proves the payload shape, not the accuracy of title relationships, rights, artwork, ratings or media references.
Should every partner receive the same feed?
Use one governed source record, but map it into each partner's required schema, vocabulary and delivery policy.
What is the smallest useful pilot?
Choose a representative set with a movie, a multi-season series, localized metadata, restricted content and more than one territory. Validate the full ingest and correction loop.
Which metric should leadership watch?
First-pass acceptance is useful when paired with reject cause and time-to-resolution. A high acceptance rate can still hide slow, expensive exceptions.
Primary references
Check the current official guidance
Rules and platform specifications change. Follow the current source for the intended territory and service.
Continue the review
Use the blueprint with live evidence
Move from the operating model into market, issue, brand and title records.