Rights and contracts
Media licensing territory constraints: modeling them in a workflow task
By SourceX Editorial · Reviewed by Noah Loul ·
10 sections
Short answer
To represent territory restrictions in a media licensing workflow task, store each signed grant as its own row with country codes, an explicit included or excluded list type, channels, dates, exclusivity and a clause link. Then ask one concrete question, such as whether this cut may go to this channel in this country on this date, and verify the answer against delivery logs rather than a status field.
License terms
Media licensing territory
Key takeaways
- A territory list means nothing until you store whether it includes or excludes; the same country code can mean only here or everywhere but here.
- Component grants such as music, talent and stock licenses often limit an asset more tightly than the client agreement.
- Under U.S. copyright law an exclusive license is a transfer even when it is limited to one territory, so exclusive and non-exclusive territories belong in separate rows.
- When grants conflict and the contracts are silent, the task should hold or escalate, which is also the default in the W3C ODRL rights language.
- A case is verified only when the decision cites grant rows and the outcome cites a delivery or takedown record.
How do you represent territory restrictions in a media licensing workflow task?
Represent them as a constraint record, not a label. Each grant that touches the asset gets its own row: the permitted or excluded territories as country codes, the channels and uses it covers, the dates it runs, whether it is exclusive in each territory, and a link to the signed document and clause that says so. The task then asks one concrete question, such as whether a specific cut may go to a specific channel in a specific country on a specific date, and the answer has to be derived from those rows.
That separation matters because a media license can permit one territory and exclude another for the same asset, and the excluded one is often hidden in a different document: a music sync license, a talent agreement or a stock footage license rather than the client contract. A task built from the status field alone (approved, delivered, live) teaches the wrong lesson, because the status records what happened, not whether it was allowed.
This guide is written for creative operations and rights teams at agencies, studios, publishers and production companies who want their delivery history to become reusable task cases. It shows the fields, the conflict rules, an illustrative worked case and a counterexample. It is general information about record design, not legal advice; counsel should confirm how any specific grant is read.
Why does a single territory field fail on real media grants?
A single field cannot say whether its list is an inclusion or an exclusion, and the two mean opposite things. The DDEX knowledge base, which documents how music and media deals are exchanged between labels and digital services, puts it plainly: a deal carries either a list of territories where it applies or a list where it does not, never both. A code for Norway in the first list means the deal is available only in Norway; the same code in the excluded list means it is available everywhere except Norway.
At a glance
Why does a single territory field fail on real media grants
The second failure is that territory rarely travels alone. DDEX models how a consumer obtains content (its UseType) and how the consumer pays (its CommercialModelType) as separate values, and its guidance says a deal covering both streaming and offline use should carry both use types rather than one convenient headline value. A creative grant behaves the same way: broadcast in Canada and online video in Canada are two different permissions, and either can exist without the other.
The third failure is exclusivity. Under the U.S. Copyright Act, an exclusive license counts as a transfer of copyright ownership whether or not it is limited in time or place, and a transfer has to be in a signed writing to be valid. So a grant that is exclusive in one country and non-exclusive elsewhere is two different legal objects, and the task record should hold them as two rows, each pointing to its own signed source.
- Inclusion versus exclusion: store the list type explicitly, never infer it from the field name.
- Worldwide: store it as a value, and treat an exclusion list that removes the whole world as an error to send back, not a constraint.
- Channel and use: store them per grant, because the same territory can be cleared for one channel and closed for another.
- Exclusivity: store it per territory and per grant, with the clause reference.
What should the territory and channel constraint task pack include?
The pack has two parts: the grant rows, which describe what each signed document allows, and the task row, which states the request and the decision. The field set below borrows its vocabulary from two published standards so that a reviewer outside your company can read it: W3C ODRL, the rights expression language, defines a spatial constraint as a named geographic area whose code source must be stated (ISO 3166 country codes are its example), a delivery channel constraint, a media context constraint and set operators such as is any of and is none of; DDEX supplies the inclusion and exclusion convention. Every row is filled from a document you can link, or it is marked unknown.
| Field | What it holds | Where it comes from | Check before the case is used |
|---|---|---|---|
| Asset and version | The exact cut, edit or file the request concerns, with version number and the components it contains | Asset management or review tool version history | The version in the request matches the version the grants cover |
| Grant source | One row per signed document: client agreement, talent release, music license, stock license, location release | Contract repository or rights database, clause by clause | Every component of the asset has at least one grant row or is marked owned |
| Territory list and list type | Country codes with the code source named, plus whether the list includes or excludes | The grant clause, transcribed by the rights reviewer | List type is stated; no row has both lists; Worldwide is explicit |
| Channel and use | Broadcast, online video, paid social, out of home, in-store, print, internal or other, per grant | The grant clause and any media plan it references | Each channel in the request appears in every grant that covers a component |
| Period and holdbacks | Start and end date-times with timezone, and any window during which a channel or territory is closed | Grant clause and amendments | The requested go-live date falls inside every relevant period |
| Exclusivity | Exclusive, non-exclusive or unknown, per territory and grant | Grant clause | Unknown values are routed to the rights reviewer, not defaulted |
| Precedence rule | What the contracts say happens when grants conflict, or that they say nothing | Order of precedence clause, or counsel note | A silent contract is flagged, never filled with a guess |
| Approvals and revisions | Who approved which version for which market, with timestamps | Review tool approvals, client sign-off emails, change orders | The approval names the same version and market as the request |
| Decision and reason | Deliver, deliver with restriction, hold or refuse, with the grant rows that drove it | Rights reviewer or traffic coordinator record | The reason cites row identifiers, not a free-text summary alone |
| Outcome evidence | What actually went out, where, when, and any later takedown or extension | Delivery logs, platform availability settings, traffic instructions | The outcome is checked against the decision and differences are recorded |
Which grant wins when two documents disagree about a territory?
Whichever one the contracts say wins, and if they say nothing, nobody in the task should decide by default. Record the precedence clause as data and leave the conflict visible for counsel when there is none.
ODRL is useful here because it makes the choice explicit. A policy can state a conflict strategy: permissions override prohibitions, prohibitions override permissions, or the whole policy is void when any conflict appears. If no strategy is stated, the specification's default is void, and if merged policies carry different strategies, the merged policy is void too. That default is the right instinct for training data: an unresolved conflict should become a hold or an escalation in the task, not a confident deliver.
In practice, the common real-world pattern is a broad client grant sitting on top of a narrow component license. The client agreement permits online video in North America and Europe, while the music license for the track in the cut covers only the United States and Canada. Reading the grants component by component, the narrowest grant on any component limits what the finished asset can do, unless the component is replaced. The task should capture that reasoning step, because it is the step most often skipped when people rely on a campaign-level status.
Which changes to a cut, a grant or an approval reopen the decision?
Any change to the asset's components, the grant documents or the approval scope can change the answer, so each one needs a dated row and a pointer to its source. DDEX's own deal model is a good habit to copy: when terms such as territories, use types or channels change, its ERN 4.2 data dictionary says a new deal should be created and the old deal's validity period ended, rather than editing the old one in place.
- A revision swaps a component: a new music bed, a new stock shot or a new voice-over brings its own grant, which may be narrower or wider than the one it replaces.
- An amendment adds or removes a territory: record it as a new grant row with its own effective date and close the old row; do not overwrite.
- A talent or location release limits use: these often restrict territory, channel or duration more tightly than the client agreement.
- A holdback or exclusivity window closes a channel for a period even though the territory is otherwise licensed.
- An approval is scoped: a client approving the North American cut has not approved the European cut, even if the files differ only in an end card.
- Portability rules cut the other way: in Europe, Regulation 2017/1128 lets paying subscribers keep using an online content service while temporarily in another member state and treats that use as occurring in their home state, so a subscriber's travel does not by itself mean the service is being offered in a new territory.
Worked case (illustrative): one cut, three markets, two answers
Everything in this case is invented for illustration. A production company holds the edit history for a 30-second spot made for a consumer brand. The traffic team receives a request to deliver the final version to an online video channel in Canada, Germany and Brazil, going live on the first of the month. The task asks a model to decide, market by market, and to cite the rows that support each answer.
| Step | Record in the pack | What it shows | Effect on the decision |
|---|---|---|---|
| Starting state | Request ticket with asset version 7, channel online video, territories CA, DE, BR | The request is specific enough to test | Three separate decisions are needed |
| Client grant | Agreement clause, inclusion list CA, US, DE, FR; channels broadcast and online video; 12-month period | Brazil is not in the client grant | BR is refused unless a new grant is added |
| Music license | Sync license, inclusion list US, CA; online video only | Germany is not covered for the music in version 7 | DE cannot use version 7 as cut |
| Talent release | Release, inclusion list US, CA, DE; broadcast and online | Talent does not block CA or DE | No change |
| Revision | Version 8 replaces the music bed with a track whose license is worldwide for online video | A component change widens the music grant | DE becomes possible with version 8 if the client approves it |
| Approval | Client sign-off for version 7 in CA only | Version 8 has no approval yet | DE is held pending approval of version 8 |
| Decision | CA: deliver version 7. DE: hold for approval of version 8. BR: refuse, no client grant | Each answer cites grant rows | Reason codes point to rows, not to a summary |
| Outcome check | Delivery log shows CA delivered; DE delivered two weeks later after approval; nothing sent to BR | The decisions held up against what happened | The case is marked verified |
Counterexample (illustrative): when the status field says approved for all markets
The tempting shortcut is to train on the campaign tracker, where the asset reads approved, all markets. In this invented counterexample, that status was set at the campaign level before a stock footage clip was added in a later revision, and the stock license excludes one of the campaign's countries. The asset was delivered there anyway and later pulled.
If the task were built from the status field, the model would learn that approved means deliverable everywhere. Built from the pack, the case shows the real sequence: a component change, an exclusion list nobody re-read, a delivery that the grants did not support, and a takedown recorded as the outcome. Keep cases like this. A delivery that should not have happened, with the evidence showing why, teaches more than ten clean approvals, as long as the dispute is settled and the parties' confidentiality terms allow its use.
The same applies to channels that cannot be limited by country. If a request asks for a post on a globally visible account and the grant is territory-limited, record the mismatch and route it to the rights reviewer; the case's lesson is the routing, not a guess about whether the post was acceptable.
Which record shows what was actually delivered, and where?
The signed grant proves what was allowed; only the delivery record proves what happened. A case is verified when the decision cites grant rows and the outcome cites a log that someone outside the creative team could check.
Good outcome evidence includes traffic instructions with destination and date, delivery confirmations from a distribution platform, availability or geo-restriction settings exported from the publishing tool, and takedown or extension notices. A closed ticket with a final message such as sent to all markets is a claim, not evidence. Where no outcome record exists, keep the case but label it decision only, so it can be used to test reasoning without being presented as a verified result.
What should be removed or held back before workflow history is licensed?
Remove what identifies people who never agreed to the use, and hold back anything your own contracts keep confidential. Licensing your delivery history is a separate permission from the grants inside it: a client agreement that let you deliver a spot does not, on its own, let you license the agreement's terms or the spot as training data, and de-identifying a record does not create that right.
- Contract terms that clients or licensors marked confidential: replace with the structured constraint values where the agreement allows, or exclude the case.
- Names and contact details of talent, crew and client staff: replace with role labels such as lead actor or client approver.
- Payment amounts and rate cards: drop them; the territory and channel logic does not need them.
- The creative assets themselves: include them only where you hold the right to license them, which is a separate question from the workflow records.
- Open disputes and pending claims: hold them until settled and cleared by counsel.
Turning rights-tracked delivery history into a licensed dataset
Production companies, agencies and publishers that already keep grant-level rights records hold something most training sets lack: decisions about what may go where, linked to the documents that decided them. Where a company chooses to license that history, SourceX manages each stage of the work: rights review, contracting, preparation and delivery.
Scope stays with you. Your company approves the scope and terms and signs the agreement before any delivery record or grant is handed over, and the process starts with a short company assessment that needs no data upload.
Frequently asked questions
How do you record a territory exclusion in a media license?
Store the exclusion as a list of country codes with an explicit list type of excluded, and name the code source, such as ISO 3166. Keep it in a separate row from any inclusion list, because a single list cannot be both. DDEX, the media metadata standard, follows the same rule: a deal carries either territories where it applies or territories where it does not, never both.
What happens when the client grant is wider than the music license?
Read the grants component by component. The narrowest grant on any component in the cut limits where the finished asset can go, unless that component is replaced in a new version. Record the reasoning as a step in the task so the decision can be checked.
Is an exclusive license in one country a different thing from a non-exclusive one elsewhere?
Yes. Under the U.S. Copyright Act an exclusive license is a transfer of ownership even when it is limited in place or time, and it must be in a signed writing. Keep exclusive and non-exclusive territories as separate grant rows with their own clause references, and have counsel confirm how a specific grant reads.
Can a campaign status field be used as the training label?
Only as one input. A status records what the team believed or did, not whether the grants allowed it. Derive the label from the grant rows and verify it against delivery records.
Does European portability mean a service is being offered in every member state?
No. Regulation 2017/1128 lets paying subscribers keep using an online content service while temporarily in another member state and treats the use as happening in their home state. It does not by itself extend a license to new territories.
Should refused and pulled deliveries be kept in the dataset?
Yes, when the facts are settled and your contracts allow it. Refusals, holds and takedowns show the decision boundary more clearly than clean approvals, provided each one links to the grant rows and outcome records that explain it.
Sources
- A DDEX deal lists either the territories where it applies or those where it does not, not both; Worldwide is an allowed value, and excluding Worldwide is discouraged. Source
- DDEX separates how a consumer obtains content (UseType) from how they pay (CommercialModelType), and a deal covering several uses lists each use type. Source
- In ERN 4.2, when deal terms such as territories, use types or channels change, a new deal is created and the old deal's validity period is ended. Source
- The W3C ODRL vocabulary defines spatial, delivery channel and media context constraints and the is any of and is none of operators; spatial values name their code source, such as ISO 3166. Source
- The W3C ODRL information model defines conflict strategies (perm, prohibit, invalid), with invalid as the default and merged policies with different strategies void. Source
- 17 U.S.C. 101 defines a transfer of copyright ownership to include an exclusive license, whether or not limited in time or place of effect, but not a nonexclusive license. Source
- 17 U.S.C. 204(a) requires a transfer of copyright ownership, other than by operation of law, to be in a writing signed by the owner of the rights conveyed or an authorized agent. Source
- Regulation (EU) 2017/1128 lets subscribers temporarily present in another member state keep using paid online content services, with the use deemed to occur in their member state of residence. Source
See if your company qualifies
A short company assessment. No data uploads are needed.