Exclusivity clauses in dataset licences: what you are actually buying

Exclusivity is a scope question before it is a price question. What the clause has to define, what it can never deliver, and when the premium is not worth paying.

Exclusivity is a scope, not a switch

A licence is either exclusive or it is not, and buyers tend to treat that as a single decision. In practice exclusivity is a set of boundaries, and a clause that leaves any of them undrawn is not exclusive in the way the buyer assumes. The boundaries are field of use, territory, term, and the definition of the corpus itself.

A grant that names a product but not a field leaves the vendor free to license the same audio into an adjacent market. A grant that names a territory leaves the rest of the world open. A grant that runs for a fixed number of years expires quietly while the model is still in service, usually with no notice obligation attached.

The most common way exclusivity leaks is a re-version. The vendor ships a new edition of the same collection — new name, new packaging, mostly the same speakers and the same sessions — and argues that the exclusivity attached to the earlier edition only.

Write the definition against facts that cannot be renamed: the speaker roster, the collection window, the recording locations, the guideline revision. Exclusivity should attach to the material, and any collection that substantially overlaps the roster should fall inside it. A percentage threshold for overlap is negotiable, but a threshold that exists beats a definition built on product names.

Exclusivity cannot recall what already shipped

A grant binds the vendor going forward. It does nothing about copies already licensed, and it cannot remove the audio from models that were trained on it before you arrived. This is why exclusivity is worth most on a new collection and least on catalogue data that has been circulating for years.

One clause closes part of that gap: a prior grants schedule. A written list, attached to the contract, of every entity that has received the material, with dates and scopes, and its accuracy made a representation. Ask for it before signing rather than after. A vendor who will not produce the list is telling you the material is widely distributed, and the exclusivity should be priced on that basis.

How exclusivity is priced, and what you are paying for

Three structures are common. A premium on the licence fee, expressed as a multiple of the non-exclusive rate. A buy-out, where the fee is large enough that the vendor takes the collection off the market entirely. And a standstill, where the vendor is paid to stop licensing the material to anyone else for a period, with no transfer of ownership involved.

All three are priced on the vendor's forgone sales, not on your benefit. The same exclusivity therefore costs more for a popular collection and less for one nobody has licensed. Before agreeing to a premium, answer honestly whether your advantage depends on the material being rare. If the model's edge comes from your architecture, your product or your users, exclusivity is a tax that buys nothing.

Verify it, and define the remedy

Exclusivity without a verification right is a promise. Two provisions make it checkable: an audit right over the vendor's licensing records for the material, and a notice obligation whenever the vendor grants a new licence that could overlap. Neither requires sight of their pricing or their other clients — a schedule of grant dates and scopes is enough.

The remedy matters more than the definition of breach. Most exclusive licences provide that a breach entitles the buyer to a refund or a fee reduction, not to an injunction, and a refund arriving after two years of competing use is not a remedy. Negotiate a defined consequence up front: a repayment calculated as a multiple of the premium paid plus a termination right, or a step-down to non-exclusive terms with the difference returned.

Questions to put to the vendor in writing

These are worth asking as a written list, because the answers become the schedule that defines the grant.

  • Which fields of use, territories and model families does the exclusivity cover, and which does it leave open?
  • Is the corpus defined by roster and collection window, or by product name?
  • What is the complete list of prior licensees, with dates and scopes?
  • Does exclusivity reach re-recordings, re-annotations and new editions that overlap the roster?
  • What notice do I receive if the vendor licenses overlapping material?
  • What is the defined remedy if exclusivity is breached?
  • Does the exclusivity survive a change of control at the vendor?

The limit of the instrument

Exclusivity controls what the vendor does next. It cannot undo the past and it cannot stop a determined counterparty from rebuilding the same collection from the same population. It is a commercial instrument, not a technical one, and the two should not be confused when the value of the data depends on nobody else having it.

This is a practical walkthrough rather than legal advice. Exclusivity drafting turns on the specific facts of the collection, so the clause should be reviewed by counsel before it is signed.

More insights

Submit a sourcing request

Tell us the language, the hours, and what the data needs to look like. You will get a real number and a real timeline — not a range. If we cannot source it well, we will tell you that instead.

  • Pilot batch before the full run, so problems surface early.
  • Consent documentation delivered with the data.
  • No medical or clinical data. No recorded telephone calls.

We reply within two business days. Your details are used only to answer this request. See our privacy policy.

Contact

Talk to a human

Send a specification and we will come back with a real number and timeline.

Submit a sourcing request

Or email hello@linguacorpus.com