Go/no-go criteria for a first data licensing deal

What decides whether a first data licensing deal should go ahead?#

A first data licensing deal should go ahead only when four criteria score well enough: rights clarity, personal data share, competitive sensitivity and internal capacity. They sit on top of a baseline question, whether your records are substantial and connected enough for a buyer to want them, which a fit check answers first.

The scorecard gives three outcomes rather than a yes or no. Many first deals end as proceed with exclusions: the core record family goes forward, and specific fields, customers or record types are carved out. That is a normal result, not a compromise.

Score each record family separately. Support tickets, job histories and personnel files rarely land in the same place, and a single company-wide verdict hides the families that could move.

The go/no-go scorecard#

The go/no-go scorecard sets out what each outcome looks like for each criterion. Fill it in with the people who run the systems and the counsel who knows your contracts, not from memory in a leadership meeting.

Rights clarity: can you show you are allowed to license?#

Rights clarity means you can point to the documents that let you license a record family for AI development. Start with customer contracts and master service agreements, since confidentiality and data use clauses often decide the answer, then check vendor terms for the systems that hold the records and the notices you have given employees.

Some record families fail on ownership rather than wording. An engineering firm's client deliverables, a manufacturer's customer-owned designs and a software company's customer code typically belong to someone else. Those land on the stop side even when the rest of the archive proceeds.

Rights clarity behaves differently from the other criteria. Weak scores on personal data, sensitivity or capacity can usually be managed with exclusions or timing. A weak rights score cannot be engineered away; it needs a contract answer, a consent or a smaller scope.

Personal data and competitive sensitivity: what has to come out?#

Personal data share measures how much of each record is about identifiable people. Support tickets that name a customer contact can usually be de-identified without losing the problem and the fix. A staffing firm's candidate files are different, because the person is the subject of the record.

Competitive sensitivity measures what would hurt if a rival saw it. Pricebooks, margin fields, bid histories and named customer lists are common candidates for removal. For each family, ask: if we strip the sensitive fields, is a useful record of work still left?

• 1Proceed when sensitive content sits in fields you can drop without breaking the record.
• 2Proceed with exclusions when removal is possible but needs review, such as names inside free-text notes.
• 3Stop when the sensitive part and the valuable part are the same thing.

Which exclusions come up most in a first deal?#

The exclusions that come up most in a first deal follow the record family, not the industry. The table below lists common carve-outs so you can check whether a family still holds a useful record of work once they are removed.

If a family needs every exclusion in its row and little remains, score it stop for now. If the exclusions are narrow and the decision trail survives, it is a strong proceed with exclusions candidate.

Internal capacity: who will actually do the work?#

Internal capacity is the criterion owners most often skip. A first deal needs an executive sponsor who can make calls, a systems lead who can run exports and answer schema questions, and counsel time for the rights review and the contract.

Timing matters as much as headcount. Starting during an ERP go-live, a peak season or an audit puts the work on people with no slack, and a rushed first deal is more likely to include something that should have been excluded. If capacity is the only weak score, schedule the work rather than stopping it.

Illustrative: an electrical contractor scores its first deal#

Illustrative: a fictional electrical contractor with a residential service division and a commercial projects division is approached about its field service history. Its records include service calls, estimates, technician notes, invoices and warranty callbacks across many years in its field service system.

Rights score proceed for residential service, because the records are company-owned and its customer terms are standard. Commercial project files score stop, because several subcontracts with general contractors restrict use of project documents. Personal data scores proceed with exclusions, since homeowner names, addresses, phone numbers and gate codes appear in technician notes. Competitive sensitivity scores proceed with exclusions for pricebook and margin fields, and capacity scores proceed once the busy summer season ends.

The owner decides to proceed with exclusions: residential service records only, identifiers and pricing removed, with work starting after peak season. The commercial division stays out until its subcontract terms are reviewed.

How SourceX applies go/no-go criteria#

SourceX builds these criteria into the SourceX five-step transaction. Rights clarity is tested at the Rights step, personal data and competitive exclusions are applied at Preparation, and the supplier makes the final call at Approval, with the option to stop at any step.

The baseline question of whether records are worth licensing is assessed with the SourceX Enterprise Data Value Framework, a SourceX methodology that gives qualitative ratings on drivers such as uniqueness, domain expertise, human-generated signal, recency, rights and AI utility, and weighs preparation cost and privacy burden against them. The scorecard is then applied to record families that clear that baseline. Nothing is shared during the initial assessment, which runs on metadata alone.