DataPayouts
Choosing a data buyer means comparing the proposed use, payment and rights across a defined set of records. DataPayouts helps companies understand those differences and decide which deal fits their needs.
Mercor, Handshake, micro1, Scale, Turing and Labelbox Prism publish programmes for acquiring or licensing company operational data. Defined.ai offers a broader data-partner programme through its marketplace. Their requirements and commercial arrangements differ, so a useful comparison starts with the records you hold and the rights each programme wants.
This guide compares their published information, checked on 8 October 2026. Programme descriptions are providers' claims. Eligibility, security arrangements and payment depend on the assessment and signed agreement.
For the wider process, read how to sell company data for AI. See the company fit check for the details used in an initial review.
Programmes at a glance
| Programme | Published focus | Commercial point to clarify |
|---|---|---|
| Mercor | Connected company records | Buyer acceptance and payment conditions |
| Handshake | Operational knowledge | Purchase rights and downstream recipients |
| micro1 | Mature companies with 30+ employees | Ongoing work and exclusivity |
| Scale | Established operators, preferably 40+ employees | Purchase commitments for repeat packages |
| Turing Lazarus | Active and winding-down company histories | Rights transferred for a one-time payment |
| Labelbox Prism | Operational data and work documents | Default exclusivity and revenue share |
| Defined.ai | Datasets offered through its marketplace | Seller share and preparation costs |
The programme sections below link to each provider's published information.
Compare the commercial role
Ask who will sign the contract, who will use the data and who owes the payment. An organisation can acquire records, prepare datasets and license them to another company within the same arrangement.
Handshake explicitly describes purchasing company data. Mercor describes licensing and preparation for delivery to a buyer. Defined.ai describes marketplace sales. These descriptions help explain the route to a sale, but the contract must identify each party's responsibilities. Mercor, Handshake, Defined.ai
Retaining ownership also leaves the question of exclusivity open. A licence can restrict your ability to license the same records elsewhere while allowing you to keep using them internally. Prism's FAQ describes this distinction. Prism FAQ
Mercor Data
Mercor seeks connected work records, including messages, email, CRM records and code history. Its published process covers scoping, connecting systems, extracting and de-identifying records, and payment after delivery to the buyer. Mercor says it covers extraction, anonymisation and transfer costs. Mercor Data
The programme page does not state a standard exclusivity term or licence duration. Ask whether payment requires buyer acceptance, how a rejected dataset is handled and which later uses the agreement permits.
Handshake employer data partnerships
Handshake says it purchases company data to build environments where AI models can practise real work. Handshake says it agrees the scope upfront, uses an NDA and de-identifies the data before use. Handshake programme announcement
Its privacy policy says business data may be made available to AI labs and similar partners for model development, training and evaluation. Handshake privacy policy
The announcement does not specify a standard exclusivity period, payment trigger or minimum company size. Ask which recipients can access the prepared data and whether the agreement transfers ownership or grants a licence.
micro1 company data partnerships
micro1 seeks operationally mature companies with at least 30 employees. It prioritises US companies and primarily English-language records. Examples include SOPs, knowledge bases, CRM data and project histories. Companies retain ownership of the underlying data. micro1 programme
Its process moves from evaluation to discovery and an agreement. micro1 says companies can review representative samples before prepared data is used. The published page leaves exclusivity and payment timing to be clarified. Ask whether participation requires continuing updates or staff feedback, and how those obligations affect compensation.
Scale data partnerships
Scale prefers companies with at least 40 employees, while considering partnerships individually. It seeks records from established systems across messaging, documents, business software and data warehouses. Scale describes recurring revenue opportunities, approval for each data package and anonymisation within the company's environment. Scale data partnerships
Ask which data packages would be paid for, how frequently they would be required and whether the agreement commits to a minimum purchase. The public page does not establish a standard exclusivity term or guaranteed recurring payment.
Turing Project Lazarus
Turing's Project Lazarus seeks operating histories from active and winding-down companies. The programme describes intake, sampling and an offer for a one-time payment. Its examples include code, tickets, internal documentation and records of business decisions. Project Lazarus
Turing also describes acquiring private repositories and supporting operational records from defunct companies. Turing programme description
Ask exactly which rights would transfer, what the company can continue using and what authority is required if it is winding down. The published descriptions do not replace the proposed acquisition or licence agreement.
Labelbox Prism
Prism seeks operational records and work documents, including cross-system workflows, reviews and support histories. Its FAQ describes an upfront licence fee plus a share of downstream revenue. It says partnerships are usually exclusive by default, while the company retains ownership and continued use. Prism FAQ
Ask how exclusivity applies to the dataset, related records and future updates. Get the revenue-share calculation and reporting rights in writing. Prism's process page mentions revocation with 90-day notice. Its FAQ describes deletion within 30 days after specified triggers and says revocation does not undo completed model training. Ask when each provision applies. Prism process
Defined.ai data partners
Defined.ai's data-partner programme covers audio, video, images, text and multimodal datasets. It describes licensing income from marketplace sales and checks of data quality, metadata and licensing rights. Its separate AI service-supplier programme offers collection and annotation projects. Defined.ai partnership programmes
For an existing archive, ask about the data-partner route, buyer demand and any preparation costs. The public programme page does not specify a standard seller revenue share or exclusivity term. A service-supplier application concerns providing work and requires a separate assessment of fit.
Questions to ask before choosing a programme
Ask each programme to answer the same questions against the same proposed dataset:
- Who signs the agreement, pays us and receives the data?
- Which records are included, and can we approve the final package?
- Does the agreement permit training, evaluation, retrieval, redistribution or sublicensing?
- Is the licence exclusive, and which datasets, buyers, uses and dates does that cover?
- What makes payment due, and can costs or rejected records reduce it?
- Are updates, staff interviews or other continuing work required?
- What happens to source copies, prepared datasets and further use after termination?
These questions are a starting point for commercial and legal review. De-identification and an NDA address only part of the decision. Confirm the rights you hold in employee, customer and third-party material before sharing samples.
Compare your options with DataPayouts
We start with a description of your company, record types and any existing buyer discussions. We help you compare requirements and proposed terms, and identify questions to resolve before you choose a deal. Keep the initial description free of customer details and confidential samples. Use the data licensing readiness guide to organise that information before approaching programmes.
Compare written offers against the same scope. A larger headline payment may cover more data, broader rights or continuing work. The amount alone cannot establish which offer best fits your company.