Reference

Data licensing jargon, translated into everyday English

Each buyer and broker has its own words for the same idea. micro1 talks about Company Data Partnerships, Handshake about AI data partnerships, and Ventra stresses license, not sale. Here they all are, in one place, in ordinary language.

These are working definitions, not legal ones. What any term means in your own contract is something to settle with the buyer and your lawyer.

AI Data Offers research deskChecked against our methodology Updated October 28, 2026 3 min read 14 sources

In this article

micro1
Ventra
Jottings
Handshake AI

Contract and money terms

Data payout
The sum a business receives for licensing its records, counted before middleman fees. More in how licenses are priced.
Company data partnership
An arrangement where an AI firm pays a business for a license to copies of its work records, while the originals stay with the business.
Data license
The contract granting a buyer rights to a copy of particular records, on agreed conditions.
Data manifest
An inventory of your records by system, years covered and approximate volume. micro1 estimates 1 to 3 hours to prepare.
Exclusive license
A promise not to license the same records to any other buyer, either for a set time or permanently.
Non-exclusive license
You may license the same records to additional buyers.
Perpetual
Has no end date.
Irrevocable
You cannot withdraw it.
Sublicense
The buyer hands the data on to a further company.
Revenue share
Payment each time the data is licensed on, in place of a single fixed amount.
Referral fee
A finder's payment for bringing a business and a buyer together. micro1 offers as much as $50K.
Success fee
A broker's commission, payable only when the deal completes. Jottings quotes 20 to 30%.

Term	In everyday words
Data payout	The sum a business receives for licensing its records, counted before middleman fees. More in how licenses are priced.
Company data partnership	An arrangement where an AI firm pays a business for a license to copies of its work records, while the originals stay with the business.
Data license	The contract granting a buyer rights to a copy of particular records, on agreed conditions.
Data manifest	An inventory of your records by system, years covered and approximate volume. micro1 estimates 1 to 3 hours to prepare.
Exclusive license	A promise not to license the same records to any other buyer, either for a set time or permanently.
Non-exclusive license	You may license the same records to additional buyers.
Perpetual	Has no end date.
Irrevocable	You cannot withdraw it.
Sublicense	The buyer hands the data on to a further company.
Revenue share	Payment each time the data is licensed on, in place of a single fixed amount.
Referral fee	A finder's payment for bringing a business and a buyer together. micro1 offers as much as $50K.
Success fee	A broker's commission, payable only when the deal completes. Jottings quotes 20 to 30%.

The players

AI lab
A developer of large AI models, for example Google, OpenAI, Anthropic or Meta.
AI data broker
An intermediary who links a business to a data buyer, usually paid by the buyer.
Data marketplace
A website where datasets are listed for any buyer to find. Datarade's commission is 15 to 30%.
Consumer privacy ombudsman
An independent official the court appoints in certain bankruptcy sales involving personal data.
Instant Form
Facebook's and Instagram's built-in lead form, which most company data ads use.

Term	In everyday words
AI lab	A developer of large AI models, for example Google, OpenAI, Anthropic or Meta.
AI data broker	An intermediary who links a business to a data buyer, usually paid by the buyer.
Data marketplace	A website where datasets are listed for any buyer to find. Datarade's commission is 15 to 30%.
Consumer privacy ombudsman	An independent official the court appoints in certain bankruptcy sales involving personal data.
Instant Form	Facebook's and Instagram's built-in lead form, which most company data ads use.

Privacy vocabulary

De-identification
The general name for any method that separates data from the individuals it describes.
Pseudonymization
Replacing names with codes so that one person's records still connect without revealing who they are.
Anonymization
A loose term. Sometimes it equals de-identification; sometimes it means removal that cannot be reversed.
PII
Personally identifiable information, such as names, email addresses, phone and account numbers.
Indirect identifier
A detail like a date of birth or ZIP code that, combined with others, can reveal who someone is.
Re-identification
Figuring out a person's identity from data that was supposed to be cleaned.
SOC 2 Type II
An independent auditor's report on how well a company safeguards data over a period of time.

Term	In everyday words
De-identification	The general name for any method that separates data from the individuals it describes.
Pseudonymization	Replacing names with codes so that one person's records still connect without revealing who they are.
Anonymization	A loose term. Sometimes it equals de-identification; sometimes it means removal that cannot be reversed.
PII	Personally identifiable information, such as names, email addresses, phone and account numbers.
Indirect identifier	A detail like a date of birth or ZIP code that, combined with others, can reveal who someone is.
Re-identification	Figuring out a person's identity from data that was supposed to be cleaned.
SOC 2 Type II	An independent auditor's report on how well a company safeguards data over a period of time.

AI vocabulary

AI agent
A program that completes an entire job, closing a ticket for instance, instead of only answering a question.
RL environment
A simulated version of real software in which an AI agent attempts tasks and earns rewards for getting them right.
Operational data
The routine records a company generates: mail, chat, tickets, job notes and SOPs.
Synthetic data
Artificially generated records designed to look real, which tend to contain giveaway mistakes.
Token
A word or fragment of a word; AI training volumes are measured in them.

Term	In everyday words
AI agent	A program that completes an entire job, closing a ticket for instance, instead of only answering a question.
RL environment	A simulated version of real software in which an AI agent attempts tasks and earns rewards for getting them right.
Operational data	The routine records a company generates: mail, chat, tickets, job notes and SOPs.
Synthetic data	Artificially generated records designed to look real, which tend to contain giveaway mistakes.
Token	A word or fragment of a word; AI training volumes are measured in them.

Using the vocabulary on a call

If a buyer says partnership, ask whether it is a license, whether it is exclusive, whether it lasts forever and whether they can sublicense. If they say anonymized, ask what method they use, who carries it out and whether you will see a sample first. If they give you a number, ask whether their fee has already been taken off.

FAQ

Questions, answered

What does company data partnership mean?

An AI firm paying a business for a license to copies of its work records. The business keeps its originals.

Do anonymized and de-identified mean the same thing?

Not reliably. NIST avoids the word anonymization because it is used in different ways.

What does an AI data broker do?

Links a business with a data buyer for a fee, often paid by the buyer.

What is the difference between perpetual and irrevocable?

Perpetual licenses have no end date. Irrevocable ones cannot be withdrawn by you.

Find the buyers that fit your company

Answer a few questions about your size, tools and history. We show which buyers want a company like yours. Free, no files.

Match my company What is my data worth?

Keep reading

Sources

Every number on this page links back to one of these.

Spot something wrong? See our corrections policy. Some buyers pay us a referral fee. How we make money.