DataPacts

Order, Booking and Transaction Records as AI Training Data

Checked

At a glance

A company can license its order, booking and payment history to AI developers, and several acquirers name transaction, order, billing and shipment tables on their own pages, but card-data rules and consumer privacy law decide which fields may leave the company.

Independent reference. Every buyer is covered on the same terms, every figure carries its source and check date, and no buyer pays to be listed or placed higher.

Request an independent introduction

Tell us what data your company holds. We reply with the buyers that fit and what their terms leave out. Your details reach a buyer only after you agree to a named introduction.

IndependentYour details stay with us until you agreeNot selling is a valid outcome

Mercor states a deal range to the company of USD 80,000 to 900,000 for years of such tables (listing published October 20, 2026, feed re-read November 16, 2026).

Order and booking tables are one record type within AI training data licensing. No per-record price for them is on record, and a deal is a license of a copy for a defined use, not an outright sale.

Program wording was checked on November 9, 2026 and the Mercor feed and the PCI Security Standards Council page on November 16, 2026; how figures are checked explains the method.

What Order, Booking and Payment Tables Hold, and Which Parts Buyers Want

Transaction records are the structured history of what a business sold, booked, charged, refunded and shipped, and buyers ask for years of these tables in one consistent schema rather than for single receipts.

The table compares each record family with the programs that name it and the exclusion or rule that applies, as of 2026-11-16 (program wording checked 2026-11-09, Mercor feed re-read 2026-11-16).

Record family Programs naming it, in their own words Stated exclusion or rule that applies Source
Orders and line items "transaction, order, inventory or shipment tables" (Mercor Payments & Fintech) personal information rules of the customers' states and countries api.data.mercor.com
Bookings and reservations none of the programs on record names bookings Spirit consumer privacy ombudsman: transactions and PNR excluded in any form Dkt. 1684
Payments, billing and settlement Stripe among the systems named (Troveo); "billing and operations" (Mode); billing records (Mercor SaaS & Subscription) payment card and financial account data excluded unless the Order Form allows it (micro1 s.5.7(e)); PCI DSS troveo.ai; data.mode.inc; micro1.ai
Refunds, returns and inflight or add-on purchases none names them excluded in the Spirit record Dkt. 1684
Inventory, shipment and dispatch logs "procurement, shipping, dispatch logs", "product catalogs and inventory workflows" (Handshake AI); "logistics and fulfillment processes" (micro1); "ERP, WMS" (Sell Business Data) none stated joinhandshake.com; micro1.ai; sellmybusinessdata.com
Store and commerce platform histories Shopify (Troveo, Corpus Bridge) platform terms not recorded troveo.ai; corpusbridge.com
Service orders, quotes and scheduling "service calls, quotes, job notes ... dispatch"; "scheduling, billing and coding" (HUMXN) none stated humxn.io

The records sit among the data types AI companies buy.

Card data and personal identifiers are what buyers exclude or strip, and the sequence of events is what they want.

Acquirers and Intermediaries That Name Order, Payment and Commerce Tables

The programs that name transaction records split into acquirers that take a license and pay the company, and intermediaries that look for a buyer and license or broker on the company's behalf.

The table compares ten of them by the records they name and the condition each states, as of 2026-11-16, every value being the company's own statement checked 2026-11-09, in alphabetical order.

Company (role) Transaction records named Stated condition or exclusion Source
AfterQuery (licensee) "finance records" customer identities masked; payroll, privileged legal material and credentials never used afterquery.com
Corpus Bridge (broker; brokerage partner approaches labs) QuickBooks, Xero, NetSuite, SAP, Shopify "Established businesses" corpusbridge.com
Handshake AI (company data program) "procurement, shipping, dispatch logs"; "product catalogs and inventory workflows" Retail & e-commerce and Logistics & supply chain among 11 industries; "scoped, de-identified, and confidential" joinhandshake.com
HUMXN (company data program) service calls, quotes, dispatch; scheduling and billing retail & e-commerce and hospitality & restaurants among 24 industries; exclusive or non-exclusive humxn.io
Mercor (company data program) transaction, order, inventory or shipment tables see the listing table below api.data.mercor.com
micro1 (company data program) "logistics and fulfillment processes" payment card or financial account data excluded unless the Order Form allows it (Data License Master Terms s.5.7(e)) micro1.ai; micro1 terms
Mode (data vendor) "billing and operations" company size rules differ across its own pages data.mode.inc
Replay (licensee) "analytics", "product logs", "financial records" only de-identified data; "not to foreign adversaries or through brokers" replay.ai
Sell Business Data (seller-side adviser) "ERP, WMS, CRM, accounting" "licenses one copy for a defined purpose" sellmybusinessdata.com
Troveo (marketplace, licenses on the seller's behalf) QuickBooks, Stripe, NetSuite, Shopify no OAuth access or install required; anonymized package with a manifest troveo.ai

Programs that run a standing intake are compared under company data partnership programs, which covers the other record types. A cell reading "none stated" means no exclusion is published, not that the records are cleared, and an exclusion binds that program only.

Mercor's Payments and Commerce Listing: Fit, History and Stated Range

Mercor publishes one listing for transaction tables, and it names the companies that fit, the history it wants and a deal range to the company of USD 80,000 to 900,000 (Mercor states, checked November 9, 2026, feed re-read November 16, 2026). The table lists each field of the listing as of 2026-11-16.

Field Mercor states Source
Listing and publication date Payments & Fintech Company, published 2026-10-20 api.data.mercor.com/data-acq/api/v1/listings
Summary "License years of transaction, order, inventory or shipment tables from a payments or commerce business." .../listings/payments-fintech
Companies that fit payment processor, merchant acquirer, POS, e-commerce platform, logistics or market-data company .../listings/payments-fintech
History and format 3 to 5+ years; consistent schema; ideally refreshed on an ongoing basis .../listings/payments-fintech
Requirements and not-a-fit lists empty in the feed .../listings/payments-fintech
Deal range to the company USD 80,000 to 900,000 .../listings/payments-fintech

The program behind the listing is described in the Mercor data buyer profile. Mercor's program page names retail and e-commerce and logistics among its target industries (checked November 9, 2026).

The range is a whole-deal figure, never a price per order, record or gigabyte, and the empty requirement lists mean nothing beyond the description is stated.

Card Numbers and Payment Credentials: PCI DSS Scope and Program Exclusions

Payment card data stays out of an AI training license in practice: PCI DSS applies to every entity that stores, processes or transmits cardholder data, and at least one acquirer's master terms exclude card and account data unless the order form allows it.

The table compares the rules and terms as of 2026-11-16, quoting each issuer's own text.

Rule or term What it says Who it binds Source
PCI DSS (PCI Security Standards Council) applies to "Entities that store, process, or transmit cardholder data (CHD) and/or sensitive authentication data (SAD) or could impact the security of the cardholder data environment (CDE)" merchants, processors, acquirers, issuers and service providers pcisecuritystandards.org (checked 2026-11-16)
micro1 Data License Master Terms s.5.7(e) the licensor warrants that the data contains no payment card data or financial account data, "Except as may be otherwise agreed in an Order Form" licensors to micro1 micro1.ai (last updated 2026-11-12, re-read 2026-11-16)
AfterQuery data partnerships credentials never used; customer identities masked sellers to AfterQuery afterquery.com (checked 2026-11-09)
PCI DSS version and storage rules for sensitive authentication data not stated here n/a n/a

A retailer or booking engine that takes cards is inside the standard whether or not it licenses data. micro1's grant is nonexclusive, irrevocable and perpetual (s.1.2, re-read November 16, 2026), so a card number delivered by mistake cannot be recalled.

Bank, fintech and open-banking account data fall under separate rules, covered under accounting, ERP and financial records.

Consumer Privacy Rules for Purchase and Booking Histories

Five bodies of law decide whether purchase and booking histories may be licensed: California's CCPA, other state privacy acts, the FTC Act, the GDPR and the Justice Department's bulk data rule. This section is information, not legal advice.

California: Sale, Opt-Out and Deidentified Order Data

Licensing customers' order history to an AI developer for money is a sale of personal information under the CCPA, so California customers keep the right to opt out of it (Cal. Civ. Code 1798.120, in force since January 1, 2020, statute version effective January 1, 2026).

Order data counts as deidentified under Cal. Civ. Code 1798.140(m) only when the business takes reasonable measures against association with a person, publicly commits not to reidentify it and contractually binds recipients to the same.

A risk assessment comes before any sale (Cal. Code Regs. tit. 11, sections 7150 to 7157, in force January 1, 2026), and California CCPA rules for AI training data sets out the duties.

Other US State Privacy Acts: Monetary Sale and Sensitive Data

Other state privacy acts reach the same records with their own definitions, and some forbid selling parts of a purchase history outright. Virginia defines a sale as an exchange for monetary consideration and requires consent for sensitive data (Va.

Code 59.1-575 and 59.1-578, in force January 1, 2023), and Florida bars covered businesses from selling sensitive data without prior consent (Fla. Stat. 501.715, in force July 1, 2024).

Maryland bars selling the personal data of a consumer the controller knew or should have known is under 18 (Md. Code Com. Law 14-4707(a)(5), in force October 1, 2025), a rule that reaches order histories of minors; see also Maryland's ban on selling sensitive data.

FTC Section 5: Privacy Promises Made at Checkout

A company that told customers at checkout how their data would be used cannot quietly widen that promise to cover AI training, because the FTC treats surreptitious, retroactive policy changes as potentially unfair or deceptive (FTC staff post of February 13, 2024, under 15 U.S.C. 45).

Past FTC orders have required deletion of models and algorithms built on unlawfully obtained data (FTC staff post of January 9, 2024).

GDPR: Reusing Order Records for a New Purpose

Order records of people in the EU were collected to fulfil a purchase, so licensing them for AI training is further processing that must pass the compatibility test of GDPR Article 6(4) or rest on its own legal basis (Regulation (EU) 2016/679, applies from May 25, 2018).

The factors are the link between the purposes, the context of collection, the nature of the data, the consequences and the safeguards; the test is set out under purpose compatibility under Article 6(4).

Bulk Identifiers and Foreign Recipients

A transaction table that carries covered personal identifiers of more than 100,000 US persons, or personal financial data of more than 10,000, is bulk data under the Justice Department's rule, which restricts giving countries of concern access (28 CFR 202.205, in effect since April 8, 2025).

Booking and Transaction Systems Kept Out of a Pending Bankruptcy Data Sale

The one AI data sale on a public docket that touches booking records leaves them out: the consumer privacy ombudsman's report in the Spirit Airlines chapter 11 lists consumer transaction systems, PNR among them, as excluded in any form (Dkt. 1684, October 5, 2026).

The excluded systems are transactions and PNR, refunds, inflight purchases and Wi-Fi sales. At the auction of August 14, 2026, the debtors named Google LLC successful bidder at USD 10,000,000 and Mercor.io Corporation alternate bidder at USD 7,500,000 (Notice of Auction Results, Dkt. 1463), and the court had not approved the sale as of November 9, 2026, with the hearing set for November 17, 2026 (Dkt. 1779); the record is summarized under Spirit Airlines' data archive sale (In re Spirit Aviation Holdings, Inc., No. 25-11897 (SHL), Bankr. S.D.N.Y.).

The sale is pending, so the terms may change; the AI data licensing deal tracker records any ruling with its docket entry and date.

Preparing Transaction Tables: Schema, History, Identifiers and Exclusions

Buyers state what they will not take and who removes identifiers, and those statements decide which columns of an order table survive into a dataset (all checked November 9, 2026).

  • Consistent schema and 3 to 5+ years of history: Mercor's listing asks for both.
  • Card and account fields out: micro1 s.5.7(e) excludes payment card and financial account data unless the Order Form allows it.
  • Customer identities masked: AfterQuery states that identifiers are removed or pseudonymised.
  • De-identification by the program: Handshake AI states that all data is de-identified, and Replay takes only de-identified data.
  • Export terms of the commerce platform: the merchant agreement is the place to check, because no platform's terms on AI training are recorded here.

AfterQuery states that no process removes all risk of re-identification on its own; methods are compared under de-identification before licensing.

Licensing Transaction Records: Benefits, Risks, Downsides and Alternatives, Including Not Licensing

Licensing transaction history can earn from records the company already keeps, but the records describe customers, privacy law attaches to them, and a model trained on them cannot be untrained.

Benefits

  • Mercor states a whole-deal range for payments and commerce tables (see the listing section above), a claim that depends on the company's data.
  • Replay states an upfront payment plus a perpetual revenue share (checked November 9, 2026).

Risks

  • A sale of California customers' personal information triggers opt-out rights and a risk assessment (Cal. Civ. Code 1798.120).
  • A privacy promise made at checkout can bar the reuse (15 U.S.C. 45).
  • Bulk identifiers above 100,000 US persons bring the Justice Department rule into play (28 CFR 202.205).

Downsides

  • micro1's grant is nonexclusive, irrevocable and perpetual (s.1.2).
  • AfterQuery states that no process removes all re-identification risk on its own.
  • No per-record price is on record, so an offer has no benchmark.

Alternatives

  • A de-identified subset, which the CCPA treats differently when the three conditions of Cal. Civ. Code 1798.140(m) are met.
  • Workflow documentation without customer rows, as micro1 describes for logistics and fulfillment processes.
  • A non-exclusive license to one buyer, as HUMXN describes.
  • Not licensing at all, which keeps every right in the records.

Not licensing is a valid outcome.

Neighboring Record Types and the Seller Steps for Order Data

This page covers a company's own order, booking and payment tables; ledgers, market data, customer records and the retail and transport seller views each have their own page.

The seller steps for order data follow the general process:

  1. Read the complete seller guide for the order of work.
  2. Check records barred across sectors and remove them from the set.
  3. Compare every license clause before signing.
  4. Review upfront, milestones, revenue share for how buyers pay.

Frequently Asked Questions

How much do AI companies pay for transaction or order data?

No per-record price for transaction or order data is on record. Mercor states a whole-deal range for its payments and commerce listing, a claim and not a payment. The public price points are compared under what company data is worth.

It depends on the privacy promise made to customers and the law of each customer's state or country. A sale needs an opt-out in California and a lawful basis in the EU. The duties are covered under customer consent before licensing.

Can card numbers stay in a licensed transaction dataset?

Card numbers do not stay in practice. PCI DSS applies to entities that store, process or transmit cardholder data, and micro1 s.5.7(e) excludes card and account data unless the Order Form allows it.

Is de-identified order data still personal information?

Not under the CCPA when the three conditions of Cal. Civ. Code 1798.140(m) are met: reasonable measures against association, a public commitment not to reidentify and binding contracts with recipients. GDPR anonymity is a separate test.