Don't Do It Yourself

How to Sell Payments and Fintech Data to AI

A payments or fintech company sells its data to AI by licensing years of transaction, order, inventory or shipment tables to a buyer like Mercor.

Mercor pays payments and fintech companies $80,000 to $900,000. It wants 3 to 5 or more years of history in a consistent schema.

The hard part is legal, not technical. Consumer transaction data falls under the Gramm-Leach-Bliley Act, and card numbers fall under PCI DSS.

For the full process across every industry, start with our guide to how to sell your company’s data to AI companies.

Key takeaways

  • Mercor’s payments and fintech listing pays $80,000 to $900,000 for transaction, order, inventory or shipment tables.
  • It wants 3 to 5 or more years of history in a consistent schema, ideally refreshed on an ongoing basis.
  • Regulation P bars sharing consumers’ nonpublic personal information with an outside company unless they had notice and a chance to opt out.
  • Data that doesn’t identify a consumer, such as blind or aggregate data, isn’t covered by that definition.
  • PCI DSS requires card numbers to be unreadable wherever they’re stored, so strip or tokenize them first.

What fintech data do AI companies buy?

AI companies buy long, consistent tables of real transactions: payments, orders, inventory and shipments. Mercor wants 3 to 5 or more years of history from processors, acquirers, point-of-sale systems, e-commerce platforms, logistics firms and market-data companies.

Mercor’s payments and fintech listing, read November 9, 2026, names the companies it wants:

  • Payment processors
  • Merchant acquirers
  • Point-of-sale companies
  • E-commerce platforms
  • Logistics companies
  • Market-data companies

It asks for a consistent schema and, ideally, data refreshed on an ongoing basis. A clean table that keeps updating is worth more than a one-off dump.

Mercor’s listing doesn’t set a company size, and it doesn’t restrict location. Its referral page says the full referral fee applies to companies with 30 or more employees.

How much is fintech data worth to AI companies?

Mercor’s payments and fintech listing pays $80,000 to $900,000. That floor is higher than most of Mercor’s finance listings, which start at $25,000 to $50,000. The price depends on years of history, schema consistency and whether the data keeps updating.

Here’s how it compares with Mercor’s other finance listings, read November 9, 2026:

Mercor listing Company earns Size bar
Payments and fintech $80,000 to $900,000 None stated
Investment firm $50,000 to $800,000 5 to 200 employees
Accounting and tax firm $25,000 to $865,000 5 to 200 employees
Hedge fund or quant firm $25,000 to $1.1 million 5 to 500 employees

Mercor also shows an $80,000 referral fee for this listing, including a $10,000 bonus. That’s paid to someone who refers a company, not to the company.

Our guide to how much company data is worth to AI compares published payouts across industries.

Does the Gramm-Leach-Bliley Act let you sell transaction data?

Only data that doesn’t identify a consumer, or with notice and opt-out. Regulation P bars a financial institution from disclosing consumers’ nonpublic personal information to a nonaffiliated company unless each consumer got a privacy notice, an opt-out notice and a fair chance to opt out.

The rule is in 12 CFR 1016.10. An AI data buyer is a nonaffiliated third party, so the notice and opt-out conditions apply to anything that counts as nonpublic personal information.

The definition, in 12 CFR 1016.3, is broad. It covers information a consumer gives you, information from any transaction with them, and anything else you learn while serving them.

It also has a gap that matters here. Personally identifiable financial information doesn’t include “aggregate information or blind data” without personal identifiers such as account numbers, names or addresses.

Three more points shape what you can license:

  • Consumers only. Regulation P covers products for personal, family or household use. Business and commercial accounts are outside it.
  • The service-provider exception doesn’t fit. It requires a contract barring the recipient from using the data except for your purposes. A buyer training models for others isn’t doing that.
  • Some firms use a different rule. Regulation P excludes broker-dealers and registered investment advisers, which follow the SEC’s Regulation S-P instead.

The service-provider wording is in 12 CFR 1016.13, and the scope limits are in 12 CFR 1016.1.

California’s CCPA steps aside for GLBA data. Civil Code 1798.145(e) exempts personal information collected under GLBA, except for the data breach lawsuit section, 1798.150.

Data you collect outside GLBA, such as marketing lists or e-commerce order records, can still fall under the CCPA.

Our guide to selling investment firm data to AI covers Regulation S-P for advisers and wealth managers.

Does the FTC Safeguards Rule apply?

Yes, for non-bank financial companies. The FTC’s Safeguards Rule covers businesses such as check cashers, companies that wire money for consumers and tax preparers. It requires an information security program, and an FTC notice within 30 days of a breach affecting 500 or more consumers.

The FTC’s guide to the rule says what counts is the activity, not the label. A company doing financial activities is covered whatever it calls itself.

For a data deal, the rule matters in two places. Your export process is part of your security program, and a breach at the buyer can still trigger your notice duty.

Write the buyer’s security standards, a breach-notice deadline and a deletion date into the license.

Can you license card data under PCI DSS?

Not the card numbers. PCI DSS requirement 3.5.1 says the primary account number must be unreadable anywhere it’s stored, and sensitive authentication data such as CVV codes and PINs can’t be stored after authorization at all. Strip or tokenize card data before any export.

The PCI Security Standards Council glossary defines cardholder data as, at minimum, the full card number. Name, expiration date and service code count too when stored with it.

A PCI SSC FAQ confirms only the card number itself has to be made unreadable. The other fields still need PCI protection inside your card data environment.

For a license, the simplest rule is that card data doesn’t leave. Replace card numbers with tokens or a hashed key that only links transactions to each other.

A model learns spending patterns from a stable token as well as from a real card number. You give up nothing by removing it.

Do bank partner and processor contracts allow it?

Check before you assume. Sponsor banks, processors, card networks and merchants all have contracts that can limit what you do with transaction data. Mercor’s other listings rule out data restricted by customer contracts, and the same logic applies here.

Read these agreements for data-use, confidentiality and “no resale” clauses:

  • Your sponsor or partner bank agreement
  • Processor and acquirer agreements
  • Card network and scheme agreements
  • Merchant agreements, if you process for merchants
  • Your published privacy notice

Merchant data is the trap for processors and platforms. The transactions may run through your systems, but your merchant agreement decides whether you can license them.

If a contract is silent, ask the counterparty in writing. A short consent letter from a bank partner is cheaper than a dispute after the deal.

How do you de-identify transaction data?

Remove everything that points to a consumer, then check the combinations. Drop names, account and card numbers, addresses, emails and device IDs, replace account keys with consistent tokens, and coarsen exact timestamps, locations and rare amounts that could single someone out.
Field What to do
Card numbers (PAN) Remove, or replace with a token that can’t be reversed
CVV, PIN and track data Shouldn’t exist in storage at all; confirm none are in the export
Names, emails, phone numbers Remove
Account and customer IDs Replace with consistent tokens so transactions still link
Street addresses and exact geolocation Generalize to region or 3-digit ZIP
Exact timestamps Round, or shift consistently
Merchant names for small merchants Generalize to category if the merchant agreement requires it
Unusually large or rare amounts Bucket into ranges

Keep the tokens consistent across tables. A model needs to see that one account placed 40 orders, and a stable token shows that without saying who it was.

Mercor says it signs an NDA, runs the export and removes or protects personal information. Someone on your compliance team should still read a sample before delivery.

Our guide to anonymizing data for AI training covers the tools and standards in more detail.

Who buys payments and fintech data?

Mercor is the only buyer publishing a payout range for payments and fintech companies. micro1 lists finance and accounting processes among the operational data it buys from companies with 30 or more staff.
  • Mercor: transaction, order, inventory and shipment tables, 3 to 5 or more years. Pays $80,000 to $900,000.
  • micro1: finance processes, reconciliations, approvals and reporting workflows, for companies with 30 or more full-time staff. Its tiers run from $100,000 to over $3 million. You can apply to micro1 here.

Our list of companies that buy business data for AI has each buyer’s terms.

What are the steps for a fintech company, start to finish?

A fintech company licenses its data in seven steps: inventory the tables, sort consumer from business data, read partner contracts, strip card data, de-identify the rest, have counsel review the license, then check a sample before delivery.
  1. Inventory the tables. Note years of history, schema changes and how often each one refreshes.
  2. Sort consumer from business data. Consumer accounts fall under Regulation P. Business accounts don’t.
  3. Read your partner contracts. Bank, processor, network and merchant agreements, plus your privacy notice.
  4. Strip card data. No card numbers, CVVs, PINs or track data in the export.
  5. De-identify the rest. Tokens for accounts, coarser times and places, buckets for rare amounts.
  6. Have counsel review the license. Our AI data licensing agreement checklist lists the clauses to check, including a ban on re-identification.
  7. Check a sample. Try to identify a customer from 50 de-identified transactions before anything ships.

Start with business-account data and internal records you own, such as fraud rules and reconciliation logs. Neither needs consumer opt-outs.

Common mistakes when selling fintech data to AI

The most common mistake is treating de-identification as optional because the buyer promises to do it. These are the others:

  • Leaving card numbers or CVVs in an export table
  • Treating a data sale as a service-provider disclosure under Regulation P
  • Licensing merchant transactions without reading the merchant agreement
  • Forgetting the sponsor bank’s data-use clause
  • Keeping exact timestamps and locations that single out a customer
  • Reading Mercor’s $80,000 referral fee as what the company gets paid

Frequently asked questions

Can a fintech company sell its transaction data to AI companies?

Yes, if the data is stripped of anything that identifies a consumer and your bank, processor and merchant contracts allow it. Mercor pays payments and fintech companies $80,000 to $900,000.

How much do AI companies pay for payments data?

Mercor’s payments and fintech listing pays $80,000 to $900,000. It wants 3 to 5 or more years of transaction, order, inventory or shipment tables with a consistent schema.

Does the Gramm-Leach-Bliley Act stop you selling transaction data?

It limits it. Regulation P bars disclosing consumers’ nonpublic personal information to a nonaffiliated company unless they got a privacy notice and a chance to opt out. Data with no personal identifiers falls outside that definition.

Can you include card numbers in a data license?

No. PCI DSS requires the primary account number to be unreadable wherever it’s stored, and sensitive authentication data can’t be stored after authorization at all. Strip or tokenize card data before export.

Does the CCPA apply to fintech data?

Partly. The CCPA doesn’t apply to personal information collected under the Gramm-Leach-Bliley Act, except for its data breach lawsuit section. Data you collect outside GLBA, such as marketing or e-commerce records, can still be covered.

What fintech data is safest to license?

Business-account transactions with no consumer identifiers, aggregated or tokenized consumer data, and internal operational records such as fraud rules and reconciliation logs you own outright.

Last updated: November 9, 2026

Disclosure: some links on this page are affiliate links. If you sign up through one, we earn a commission at no extra cost to you.

This isn’t legal advice. Have counsel review any data deal, and your compliance team review anything touching consumer or card data, before you sign.

Don't Do It Yourself
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.