Stored Credential Framework: Card-on-File Rules
Payments Glossary

What Is a Stored Credential? Card-on-File Rules Explained

Learn how the stored credential framework works, from customer consent to CIT and MIT flags, so saved-card charges see fewer declines and disputes.

On this page 7 sections
  1. How the Stored Credential Framework Defines a Saved Card
  2. Customer-Initiated vs. Merchant-Initiated Transactions
  3. COF Consent Requirements: What to Get Before You Save a Card
  4. Example: Saving a Card for No-Show Fees at a Salon
  5. How Proper Flagging Improves Approvals and Disputes
  6. How Card-on-File Transaction Rules Play Out by Industry
  7. Audit Your Recent Card-on-File Charges This Week

Key takeaways

  • Saving a card is a transaction event of its own. Flag the first charge or zero-dollar verification as the one that stores the credential, and keep the identifier the issuer returns.
  • Every later charge is either customer-initiated or merchant-initiated. That label tells the issuer whether to expect the cardholder to take part.
  • Consent you can't produce is consent you don't have, so keep the agreement next to the card reference.
  • Indicators such as "no-show" belong to specific business types. Ask your processor which ones your merchant category can send.

A stored credential is a card number, or a token that stands in for it, that a business saves so it can charge the customer again later. The stored credential framework is the set of card network rules that governs the practice. You need the customer's permission before you save the card. You also need to label the transaction that stores it and every charge that follows, so the issuing bank knows what it is looking at. Follow the framework and your card-on-file charges look like agreed payments from an existing customer. Skip it, and the same charges can look like a stranger keying in a stolen card number.

This matters most for businesses that charge a card when the customer isn't at the counter. Subscription businesses renew plans while the customer sleeps. Hotels settle minibar charges after the guest has left. Salons collect no-show fees from clients who never arrived, and home care agencies bill private-pay families for last week's visits.

How the Stored Credential Framework Defines a Saved Card

Visa and Mastercard each publish their own version of these rules. At a practical level they work the same way, and they are network rules, not law. Your processor and gateway control how you send the required data. Field names and report columns therefore differ from one setup to the next, even though the underlying rule doesn't.

A card becomes a stored credential when you or your payment provider keep it for future use. A card keyed once for a phone order and never saved doesn't qualify. One saved in your gateway's vault after a customer checks "save my card for next time" does, and so does a card a customer authorizes for monthly billing.

What the rules care about most is the moment of storage. The first transaction on a newly saved card should be flagged as the one that establishes the stored credential. Ideally the customer takes part in it, in person or at online checkout, so you can collect the card security code and run 3-D Secure authentication where you use it. Sometimes you're saving a card without owing a charge yet, as with a booking you'll only bill on a no-show. In that case a zero-dollar authorization to verify the card can serve as the initial transaction.

The issuer's response to that first transaction includes a reference identifier. Later charges on the stored card send it back, which ties each new charge to the original setup the customer approved. Most modern gateways capture and resend it automatically when you charge a saved card through their vault. Trouble usually starts when staff work outside the vault, copying card numbers into booking notes or spreadsheets and re-keying them later. That habit also breaks PCI DSS. The security code can't be kept after authorization, and a card number sitting in a notes field is cardholder data in an unprotected system.

Customer-Initiated vs. Merchant-Initiated Transactions

Every charge on a stored credential after the first falls into one of two groups, and the flag you send tells the issuer which one.

A customer-initiated transaction (CIT) is one the cardholder actively starts. A shopper who picks a saved card at checkout is making one. So is a client who calls and says "put it on the card you have." Because the customer is participating, the issuer can reasonably expect authentication data such as the security code or a 3-D Secure result where it's available.

A merchant-initiated transaction (MIT) is one you start under an agreement the customer already gave you. The customer doesn't take part in that specific charge. The networks recognize several kinds:

Type of chargeWho starts itTypical exampleKey term to state up front
RecurringMerchantMonthly gym membership or software planAmount and billing frequency
InstallmentMerchantEquipment paid off in monthly partsTotal and payment schedule
Unscheduled card on fileMerchantHome care billed weekly for hours workedHow the amount is calculated
Industry-specificMerchantHotel incidentals after checkoutThe event that triggers the charge
Customer-initiated card on fileCustomerReturning shopper pays with a saved cardPermission to store the card

Merchants regularly misread the industry-specific category. The networks tie indicators such as "no-show" and "delayed charge" to particular business types, with lodging and vehicle rental as the classic cases. A salon or clinic usually can't send the hotel no-show indicator just because the event looks similar. Instead it will typically send an unscheduled card-on-file MIT, backed by a clear written agreement. Ask your processor which indicators your merchant category allows. For the full mechanics, see our breakdown of merchant-initiated vs. customer-initiated transactions.

The networks expect you to get the cardholder's agreement before you store a card and to be able to produce that agreement later. Exact wording varies by network and transaction type, so confirm current terms with your processor. A workable consent record covers these points in plain language:

  • That you will store the card, and why.
  • Which kinds of charges you may make, such as recurring billing or event-triggered fees.
  • The amount, or how you'll calculate it when it varies.
  • When charges happen, or what event triggers them.
  • How the customer cancels or changes the arrangement.

Keep the record somewhere you can pull it up within minutes. A timestamped checkbox log from online booking works. So does a signed intake form scanned into the client's file. When a dispute arrives, you'll need to match that record to the specific card reference. If the terms change, for example through a price increase or a new billing day, the networks expect you to tell the customer before the new terms apply.

Separate from network rules, some states and federal regulators impose their own requirements on automatic renewals and recurring billing, typically around clear disclosure and an easy way to cancel. Those are law, and they apply whether or not your card flags are perfect. This is general information, not legal advice. Have a qualified attorney review your terms if you bill on a recurring basis. For the security side of saving cards, read our guide on how to store a customer's card on file safely and legally.

Example: Saving a Card for No-Show Fees at a Salon

Here is an illustrative setup. The numbers are assumptions chosen for the example, not benchmarks.

Assumptions: A salon books color appointments that run $160. Its posted policy charges half the booked service, or $80, when a client misses an appointment without a day's notice. Clients book online.

Done within the framework:

  1. At booking, the client accepts a policy that says in plain words that the salon will store the card and may charge the no-show fee.
  2. Next, the booking system runs a zero-dollar verification flagged as the initial stored-credential transaction, with the security code. The gateway keeps both the card token and the issuer's reference identifier.
  3. When the client doesn't show, the salon charges the fee as a merchant-initiated, unscheduled card-on-file transaction, or whichever indicator its processor specifies for the category. That charge points back to the original identifier.
  4. A receipt goes to the client's email under a billing descriptor the client recognizes.

Done outside it: Someone at the front desk writes the card number in the appointment notes. After the no-show, a staff member keys it into the terminal with no security code and no stored-credential flag.

It's the same client and the same fee, but what reaches the issuer is very different. In the first case, the charge arrives with context saying this is an established customer on agreed terms. In the second, it looks like a keyed card-not-present sale with no authentication, which makes a decline more likely. Suppose the client then disputes the keyed charge as unauthorized.

The salon has no consent record tied to that card, so it is likely to lose. Losing means returning the fee and paying whatever chargeback fee the processing agreement sets. That fee comes from the processor, not the networks, and it varies. A salon can end up losing more on a disputed no-show than the fee was worth.

Salon owners setting this up can compare booking-friendly options on our salon and spa payments page.

How Proper Flagging Improves Approvals and Disputes

Issuers make fraud decisions in milliseconds with limited information. A correctly flagged MIT tells them that the cardholder agreed to an arrangement and that this charge belongs to it. Because the reference identifier points back to an authenticated setup, the issuer isn't judging the charge in isolation. Without those signals, a charge with no customer present and no security code sits in the riskiest category an issuer sees.

Flagging also shapes what happens after a decline. Network rules separate declines you may retry from declines you may not, and repeated retries on a hard decline can draw network attention and fees. A sound recovery process reads the decline code before deciding whether to retry or ask the customer for a new card. Subscription businesses call that process dunning, and our explainer on recovering failed subscription payments covers retry timing.

Stored cards also go stale as they expire or get reissued after fraud. Account updater services and network tokens for stored cards keep credentials current without asking every customer to re-enter a card.

How Card-on-File Transaction Rules Play Out by Industry

Subscriptions and Memberships

Free trials that convert to paid plans draw the closest scrutiny from networks and regulators alike. Your consent screen should state the post-trial price and the date billing starts, with cancellation instructions right beside them. Flag each renewal as a recurring MIT, never as a fresh keyed sale.

Hotels and Lodging

Lodging gets the most specific indicators of any category. A hotel can send incremental authorizations while the guest is in-house and delayed charges for minibar use or damage after checkout. Guaranteed reservations add a separate no-show indicator. Each one depends on terms the guest accepted at booking or check-in, which makes the registration card and reservation confirmation your consent records. Our hotel payments page covers property setups.

Home Healthcare and Staffing

Private-pay clients are often billed weekly for a variable number of visit hours. The agreement should state the hourly rate and the billing day, and it should explain that each charge depends on approved timesheets. Each weekly charge then runs as an unscheduled card-on-file MIT. Agencies billing from an office often use a virtual terminal with a secure card vault instead of keying cards from paper files. For agency-specific setups, see our home healthcare and medical staffing page.

Audit Your Recent Card-on-File Charges This Week

Pull a handful of charges you made on saved cards last month from your gateway or processor report. Look for a stored-credential or merchant-initiated flag on each one, and confirm that it references an original transaction. Then try to find the customer's consent for every charge. If you can't do that quickly, a dispute will expose the same gap. Where any check fails, ask your processor which fields your payment gateway sends and which indicators your merchant category allows. For a second look at your setup, request a free quote from MannyPay, and a specialist will follow up with you.

About the author

Adrian Reyes

Adrian Reyes is Senior Payments & Fintech Editor for Beyond the Swipe, covering credit card processing, merchant accounts, payment technology and processing fees.

This article is for general educational purposes and is not legal, tax or financial advice. Card network rules and state laws change; confirm current requirements for your business.

Related reading