Payments and POS for a recovery studio: the spread, the failed charge, and the floor

Every studio platform comparison starts at the subscription tier and usually ends there, because the subscription tier is the one figure a vendor prints in a size you can read. It is rarely the largest line. A studio running memberships, drop-ins and a small retail shelf moves nearly all of its revenue across a card, and the spread between what the card networks charge and what your platform bills you is applied to every dollar that crosses the counter. A subscription is a fixed cost you can put in a spreadsheet and compare. Processing is a variable cost on your entire top line, and it grows with you.

Seventeen studio and wellness platforms were checked in September 2026 for what they publish about payments. Five publish a card processing rate at all, and only four of those put it on a pricing page. Three name Stripe as the processor that actually moves the money. Four market their own payments product as a headline feature while publishing no rate for it. The rest publish neither a rate nor a processor. That is the disclosure shape of the category, and its practical meaning is that the most decision-relevant number in your software contract is one you have to extract in a sales call rather than read on a page.

Rates and tier prices are deliberately absent from what follows; the shape of each line, and the question that makes a vendor state it, is the part that keeps.

The spread is the price, and five vendors print it

Card acceptance has three cost components and only one is negotiable. Interchange is set by the bank that issued the member's card and varies by card type, so a premium rewards card genuinely costs you more to accept than a basic debit card. Assessments are set by the card network, are small, and are identical for everyone. The markup on top is your processor's revenue, and that markup is the spread. A platform that bundles processing is selling you software and a markup in one price, and it has every incentive to present the combination as a single convenient number rather than as two lines you could shop separately.

Of the seventeen platforms checked in September 2026, five publish a rate, and they publish it in five different shapes. One prints its in-person and virtual card rates directly on its pricing page[1]. One publishes its own percentage stated explicitly as an amount charged on top of the underlying Stripe fees[2], which is the most legible of the five because it separates the platform's take from the network's cost. One publishes a percentage range plus per-transaction fees that vary by payment method[3], which is closer to how an acquirer actually prices and closer to what your monthly statement will look like. One publishes a processing-fee table broken out by country[4], which is the shape that matters if you operate across a border and the shape that tells you least if you do not. And one publishes a flat card-present rate, but not on its pricing page: it appears on a page comparing the product against a named competitor[12], where the rate is doing argumentative work. Read that fifth one for what it omits, because a card-not-present rate is the one a studio selling memberships online will actually pay most often.

The rest publish nothing, and four of those market their own payments product prominently while leaving the rate to a conversation. The absence is a decision rather than an oversight: a published rate is a number a competitor can undercut and a customer can benchmark. Read an unpublished rate as a question you are obliged to ask in writing before signing, and ask it in the form that produces a usable answer. Not what is your rate, but what will my total effective rate be across my actual card mix, including every fee on the statement.

• Interchange: set by the issuing bank, varying by card type and by how the card was presented. Your average matters far more than any headline rate, because your mix is not the vendor's example.
• Assessments: set by the card network, small, fixed, identical for every merchant. Nothing to negotiate.
• Processor markup: the negotiable component, and the one a bundled platform price is designed to obscure.
• Per-transaction fees: a few cents on every authorisation, which on a studio selling many small drop-ins is a larger share of the total than the percentage alone suggests.
• Ancillary fees: monthly minimums, gateway, PCI, batch, chargeback and early termination fees. Ask for the full schedule, then compute total fees divided by total card volume.

Integrated processing versus bringing your own

Integrated processing means the platform is your processor or resells one, so every payment object in the system is native: a membership charge, a retail sale, a late-cancellation fee and a gift card land in one ledger and reconcile without anyone touching a spreadsheet. Bringing your own means you keep a merchant account you negotiated and the platform connects to it. On a normal Tuesday the functional difference is small. The difference in leverage is not small at all.

With integrated processing, the cost of leaving your processor becomes the cost of leaving your software, and the vendor is aware of that. Your renewal stops being a conversation about a rate and becomes a conversation about migrating an entire operation, so the renegotiation that would be routine on a standalone account is the one nobody opens. With a separate merchant account the rate is its own contract on its own renewal cycle, and you can put it out to bid without touching a booking record.

The argument the other way is also serious. Platforms increasingly build features that assume they control the payment rail: card on file enforcing a no-show policy, automatic membership retries, tipping, split payments, refunds initiated from inside the appointment record. Three platforms in the set name Stripe as the processor behind their payments, a useful middle position where the money moves on a rail you can recognise but the account is provisioned through the platform. At low volume the integrated path is usually correct, because the staff hours saved on reconciliation outweigh a spread you have no volume to negotiate. Find your crossover by computing annual processing cost and asking an independent acquirer to quote the same mix.

The failed recurring charge is the churn you never see

In a membership business the largest silent churn channel is not the cancellation form. It is a card that expired. The member decided nothing: their bank reissued the card, or a billing address changed, or a daily limit was hit, or an issuer's fraud model declined a recurring charge it had approved eleven times before. The charge fails, the platform retries on whatever schedule it shipped with, the retries fail, the membership lapses, and the member discovers it weeks later when their check-in stops working, assuming they come back at all.

This is worse than an ordinary cancellation in three ways. It is invisible in reporting, because a lapsed membership and a cancelled membership look identical in a revenue report unless somebody deliberately built the distinction. It is invisible at the desk, because nobody working Tuesday afternoon knows that Tuesday is the day this member's card stopped working. And it is recoverable in a way a real cancellation is not, because the member still wants the service and will fix it in under a minute if a person tells them. That makes it the cheapest recoverable revenue in the building and the easiest to lose by default.

So the first payments question to put to a platform is not about the rate. It is: what happens on the day a recurring charge fails, and who finds out. Ask to see the actual screen a staff member would see, in the interface they already have open. A platform that surfaces failed payments only inside a billing report nobody opens will quietly convert card failures into churn on your behalf.

Retry and dunning design a front desk can act on

Dunning is the sequence of retries and messages that runs between a failed charge and a closed account, and most platforms ship a default designed for a software subscription rather than for a business the customer walks into. That distinction is the whole opportunity: you have a recovery channel a software company does not, because the member appears in your building on a cadence you can predict, and a ten-second conversation at the desk resolves what four emails will not.

Two mechanics do most of the work. Account updater services, offered by the card networks and passed through by many processors, refresh a stored card automatically when the issuer reissues it, which removes a large share of expiry failures before they become declines. Ask explicitly whether your platform's processing includes it, because it is a rail-level capability a platform can enable or ignore and almost none advertise. The second is retry timing: retrying the same declined card an hour later mostly reproduces the decline, while spacing attempts across several days and across a payday boundary catches the insufficient-funds cases a same-day retry cannot.

• Separate a failed payment from a cancellation in every report you look at. If the platform cannot distinguish them, any churn figure you quote carries the defect.
• Put failed payments on the front desk check-in screen, not only in a billing report. The member's next visit is your best collection channel and it expires when they stop visiting.
• Space retries across days and across a payday, and cap the sequence. An unbounded retry loop teaches an issuer to flag your descriptor as suspicious.
• Send the update-card link in the channel the member actually reads, and make it work without a login. A password reset dropped into a payment recovery loses most of the recoveries.
• Decide the access policy during dunning before you launch. Day-one lockout converts a card problem into a cancellation; unlimited grace converts it into unpaid usage.

Consent and cancellation: what actually applies to auto-renewal

Recurring billing is a regulated thing to sell, and the federal position moved recently in a way many operator checklists still have wrong. The Restore Online Shoppers' Confidence Act is a statute, and its section 8403 is the live hook for an online negative-option charge: the seller must clearly and conspicuously disclose all material terms before obtaining billing information, obtain express informed consent before charging the account, and provide simple mechanisms for the customer to stop the recurring charges[9]. A membership sold through your booking site sits squarely inside that, and so does an upsell added during an online checkout. Watch for one substitution in particular, because it is common in operator write-ups: the requirement to take the account number directly from the consumer is section 8402, and section 8402 governs post-transaction third-party sellers rather than a studio charging for its own membership.

The 2024 negative-option rule commonly called click to cancel is a separate matter and gets misreported constantly. It was vacated and is not in force[11]; the Commission refers to its pre-2024 negative-option rule as the current one, and has since returned to an advance notice of proposed rulemaking on negative-option marketing[10]. Any checklist telling you to follow click to cancel as current federal law is out of date. The parts that went with it are narrower than most summaries imply: proof of consent retained for a set period, and cancellation offered in the same medium used to sign up. Both remain sound design on their own merits. The statute underneath still requires you to give the customer a simple way to stop the charges, so a cancellation flow built to frustrate is not something the vacatur made safe. State auto-renewal statutes impose their own requirements that a federal vacatur does nothing to change, and those are the rules most likely to reach a single-location studio.

The software consequence is testable: your platform has to store proof. When a member disputes a charge eight months later, the artifact that ends the argument is a timestamped record showing which terms were displayed and that the member affirmed them. Ask whether the platform retains a versioned copy of the membership agreement as presented at signup, rather than a boolean field that says accepted. A checkbox with no record of what sat next to it is not evidence of anything.

Chargebacks when the product was a session

A chargeback in a service business is a documentation problem far more than a fraud problem. The recurring patterns are a member who forgot about a recurring charge, a member who believes they cancelled, a member who says they never attended, and occasional card fraud on a drop-in purchase. Each is winnable or unwinnable based on what your system recorded at the time, which means the work happens months before the dispute arrives.

Two cheap fixes remove whole categories of dispute. The first is the billing descriptor: a member who sees an unfamiliar legal entity name on a statement calls their bank instead of calling you, and that call becomes a chargeback rather than a refund. Set the descriptor to the studio's trading name. The second is a pre-billing notice on annual or high-value recurring charges, which converts a surprise debit into an expected one. Then ask any platform whether dispute evidence is assembled automatically or by a staff member exporting screenshots at the end of a shift.

• The signed agreement as presented, with a timestamp and the version of the terms the member actually saw.
• The booking record, showing when the appointment was made and by whom.
• A check-in or door event proving attendance, the single strongest artifact in a did-not-attend dispute.
• The cancellation and no-show policy the member accepted, plus any notice you sent before the charge.
• The receipt, and the descriptor exactly as it appeared on the member's statement.

Surcharging, cash discounts, and who carries the fee

One platform in the set publishes customer-paid card-fee surcharging as a feature[7], which is a genuine product decision worth naming: it moves processing cost off your margin and onto the member's receipt. Whether you may use it is a legal and card-network question rather than a software question. Network rules impose registration, disclosure and cap requirements, several states restrict or prohibit the practice, debit is treated differently from credit, and the rules move. A platform advertising the capability is not telling you that you are permitted to switch it on, so that is a conversation with counsel in your state first.

The operational question is separate and quieter. A surcharge is visible at the moment of purchase, on a purchase a member repeats weekly, in a business whose economics depend on a habit you are building. A cash discount framed as a discount lands differently from a fee framed as a fee, even when the two are arithmetically identical. Plenty of operators who could surcharge decide the friction is not worth the recovered points and price processing into the membership rate instead.

The floor: terminals, retail, and what the desk needs at closing time

A recovery floor has payment needs a salon appointment book does not model well. Sessions are short and overlap. A member arrives for a contrast circuit, adds a compression slot because the room is free, buys an electrolyte sachet and leaves, and all of that should land as one transaction against one member record rather than three. Retail is small-basket and frequent. Tipping is inconsistent by modality: assisted stretch and massage attract it, a self-serve sauna does not, and a terminal that prompts for a tip on a room booking irritates people.

Hardware matters more than it appears to. Ask whether the terminal is the platform's own and whether it is mandatory, because proprietary hardware is another switching cost stacked on the processing contract and is rarely priced as one. Ask whether it functions offline, since a studio that cannot take a payment during an outage sends paying members home. Ask whether card on file, tap to pay and any kiosk flow write to the same customer record, because the value of an integrated stack evaporates the moment one payment path silently creates a duplicate customer.

All of it resolves at the desk, which is also where the software stack stops. Praxium moves no money and stores no card: it is a directory carrying studio profiles for city and modality searches, plus an optional protocol layer that reads whatever booking system a studio already runs. Nothing on this page is a comparison with it, because a payments contract is not the thing it sells.

The questions that expose a payments contract

Most of what decides the cost of a payments stack is unpublished, so the evaluation is a set of questions rather than a comparison of pages. Put them in writing, send them to every shortlisted vendor at the same time, and keep the replies. Then do the arithmetic once on your own numbers: take last year's card volume and transaction count, apply each vendor's quoted structure including the per-transaction component, add the subscription and any per-location multiplier, and compare totals. That ranking is frequently not the ranking that falls out of comparing tier prices.

• What is the total effective rate on my card mix, including per-transaction, monthly minimum, gateway, PCI, batch and chargeback fees? Ask for a sample statement.
• Is processing mandatory, or may I connect my own merchant account? If mandatory, what happens to the rate at renewal, and is there a rate-change notice period?
• What is the default retry schedule on a failed recurring charge, is account updater included, and where does the failure appear for front desk staff?
• Do you retain a versioned copy of the membership agreement as presented at signup, with a timestamp and device record?
• On exit, do stored card tokens transfer to another processor, and what is the documented process and cost for that transfer?