NEW YORK — A consumer opens one app, sees one balance, and expects one place to call when the money freezes. The stack underneath is rarely that simple. A nonbank platform markets the product. Middleware may route deposits and keep a sub-ledger. A chartered bank holds funds for benefit of customers. Each firm has a different regulator, a different contract, and a different view of whose ticket it is.

Embedded finance sells that split as progress. Distribution improves when checkout, payroll, or a consumer app can offer banking surfaces without building a bank. The same split is the failure mode. When records diverge, support scripts point sideways. The brand on the screen is not the entity that can move the money.

Complaint routing is the real design problem in embedded finance: UX unity without a single accountable owner is a product, not a durable system.

What Synapse made concrete

Synapse Financial Technologies sat in the middle of that stack. According to the Consumer Financial Protection Bureau’s enforcement action page, Synapse provided technology and software that bridged nonbank fintech platforms offering banking services to consumers and traditional partner banks. On 22 April 2024, Synapse filed for chapter 11 bankruptcy protection.

The Bureau alleged that Synapse violated the Consumer Financial Protection Act by failing to maintain adequate records of where consumers’ funds were and by failing to keep those records matched to partner banks’ books. Partner banks determined that the total they held for consumers was less than the total reflected in Synapse’s records — a shortfall the Bureau put between $60 million and $90 million. Consumers had no access to funds for weeks or months while banks reconciled. Many did not receive the full amount of their recorded balance. On 21 August 2025 the Bureau filed a complaint and a proposed stipulated final judgment; the court entered the order on 12 September 2025, including injunctive relief and a $1 civil money penalty that opened a path to civil-penalty-fund redress.

That sequence is not a branding dispute. It is a ledger dispute with human consequences. The customer experience was one app. The legal path ran through a bankrupt middleware firm, several partner banks, bankruptcy trustees, and federal enforcement. Embedded finance had delivered a single interface. It had not delivered a single owner of the complaint.

What supervisors already said about duty

US bank supervisors did not invent third-party risk after Synapse. On 6 June 2023 the OCC, Federal Reserve Board, and FDIC issued interagency guidance on risk management for third-party relationships. Effective contracts, the guidance says, typically specify whether the banking organization or the third party is responsible for responding to customer complaints or inquiries. If the fintech firm handles complaints, contracts should require timely response and enough usable data for the bank to analyze volume, nature, and trends. If the bank handles them, the fintech firm must notify the bank promptly. Either way, both sides are expected to monitor trends and remediation.

On 25 July 2024 the same three agencies issued a joint statement on bank arrangements with third parties to deliver deposit products. The statement does not create new legal duties, the agencies said. It restates a hard line: a bank’s use of third parties to perform activities does not diminish the bank’s responsibility to comply with applicable laws and regulations. The same package included a request for information on bank-fintech arrangements spanning deposits, payments, and lending.

Those documents are the supervisory answer to the product pitch. The screen can be white-labeled. Liability for consumer protection, deposit recordkeeping, and complaint handling still attaches inside the regulated perimeter. Middleware failure does not convert a deposit arrangement into a pure software problem.

Why the brand still loses first

Customers escalate where they live. They open the app. They post the brand name. They file a CFPB complaint naming the company on the card. Banks and program managers may be correct that a contract allocated day-to-day support to another party. Correct allocation does not stop the first public blast radius from hitting the face brand — or stop examiners from treating complaint quality as a signal about the bank’s third-party program.

The operational gaps show up in ordinary tickets long before bankruptcy. A dispute needs transaction data the platform has and the bank does not, or bank core data the platform cannot query in real time. A fraud claim needs authority to freeze a card the issuer controls. A wind-down needs a sequence for moving balances when a partner exits. Those are complaint-path designs. They are also partnership designs. Firms that ship embedded products without a written path for first response, decision rights, regulator contact, and public communications discover the missing RACI only when volume spikes.

Failure mode What the customer sees Who can actually clear it
Ledger mismatch Balance frozen or wrong Bank + middleware reconciling books
Card authorization error Declines at checkout Issuer processor + program manager
Marketing claim vs contract “FDIC insured” confusion Bank disclosures + platform copy owners
Partner exit / insolvency App still branded, money stuck Bank wind-down, bankruptcy court, agencies

Embedded finance will keep growing because distribution is valuable. The durable products treat complaint ownership as launch criteria. The brand on the glass can stay unified. The contracts, data feeds, and escalation trees still have to name who answers, who decides, who talks to supervisors, and who can move the money when the middleware is gone.

Source notes