What Happens to Your Money When a Neobank Fails
Written with AI assistance and reviewed by the NorwegianSpark SA editorial team.
Most people choose a digital banking app on fees, the interface, and how fast the account opens. Almost nobody checks the one thing that decides what happens to their balance if the provider goes under — and that thing is not the brand, the country, or the size of the company. It is the licence.
This guide uses two real cases we have written up separately, because they failed in different ways and produced different outcomes for customers.
Three structures that all look like a bank in an app
A licensed bank (credit institution). Takes deposits on its own balance sheet, lends them out, and is covered by a statutory deposit guarantee scheme. In the EU that is EUR 100,000 per depositor per institution under Directive 2014/49/EU; Switzerland privileges CHF 100,000.
An e-money or payment institution. Issues e-money and moves payments, but does not take deposits. Article 1(2) of that directive applies it to credit institutions, and e-money is expressly not treated as a deposit — so there is no deposit guarantee scheme. Instead, customer funds must be safeguarded: held separately from the firm's own money, so they are insulated from the firm's creditors in an insolvency.
A brand sitting on top of somebody else's licence. The app you use has no licence at all; a partner bank or EMI holds the money. This is extremely common and entirely legitimate, but it means the protection follows the partner, not the brand.
The third case is the one people get wrong most often, because there is nothing on the app to tell you.
Case one: a licensed bank fails
FlowBank was a Swiss bank with a genuine banking licence. FINMA withdrew that licence in March 2024 and opened bankruptcy proceedings in June, citing capital breaches, deficient organisation and anti-money-laundering failures.
What happened to depositors: privileged deposits up to CHF 100,000 per client were protected, and FINMA's own assessment was that they "can be repaid in full out of the bank's available funds", without the industry deposit insurance scheme needing to be involved at all.
The important detail is that the regulator intervened while there was still money. That is the practical value of a banking licence: continuous supervision, and an authority with the power to close the institution before the hole gets bigger. The full sequence is in our FlowBank write-up.
Case two: a fintech brand fails
Nuri, formerly Bitwala, was not a bank. Its euro accounts were provided by a licensed German partner bank; Nuri was the brand and the app. It filed for insolvency in August 2022 and wound down by December.
What happened to customers depended entirely on which product they held. Euro balances sat at the licensed partner bank, outside Nuri's insolvency and inside the deposit guarantee scheme. But funds committed to Nuri's interest-bearing crypto product depended on a third-party lending platform that had itself gone bankrupt — outside any guarantee scheme, because a deposit guarantee scheme protects deposits, not assets lent out for yield.
One app, two products, two completely different legal positions, presented on the same screen. The full Nuri story is worth reading for that reason alone.
What the two cases teach, together
Protection attaches to the entity and the product, never to the app. The question is never "is this company trustworthy". It is "which entity holds this specific balance, under which licence".
Safeguarding is not a guarantee scheme, and both are real. Safeguarded funds are segregated and should be returned in a wind-down — but the process is an insolvency process, it takes time, and there is no scheme paying out a fixed amount on a fixed timetable. A deposit guarantee scheme is a promise of a specific sum. If you need certainty and speed, that difference is the whole answer.
Yield is the tell. A return meaningfully above what banks pay is not a better deal on the same product; it is a different product with credit risk attached. That was precisely what separated the protected and unprotected halves of a Nuri customer's balance.
The limit is per institution, so concentration is the risk you control. Holding more than the guaranteed amount at one institution leaves the excess unprotected no matter how safe the institution looks. Splitting balances is the entire remedy and it costs nothing.
What a wind-down actually looks like from the customer side
The abstract version — "funds are safeguarded and returned" — skips the part people care about, which is what the weeks after an announcement feel like.
The card stops working, usually first and usually without warning. Card processing depends on the provider's own relationships, and those end quickly. Direct debits and standing orders fail around the same time, which is how most people find out something is wrong.
Incoming payments start bouncing. Salary, invoices and refunds sent to an account that is winding down may be returned to sender, which sounds tidy and is not: it means chasing every payer to re-issue with new details, at exactly the moment you have no working account.
There is usually a withdrawal window, and it usually holds. In an orderly wind-down customers are told to move their money by a date, as Nuri's were. Meeting that date is the single most important thing you can do, and the reason to act on the first announcement rather than the last reminder.
If it is not orderly, the timetable belongs to a liquidator. Once an insolvency practitioner is appointed, the process runs on legal deadlines rather than customer convenience, and correspondence replaces the app.
The operational lesson is separate from the protection lesson and just as valuable: do not run your only account anywhere. A second account at a different institution, with the same salary and direct-debit details on file, converts a crisis into an afternoon of admin. It costs nothing to open one before you need it.
What to actually do, in order
- Find out which legal entity holds your money and what licence it has. The provider's terms name it; the regulator's register confirms it. How to check whether a bank is actually licensed is the procedure.
- Classify each balance as operational or stored. Operational money can sit with an e-money institution; money you will not touch for a year belongs somewhere a guarantee scheme covers.
- Check you are under the limit at each institution, and split if you are not.
- Treat any interest-bearing feature inside a banking app as an investment product and read it on those terms.
- Re-check annually. A licence is a live status, as FlowBank demonstrated.
The setup that makes all of this a non-event
You can reduce this entire subject to an afternoon of admin, once, and then stop thinking about it.
One licensed bank for stored money. Savings and anything you will not touch for a year sit at a credit institution, inside the guarantee scheme, under the limit. If you exceed the limit, use a second institution rather than a second account at the same one — the limit is per institution and a second account does not double it.
One provider for operations. Day-to-day spending, multi-currency handling, business invoicing. An EMI is entirely reasonable here, because the balance is money passing through rather than money stored.
A dormant backup at a third institution. Opened, verified, with a card, and left empty. This is the piece almost nobody has and the one that turns a provider failure from a crisis into an inconvenience: salary redirects the same day, direct debits move within the week.
An annual review. Fifteen minutes to re-check each provider's register entry and confirm the balances are still on the right side of the limits.
That is the whole system. It does not require predicting which provider will fail, which is fortunate, because nobody can.
The part that is genuinely reassuring
Nothing above is an argument against using fintech. Both cases ended with the protected layer intact, and in the Swiss case the regulator expected depositors to be repaid in full from the bank's own funds. The system did what it is built to do.
The failures were failures of understanding, not of protection: people did not know which of the three structures they were dealing with, or that one app can contain two of them. That is a five-minute problem to fix, and fixing it costs nothing.
For live providers, our neobank ranking and business account comparison cover the current field, and the neobank safety guide is the shorter companion to this page.
Banks mentioned in this article
Affiliate disclosure: The links above are affiliate links. We may earn a commission at no extra cost to you.