How to evaluate a challenger bank: nine tests, the evidence for each, and how 11 neobanks compare

How do you evaluate a neobank or challenger bank?

Evaluate a neobank the way a bank supervisor would, then add the questions an investor asks. Nine criteria cover it: (1) the licence and who actually holds the deposits, (2) funding and capital, (3) unit economics and revenue mix, (4) credit exposure and loss rates, (5) product breadth, including the move from debit cards into credit, (6) the technology and AI the company has put into production, (7) the regulatory record, (8) customer outcomes and (9) valuation. Each has a public document that answers it: a register entry, a Pillar 3 disclosure, an annual report, a 10-K or 10-Q, an enforcement notice, a complaints table or a dated funding announcement. A criterion with no public document behind it is scored “not disclosed”, and the absence is information in itself.

The weighting depends on who is asking. A customer needs criteria 1, 7 and 8: is my money insured, is the firm in trouble with its regulator, and are complaints handled. A bank or fintech considering a partnership adds capital (2) and technology (6), because a partner's failure or outage becomes its own problem. An investor reads capital, unit economics, credit and valuation (2, 3, 4, 9). A competitor studies the revenue mix, the product roadmap and the AI stack (3, 5, 6).

Who holds the deposits, and is the money insured?

Start here: it decides what happens to a balance if something fails. A neobank with its own bank charter takes deposits itself and is covered by the national scheme: the FDIC in the US (at least $250,000 per depositor per bank, in the FDIC's words), the FSCS in the UK and a national deposit-guarantee scheme in the EU. A neobank on a partner bank is a programme manager: the customer's money sits in a pooled account at an FDIC-insured bank and the fintech keeps the ledger of who owns what. An e-money institution holds no deposits at all; it safeguards customer funds in segregated accounts.

The Synapse lesson, in the regulators' own words

The FDIC's consumer guidance on banking with third-party apps (June 2024) is explicit: funds sent to a nonbank “are not eligible for FDIC insurance until the company deposits them in an FDIC-insured bank and after other conditions are met”, pass-through cover requires records that identify “who owns the money and the specific amount that each person owns”, and “FDIC deposit insurance does not protect against the insolvency or bankruptcy of a nonbank company.” Synapse Financial Technologies, the middleware firm between fintech apps and their partner banks, showed why. In its proposed recordkeeping rule of 17 September 2024 (FIL-64-2024) the FDIC wrote that Synapse's bankruptcy “has affected the ability of consumers to access funds placed at IDIs for a number of months”, that the partner banks “encountered significant difficulties in obtaining, reviewing, and reconciling Synapse's records”, and that “in many cases, it was advertised that funds were FDIC insured.” The CFPB's complaint against Synapse (filed 21 August 2025, judgment entered 12 September 2025) put the gap between what the banks held and what Synapse's records showed at “between $60 and $90 million”, with consumers locked out “for weeks or months”; The FDIC proposal would require daily reconciliation and direct, continuous access to a third party's records; we found no final rule on the FDIC's site, so treat it as a benchmark for good practice rather than a requirement.

In the UK the Prudential Regulation Authority raised the FSCS limit from £85,000, set in 2017, to £120,000 per person per authorised firm from 1 December 2025 (PRA, 18 November 2025). E-money and payment firms are outside it: the FCA's safeguarding press release of 7 August 2025 says their funds “are not directly protected by the Financial Services Compensation Scheme” and that customers can “lose money or experience delays” if a firm fails; new safeguarding rules took effect on 7 May 2026. In the EU, Directive 2014/49/EU sets the harmonised coverage level at €100,000 per depositor per bank.

The direction of travel in 2026 is towards charters: Chime is buying its partner bank, Nubank and Revolut hold conditional OCC approvals, and Revolut's UK bank exited its restricted mobilisation phase after about twenty months. bunq shows the bar is real: the OCC rejected its application in August 2026. Licence facts in the table come from each linked profile, where the documents are cited.

How do the neobanks compare on capital, unit economics and credit?

Criteria 2 to 5 are read together, because a neobank's revenue mix decides how much capital it needs and how much credit risk it carries. Interchange on a free debit card needs little capital but earns little per customer; interest on central-bank deposits fades as rates fall; lending earns the most and is where losses come from. The table quotes each company's latest disclosed figures; “not disclosed” means the figure is not in the document cited.

Three patterns stand out. First, the profitable European banks are capital-heavy by choice: Monzo's CET1 ratio of 27.28% and Revolut's £4.9bn of pure CET1 are far above minimums, because both still hold most deposits in cash and treasury assets rather than loans. Monzo keeps 60% of deposits at the Bank of England. Second, US partner-bank models live on short-duration credit: Chime's MyPay and Dave's ExtraCash are advances repaid from the next pay cheque, with loss rates of 0.9% (MyPay, quarterly) and a 2.12% 28-day delinquency rate, while Chime still takes 64% of revenue from interchange. Third, loss rates are not comparable across models. Monzo's 9.0% is an annual expected-credit-loss charge on unsecured consumer lending; OakNorth's 0.045% is cumulative realised principal loss on secured lending to mid-sized businesses over ten years; Nubank reports delinquency, not losses. Read each against its own definition.

Product breadth: the move from debit to credit

Every profiled neobank that started with a debit card is now lending. Revolut offers loans, credit cards or buy-now-pay-later in 13 countries and launched mortgage refinancing in Lithuania. Monzo serves 4.1 million borrowers and agreed to buy the mortgage broker Habito. Chime's credit mix reached 27% of purchase volume in Q2 2026 as its Chime Card and Chime Prime tier grew. Dave is testing a Flex credit card underwritten by its CashAI model. Nubank's $39.4bn book is the largest, and its Q2 2026 release says early-delinquency improvement was “partially offset by intentional expansions into higher-risk, higher-return segments.” Our AI credit underwriting page tracks the models behind those decisions.

What does each neobank's regulatory record show?

The regulatory record is the criterion customers and partners can least afford to skip, and the one where the evidence is cleanest: fines, consent orders and supervisory measures are published by the authority that imposed them. European cases centre on financial-crime controls during fast growth; US cases on fees and disclosure for small-dollar advances.

Read the record for direction as well as size: N26's cap was lifted in 2024 and BaFin intervened again eighteen months later. Revolut's UK licence came with restrictions in July 2024 and full launch only in March 2026, and its US approval of September 2026 holds back four products pending separate non-objection (see the Revolut profile). US cases are about conduct, which can still reshape a model: Dave moved all members to a flat 5% ExtraCash fee by February 2025.

Customer outcomes (criterion 8)

In the UK, the FCA's firm-level complaints data is the best public measure. For the half-year 1 April to 30 September 2025 Monzo Bank reported 49,570 complaints opened (49,247 on banking and credit cards), a rate of 2.07 on the FCA's “context” measure, which scales complaints by the number of accounts; Starling reported 17,323, a rate of 3.62. Revolut does not appear in that release of the table. The Financial Ombudsman Service is the second check: The Guardian reported in February 2026 that the ombudsman found Monzo had wrongly rejected 34% of 3,372 fraud and scam complaints referred to it (see the Monzo profile). In the US, the CFPB's consumer complaint database can be searched by company name, including Chime, Dave and MoneyLion; we do not tabulate it here. We do not use app-store ratings: they are not audited, and the only defensible source would be the official stores themselves.

How much AI do neobanks actually run in production?

Criterion 6 asks what is live, not what is announced. The table below is computed from this site's eleven neobank profiles, each use case sourced to a company document, filing or trade report and tagged with a status. Across the eleven, 124 of 157 tracked use cases are in production. Two caveats apply. Counts reward disclosure: bunq and OakNorth publish fewer, broader items, all of them live, while MoneyLion's detail has thinned since it became a Gen Digital subsidiary. And a few “in production” entries are governance rather than customer-facing AI, such as N26's supervisory remediation programme.

The metrics behind the counts are more useful than the counts. In customer support, Nubank says AI agents handle more than 60% of Brazilian support conversations, Revolut's Assistant resolves over 75% of queries, Chime says AI powers roughly 70% of member-support interactions, Monzo's self-service handles 62% of inbound chats and bunq says Finn fully resolves 84% of queries. In credit, Nubank's in-house nuFormer model underwrites cards in Brazil and Mexico, Dave's CashAI v6 had reached about a third of users by August 2026, and Chime replaced hand-built features with transformer sequence models in four production risk models. The build-or-buy split is visible too: Nubank and Revolut train their own foundation models, while N26 (Claude via AWS Bedrock), bunq (Claude via Bedrock), Starling (Google Gemini) and OakNorth (OpenAI) build on vendor models. Ask for the metric, the date it was measured and whether a human still reviews the output, and look for the effect in cost to serve, as in Chime's stated 60% drop. The AI fraud detection and AI in retail banking pages compare the same systems with incumbent banks.

How are neobanks valued, and what are the latest valuations?

Listed neobanks are valued by the market every day; private ones only when shares change hands, and the basis matters. A primary round prices new shares the company issues; a secondary sale prices existing shares changing hands, often employees' or early investors', and can be set by a narrow group of buyers; an acquisition price includes a control premium. Growth neobanks are compared on revenue multiples, profitable ones increasingly on earnings and book value. Market capitalisations below are dated LSEG snapshots on FT Markets and move daily.

The spread is wide. On the reported range, Monzo's 2024 secondary price was about 2.6 times its FY2026 revenue of £1.7bn, and the £8bn–£10bn in the September 2026 talks would have been 4.7 to 5.9 times; Starling at £4bn would be about 4.5 times its FY26 revenue. Revolut's $75bn is a large premium to both, paid for growth (revenue +46% in 2025) and margin (£1.7bn pre-tax on £4.5bn). For the four companies valued on 2019 to 2023 rounds, N26, bunq, OakNorth and Varo, the figure is a historical reference, not a current price; Current's 2026 down round shows how far a 2021 mark can fall. Ratios here are our arithmetic on the cited figures.

How should customers, partners, investors and competitors weight the criteria?

• Customer. Find the bank that holds your money in the account terms. If the neobank is the bank (Varo, Monzo, Starling, N26, bunq, OakNorth, Revolut's UK and EU banks), deposit insurance covers its failure up to $250,000, £120,000 or €100,000. If it is a partner-bank app (Chime, Dave, MoneyLion), FDIC pass-through cover protects you against the partner bank failing, not the app; the FDIC tells consumers to be “particularly careful” with money they rely on for day-to-day expenses. Then check the regulator's enforcement page and the complaints data.
• Partner (a bank sponsoring a fintech, a vendor, a card programme). Add capital and liquidity: Chime's $1.06bn of cash and securities and Dave's $254.4m set how long either can absorb a shock. Ask for the reconciliation process the FDIC's 2024 proposal describes, daily and with direct access to the ledger, whether or not the rule is final.
• Investor. Weight unit economics, credit and valuation basis. A neobank whose revenue is mostly interchange (Chime, 64% in Q2 2026) is exposed to spending and card-network economics; one built on interest income is exposed to rate cuts, as Monzo's net interest margin fall from 4.1% to 3.7% shows. Discount a 2021 primary price, and treat a secondary price as a ceiling set by a few buyers.
• Competitor. Track the move into credit and the AI stack. The profitable neobanks are building lending books and subscriptions on top of a primary account, and the leaders run their own models in underwriting and support; the profiles above list what is live.

Applied to the eleven: on safety, the chartered European banks disclose most, but N26 carries an open BaFin order and Monzo and Starling paid large financial-crime fines in 2024–2025. On economics, Revolut, Nubank, Starling, OakNorth and Monzo are profitable on their own reporting, Chime and Dave turned GAAP-profitable in 2026, and Varo still reports losses. On franchise, Nubank and Revolut have the largest credit and AI build-outs. Our neobank hub lists the profiles and the other explainers in this series.

1. Banking With Third-Party Apps — Federal Deposit Insurance Corporation, Jun 1, 2024 (fdic.gov)
2. Requirements for Custodial Deposit Accounts with Transactional Features and Prompt Payment of Deposit Insurance to Depositors (FIL-64-2024) — Federal Deposit Insurance Corporation, Sep 17, 2024 (fdic.gov)
3. Enforcement action: Synapse Financial Technologies, Inc. — Consumer Financial Protection Bureau, Sep 12, 2025 (consumerfinance.gov)
4. PRA confirms FSCS deposit limit to be increased to £120,000 from 1 December — Bank of England (Prudential Regulation Authority), Nov 18, 2025 (bankofengland.co.uk)
5. FCA sets out changes to payment safeguarding rules — Financial Conduct Authority, Aug 7, 2025 (fca.org.uk)
6. Directive 2014/49/EU on deposit guarantee schemes — European Parliament and Council (EUR-Lex), Jun 12, 2014 (eur-lex.europa.eu)
7. FCA fines Starling Bank £29m for failings in financial crime systems and controls — Financial Conduct Authority, Oct 2, 2024 (fca.org.uk)
8. FCA fines Monzo £21m for failings in financial crime controls — Financial Conduct Authority, Jul 8, 2025 (fca.org.uk)
9. Firm-level complaints data sortable table: 2025 H2 — Financial Conduct Authority, May 15, 2026 (fca.org.uk)
10. Revolut Group Holdings Ltd Annual Report 2025 — Revolut, Mar 24, 2026 (assets.revolut.com)
11. Revolut completes fundraising process establishing $75 billion valuation — Revolut, Nov 24, 2025 (revolut.com)
12. Monzo Annual Report and Accounts 2026 — Monzo, May 19, 2026 (monzo.com)
13. Monzo Bank Limited Pillar 3 disclosures 2026 — Monzo, May 19, 2026 (monzo.com)
14. Starling delivers fifth year of profitability and accelerates global growth strategy — Starling Bank, May 21, 2026 (starlingbank.com)
15. Nu Holdings Ltd. Reports Second Quarter 2026 Financial Results — Nu Holdings, Aug 13, 2026 (nu.com)
16. Nu Holdings Ltd. Form 6-K: Statement Regarding Media Reports — Nu Holdings via SEC EDGAR, Sep 30, 2026 (sec.gov)
17. Chime Reports Second Quarter 2026 Financial Results — Chime, Aug 5, 2026 (investors.chime.com)
18. Dave Reports Second Quarter 2026 Financial Results — Dave Inc. via PR Newswire, Aug 5, 2026 (prnewswire.com)
19. OakNorth 2025 annual results — OakNorth, Mar 17, 2026 (oaknorth.co.uk)
20. Revolut fined €3.5m by Lithuania's central bank for "violations of money laundering prevention requirements" — FinTech Futures, Apr 9, 2025 (fintechfutures.com)
21. Nu Holdings Ltd. (NU:NYQ) summary: market cap — FT Markets (LSEG data), Sep 16, 2026 (markets.ft.com)
22. SoFi Technologies, Inc. (SOFI:NSQ) summary: market cap — FT Markets (LSEG data), Sep 30, 2026 (markets.ft.com)
23. Chime Financial, Inc. (CHYM:NSQ) summary: market cap — FT Markets (LSEG data), Aug 28, 2026 (markets.ft.com)
24. Dave Inc. (DAVE:NMQ) summary: market cap — FT Markets (LSEG data), Oct 2, 2026 (markets.ft.com)
25. Monzo in early talks over possible sale to Brazil's Nubank - Sky News (Reuters) — Investing.com (Reuters), Sep 26, 2026 (investing.com)
26. Starling Bank reportedly plans secondary share sale targeting £4bn valuation — Retail Banker International via Yahoo Finance, Sep 11, 2025 (finance.yahoo.com)
27. Challenger bank N26 raises $900 million at $9 billion valuation — TechCrunch, Oct 18, 2021 (techcrunch.com)
28. About bunq — bunq Newsroom, Sep 22, 2026 (press.bunq.com)
29. SoftBank leads $440 million investment in UK fintech OakNorth, valuing it at $2.8 billion — CNBC, Feb 8, 2019 (cnbc.com)
30. Neobank Current raises $80 million, reports growth — American Banker, Jun 15, 2026 (americanbanker.com)