CHI Company Investigation · Finance
RevolutCurrent accounts · Cards · Foreign exchange · Transfers · Savings · Stocks · Crypto · London, Vilnius
Ten research streams · Verdict: weakly supports · No numerical score assigned
Revolut is one of the cheapest, fastest current accounts a normal customer can hold.The finding begins the moment you stop being a normal customer.
This investigation was commissioned to test whether Revolut's automation — the thing that makes it cheap, fast and global — transfers an unreasonable share of the burden onto the customer when something goes wrong. Across fraud reimbursement, pricing, payments, investments and product design, the answer was no, and the evidence that says so is independent. The hypothesis survived in exactly one place: the interval between an automated restriction and a human who can lift it.
Normal The customer who never trips a rule
Cheap, fast, and better than the bank it replaced.
No foreign-exchange fee on the free tier up to £1,000 a month and 1% above it — plus a weekend markup — against a high-street benchmark that MoneySavingExpert puts at “3%-ish”. Cash withdrawals free to five a month or £200, then 2% with a £1 minimum, against roughly 2.5% and a £3 minimum elsewhere. International transfers at 0.15% from April 2026, capped at £250 — per-currency minimums apply, and £3–£5 flat for a non-local currency — where a bank wire is typically £15–30. Instant Revolut-to-Revolut payments; SEPA Instant in ten seconds.
Which? scored Revolut 83% in September 2025, sixth of twenty-two providers and above every large high-street bank in its set — on 121 Revolut respondents within a 6,665-person survey. The Advertising Standards Authority has never upheld a complaint against it in ten years.
Exception The customer an automated scenario has flagged
A different company answers the door.
An automated transaction-monitoring rule — one of the [101–150] transaction-monitoring scenarios recorded in the Italian decision — restricts the account. Revolut's own guidance states that customer support cannot lift restrictions and that it “can't always explain” why one was applied. Its only public speed commitment starts after its review is finished, not when the review starts. It publishes no retail review-duration commitment outside Italy and no restriction statistics anywhere.
In Italy, on Revolut's own submissions to the regulator, blocks in 2023–July 2025 averaged [51–100] days. Revolut disputes the decision and has said it will appeal.
Both columns are true at the same time, and neither is a phase the other replaced. This is not a company that used to be bad and got better, or the reverse. It is one system operating in two states, and CHI's finding is about the second one. Everything below is dated, jurisdiction-stamped and labelled by evidence type for that reason.
The test
The question was not whether Revolut is good. It was where the burden lands when the machine says no.
Automation is the whole proposition. It is why a Revolut account costs nothing to open, settles in seconds and prices currency at a fraction of a bank's rate. The investigation asked whether the same automation, in the cases it gets wrong or cannot resolve quickly, shifts the work of proving the customer is legitimate onto the customer — and whether the route back to a person who can act is real.
Hypothesis under test: Revolut's automated infrastructure externalises a disproportionate share of operational, fraud-prevention, compliance and support burden onto customers, particularly when accounts are restricted, funds become inaccessible, transactions fail, disputes occur or a human is needed.
Ten research streams were run concurrently by separate researchers from a common brief, without sight of each other's work, so that no early thesis could contaminate a later stream. The draft was then attacked twice — once by a General Counsel defence review that tested fifty-six proposed findings and defeated thirty-four of them in the form proposed, and once by a combined fact audit and verification pass. Every correction either review required has been applied. The strongest counter-case was built deliberately, not as a courtesy.
That method matters here because the hypothesis mostly failed. Five of the ten streams returned findings that point away from hostility, and two more returned mixed results in which the larger body of evidence does — several of them resting on independent data Revolut does not control. What follows reports the failures as prominently as the survivor.
Fees and marketing are marked mixed because each contains one narrow supported element — disclosed exit friction and a documented historical tightening of the free tier in the first; a headline-for-paid-tier pattern in the second — alongside a much larger body of evidence that does not support deception. Both are set out in full further down.
The distinction this page turns on
Read this before the evidence, because almost every misreading of Revolut starts here.
Banks and electronic-money firms are legally required to monitor transactions, verify identity, screen against sanctions lists, detect suspicious activity and act on what they find. Acting sometimes means restricting an account, and it sometimes means doing so without warning. CHI does not treat a lawful restriction as customer hostility, and this investigation does not argue that Revolut's restrictions are improper. Italy's competition authority did not find them unlawful either, and the Bank of Italy's opinion in that case is reported to have accepted the legality of the underlying controls and the constraints that anti-tipping-off rules place on what a firm may say.
The CHI question begins after the flag: what is the customer told, how quickly can they respond, who is able to intervene, how long can access to their own money remain restricted, is there a review clock, and is the human recourse real or nominal. Those are process questions, and process is what a regulator examined.
Hero evidence · Italy
For one branch, in one conduct period, a regulator opened the machine and wrote down what it found.
On 23 March 2026 the Italian Competition and Consumer Authority (AGCM) adopted decision n. 31910, announced on 2 April. It fined Revolut entities €11.5 million in total — the authority's own headline is “over €11 million” — across three separate practices. One of the three is the load-bearing evidence on this page, and it is the only regulatory decision anywhere that examines precisely the conduct this investigation set out to test.
Total across three practices, Italian branch, conduct period 2023 – July 2025. Never €5m: that is one component of three.
Account suspension, limitation and blocking. Found to be an aggressive and misleading practice: blocks applied without adequate prior notice, without an effective opportunity to reply and without adequate assistance, leaving customers unable to access funds “even for long periods” and unable to meet urgent financial needs.
The CHI evidenceInvestment marketing. “0% commission” investing found misleading for omitting plan caps on free trades, the foreign-exchange fee on dollar purchases above a monthly threshold, regulatory and custody costs, spreads, and the material differences of fractional shares. Revolut rewrote the Italian marketing in December 2025.
Part of the action, not the thesisItalian IBAN representations. A claim that all new clients would automatically receive an Italian IBAN, when nine cumulative conditions applied; roughly [700,000–800,000] of about three million Italian customers were ineligible and [50,000–100,000] were never invited because of a technical error.
ContextWhat the decision found, and what it did not
Found
- The process around blocks breached articles of the Italian Consumer Code governing misleading omissions and aggressive practices.
- Pre-contractual information on the conditions and manner of suspension was omissive and ambiguous.
- Customers discovered restrictions “fortuitously”, when a payment failed rather than by notice.
- Chat responses in one recorded case were “generic and repetitive” over roughly a month and a half.
- Revolut's own complaints manual anticipated recurring restriction complaints about chat delay and repetitive messaging caused by a lack of updates.
- Blocks were triggered automatically by [101–150] transaction-monitoring scenarios.
Not found
- That the blocks themselves were unlawful. AGCM made no such finding.
- That Revolut's anti-money-laundering controls were deficient. The Bank of Italy's opinion is reported to have accepted their legality and the tipping-off constraint.
- That Revolut intended to shift burden onto customers. Intent is not in this record, and the investigation does not assert it.
- That the same durations, volumes or communication failures occur outside Italy. The operating findings are Italian-branch, 2023–July 2025.
- Anything final. This is a first-instance administrative decision, not a judicial adjudication.
The operating numbers, exactly as the regulator redacted them
Every figure below comes from Revolut's own submissions to AGCM and is published by AGCM as a range, not a value. CHI publishes the ranges. No exact number is inferred, no total is derived, and no figure is applied outside the Italian branch or outside the 2023–July 2025 conduct period.
Why the brackets stay. AGCM redacted these figures at Revolut's request because they are commercially sensitive. A range is not a coy version of a number — it is the evidence. Converting “[51–100] days” into a single average, or multiplying [10,001–50,000] by the inverse of [1–5]% to produce a case total, would manufacture precision the record does not contain. CHI does neither, and the [1–5]% figure never appears on this page without the [95–99]% that belongs beside it.
Status of the decision. This is a contested first-instance administrative decision. Revolut's public position is that it “strongly disagrees with the AGCM's findings and will appeal the decision in the Italian courts”. No appeal filing was located during the investigation, and no court has ruled. Nothing on this page should be read as a judicial determination, and the word “adjudicated” is not used.
Three different timing figures appear in the decision — a [1–5] day grievance-handling target, a 15 or 35 business-day contractual complaint window, and an 88-day internal service-level agreement for total resolution of an anti-money-laundering alert. They measure three different things. CHI does not present them as an inconsistency or as a misstatement to the regulator.
Internal messages quoted in the decision, including a manager's acknowledgement about agent communication, were extracted by tooling during research and are not reproduced here: the package requires them to be re-read verbatim in the decision PDF, with the tipping-off context, before publication.
The missing clock
Revolut publishes a three-hour promise. The three hours begin after the part nobody has measured.
In a blog post of 28 February 2024 Revolut told customers that account restrictions are generally lifted “within 3 hours of confirming all the information we need”. The sentence is precise and, so far as the record shows, accurate. It is also a commitment about the end of the process. The clock that matters to a customer whose salary is sitting behind a block is the one that runs before it: the time taken to investigate, to decide what information is required, to request it, to review what comes back, and to conclude that everything needed has arrived.
Revolut's own commitment, stated in a company blog post: restrictions generally lifted within three hours of confirming all the information we need. This is the only public speed commitment on retail restrictions that the investigation located.
Everything that has to happen before the three hours can start. No retail review-duration commitment was found on any Revolut help page, in the UK or Irish terms, in the Annual Report 2025 or in the FY25 financial-crime report. The business help page gives fifteen business days, and adds that Revolut “might not be able to provide expected response times”. Revolut also publishes no statistics at all on how many accounts are restricted or how long reviews take.
Italian branch only, conduct period 2023 – July 2025, on Revolut's own submission to AGCM, in a decision Revolut disputes and will appeal. Average duration of a block. [35–50]% ran beyond a month, so [50–65]% were resolved inside one. On the same submission, [95–99]% of cases were completed within Revolut's own internal service-level agreement and [1–5]% were not. These figures describe Italy. They are not evidence of durations in the United Kingdom, Ireland or anywhere else, and this page does not use them that way.
What CHI is not saying. There is no evidence that Revolut conceals this interval, and none that it designed the gap deliberately; both claims were tested and neither is supported. The finding is narrower and factual: no equivalent public retail review-duration commitment was located outside the Italian remediation of June 2026, and a speed promise measured from the end of a process the customer cannot see is not a speed promise the customer can rely on. Revolut's Italian terms, revised on 9 June 2026, now promise a reason where the law allows and an estimated time to resolve — which is precisely the missing clock, supplied in one country.
Human recourse
The question is not whether you can reach support. It is whether the support you reach can fix the thing that is wrong.
Revolut's support is not a wall. Every customer, on every plan, has in-app chat around the clock in more than a hundred languages, with automatic escalation from the chatbot to an agent. Revolut states that a generative-AI first line resolves 75% of interactions and that it employs more than two thousand agents. Nothing in the record shows a customer denied a human, and this page does not claim that Revolut has no human support, that customers are trapped with a bot, or that standard customers cannot reach a person. Each of those was tested and none is supported.
What the evidence does support is a distinction between three different things that are usually collapsed into one word.
First question
Can the customer reach support?
Yes, and on every tier. In-app chat is available 24/7 in 100+ languages with bot-to-agent escalation. Formal complaint procedures in the UK, Lithuania and Ireland were checked against the applicable rules and found compliant: acknowledgement within three business days in the UK, then fifteen or thirty-five business days for payment complaints and eight weeks otherwise, with the Financial Ombudsman available after that.
Answer: yes
Second question
Can the customer reach a person by voice?
Chat-first, with a voice channel whose boundary is unresolved. Customer-initiated in-app calling is listed on Revolut's pricing materials as a Premium, Metal and Ultra benefit, in English only; a separate Revolut help article appears to imply broader access. The two documents conflict and CHI does not resolve the conflict against Revolut. The only public phone number in the UK and EEA is an automated card-blocking line that states it “can't connect you to a human agent”. US customers are given a number; whether it reaches a person was not tested.
Answer: unresolved
Third question
Can that person lift the restriction?
Revolut's own answer is no. Its guidance states that customer support cannot lift restrictions, and that it “can't always explain” the reason for one — a constraint that tracks genuine anti-tipping-off law and applies to every bank. A customer who is locked out of the app entirely is directed to email, for which no response time is published. So the channel that is universally available is the channel that is expressly unable to resolve the one problem this investigation is about.
Answer: not by that route
Why this is only weakly supports. The architecture above is documented in Revolut's own materials and is not in dispute. What is missing is outcome evidence. Revolut publishes no time-to-human metric and no wait-time distribution; its resolution and satisfaction figures are its own and unaudited, and CHI does not use them as evidence for Revolut either. The only independent datum on support quality is Which?'s three-star sub-score for “contacting customer services” — sitting beside an overall score of 83%, and drawn from 121 Revolut customers within a 6,665-person survey. One regulator has made a finding on Revolut's complaint handling: the Bank of Lithuania warned Revolut Payments UAB in November 2022 that not all complaints had been handled in accordance with law, and required in-app chat complaints to be registered and answered with external verification. That entity has since been merged, the remediation was mandated and externally checked, no later finding was located, and it is now nearly four years old. It is history, not a current finding.
The company's strongest ground
Revolut restricts accounts because the law requires it to, and its regulator fined it for not doing enough of it.
Any honest account of this finding has to hold two things at once. A bank that blocks an account is often doing exactly what it is supposed to do, and a bank that blocks too few is in trouble of a different kind. Revolut is under supervisory pressure to monitor more aggressively, not less, and that pressure landed during the same period the Italian conduct was examined.
The obligations are real and non-discretionary
Legal dutyTransaction monitoring, customer due diligence, sanctions screening and suspicious-activity reporting are legal requirements, not product choices. So is tipping-off: where a suspicious-activity report has been made, a firm may be prohibited from telling the customer why their account is restricted. Revolut's inability to always explain is in part a legal constraint shared by every regulated bank, and the UK regulator's own expectation is that reasons be given “where lawful” — the same qualified formula Revolut's terms use.
Its home regulator fined it for detecting too little
Supervisory pressure toward more blockingOn 8 April 2025 the Bank of Lithuania fined Revolut Bank UAB €3.5 million after a scheduled inspection found deficiencies in the monitoring of business relationships and operations, such that the bank “did not always properly identify suspicious monetary operations”. No money laundering was found. Revolut acknowledged the violations, remediated and settled. Reuters reported it as the largest fine the Bank of Lithuania has imposed.
That fine is not evidence of customer hostility and this page does not use it as such. It is evidence of the direction the regulatory ratchet turns. Nothing in the record links it causally to the Italian over-blocking, and neither regulator draws that link.
The sector data does not single Revolut out
Industry baselineThe Financial Conduct Authority's account-closure review covered 34 firms and more than 90% of the UK market. It found that financial crime and dormancy dominate closures sector-wide, with personal-account suspension rates in a central range of 0.1% to 2.3%. Revolut's own rates are not public, and no UK, Irish, Australian or US enforcement action specific to Revolut's freezes or its restriction communication was located.
And the Italian decision agrees on this point
Conceded by the regulatorAGCM did not find Revolut's blocks unlawful. The Bank of Italy's opinion in the proceeding is reported to have accepted the legality of the relevant anti-money-laundering controls and the constraints created by anti-tipping-off rules. The Italian finding is therefore narrower than it first appears, and precisely because it is narrow it is difficult to answer: it is not about whether Revolut may restrict an account. It is about what happened next.
The proportion
On Revolut's own numbers, the overwhelming majority of cases worked. That is not a defence CHI is going to hide.
The most misused figure in this decision is [10,001–50,000] cases outside Revolut's internal service-level agreement. Read alone, it sounds like a system in collapse. Read as Revolut submitted it, it is [1–5]% of cases — which means [95–99]% of cases were completed inside the same standard, on the same submission, over the same thirty months.
Two cautions before this bar is used for anything. First, “within SLA” is not the same as “handled well”, and the Italian finding is not a finding that Revolut missed its own deadlines — it is a finding about notice, opportunity to reply, assistance and the length of time money was inaccessible. A case can sit inside an internal target and still involve a customer who was never told why. Second, the complement runs the other way too: this is Revolut's own unaudited submission, redacted to a range, for one branch.
Why a small tail still matters when the product is access to money
CHI's position is not that a [1–5]% exception rate proves a broken system. In most consumer categories a [95–99]% success rate would be unremarkable and in several it would be excellent. The argument is about consequence, not frequency.
A delivery that fails is re-sent. A subscription billed in error is refunded. A current account that is restricted is different in kind: for the period it lasts, the customer cannot pay rent, cannot meet a direct debit, cannot buy food with that money, and in the ordinary case cannot be told why. Revolut's Italian submission records an average blocked balance of roughly £[1,001–5,000] and an aggregate of roughly £[51–100] million — sums that are unremarkable in aggregate and decisive in an individual month. Severity does not scale down with frequency when the affected product is someone's money.
That is the whole of the argument, and it is deliberately modest. It does not require Revolut to be worse than its peers, and this investigation could not establish that it is. It requires only that the tail be visible, and the record shows that outside Italy it is not: no published review clock, no published restriction statistics, and the one contractual promise of a resolution estimate confined, on the evidence available, to a single country.
Falsification
We tried to prove the rest of it, and could not. Here is what failed.
An investigation that only reports what it found is not an investigation. The hypothesis predicted burden transfer across the whole product surface, and it was tested across the whole product surface. In five of the domains where the prediction was strongest, the evidence ran the other way — often on independent data that Revolut does not produce, control or publish.
Fraud reimbursement
Does not supportThe starting fact is real and unflattering: by absolute count, Revolut has generated more fraud-related complaints at the UK Financial Ombudsman than any other firm in every measured period since 2023. That is a genuine signal about the size of the fraud-dispute population Revolut's product produces.
It is not a signal about refusal. When those complaints reach an independent adjudicator, they are upheld at rates that sit among the peer group, not outside it. Revolut is inside the UK mandatory reimbursement scheme on standard terms, with the £100 excess waived for vulnerable customers. On the Payment Systems Regulator's data, the rate at which scam proceeds were received into Revolut accounts fell 35% between 2022 and 2023.
The investigation does not support a thesis that Revolut systematically refuses legitimate fraud reimbursement, and this page does not advance one.
Pricing and fees
Does not support deceptionRevolut's pricing is aggressive against incumbents and the free tier is genuinely useful. Conditional fees generally appear in published material. Several fees have fallen: crypto commission from 2.5% in 2021 to 1.99% and then 1.49% in June 2023; the UK stock custody fee removed in February 2024; the weekend currency markup removed for Premium, Metal and Ultra around April 2025 and halved for Plus.
The honest counterweights are secondary and they are dated, not quiet. Between 2017 and 2021 the free monthly currency allowance on the Standard tier fell from £5,000 to £1,000, and the fee above it rose from 0.5% to 1% between 2021 and June 2023. Plus, Premium and Metal prices rose in November 2023 — announced by press release, with a lock-in option. Paid plans are twelve-month plans: a monthly-billed downgrade between day 15 and month 10 carries a two-month break fee (one month for Ultra, to day 180), and annual plans carry a fourteen-day cooling-off refund but nothing after that. Price changes normally carry two months' notice. From 30 October 2024 a £2 monthly administration fee applies to residual balances two months after closure; the scale on which it is applied is unknown.
One regulator has acted on Revolut's fee communication. In a commitment decision announced in January 2025, Poland's UOKiK closed — without a fine and without an infringement finding — a case about 2020–2023 fee changes notified by email link rather than on a durable medium; Revolut refunded overpaid fees, gave free upgrades and paid two months of cashback. That is a Polish finding about a medium of notice. It is not a finding about fee levels and it does not travel.
None of this is a dark pattern, none of it is concealment, and it is not the story of this page.
Investments and crypto
Does not supportThis is the domain where the hypothesis failed most completely.
- No enforcement finding on investment or crypto conduct from the FCA, the Bank of Lithuania, CySEC, ASIC, the SEC or FINRA. The Italian disclosure matter, which is contested, is the only regulatory finding of any kind.
- No gamification finding. The FCA's 2024 study of trading-app design named no firm.
- Negligible complaint volume. Of 5,520 new Revolut cases the Ombudsman recorded for July 2025 – March 2026, 17 were investment complaints.
- Stock transfers out exist, at USD 35 per position. Fractional holdings cannot be transferred — which is industry-standard and true of every major competitor, not a Revolut peculiarity.
- Crypto withdrawal access expanded over time, from none at all before May 2021 to broadly available on eligible plans and locations today.
- No payment for order flow is received in the UK or EEA. In the United States, Revolut Securities Inc. discloses that it receives a portion of the order-flow payment earned by its clearing partner.
- Product withdrawals — US crypto in 2023, UK business crypto purchases in 2024, a stablecoin delisting in the EEA in 2026 — were regulatory and orderly, with notice.
- MiCA authorisation was obtained through CySEC in October 2025.
The Italian “0% commission” finding belongs on this page because it is part of AGCM's action and carries a €5m penalty. It is not the thesis, and inflating it into one would misrepresent the record.
Product design and payments
Does not supportNeither AGCM nor UOKiK used the language of dark patterns, and no consumer organisation's dark-pattern research names Revolut. What the record contains is a disclosed friction asymmetry — one tap to upgrade, a fee-bearing downgrade for ten of twelve months — stated on the plan page and present in EU terms since at least October 2021. That is a real asymmetry and it is recorded above under fees. It is not concealment.
On payments, the timing commitments are tight and specified: Revolut-to-Revolut instant, SEPA Instant within ten seconds, a published per-transaction release date on merchant holds, and published dispute timelines including the observation that roughly 85% of disputes resolve within five weeks. Revolut also discloses, unusually, that low-value chargebacks are adjudicated by language models — a disclosure the investigation did not find peers making, though peer practice was not surveyed, and one treated here as neutral because there is no evidence of outcomes either way.
The serious outages in the record are historical — a third-party card processor in 2017 and 2018, an app-wide failure in November 2021 — and were addressed by bringing processing in-house. No regulator publishes a per-firm outage series covering Revolut, and the aggregator data is contradictory. CHI therefore makes no claim about Revolut's current reliability in either direction.
The fraud data, since it is the number most often quoted at Revolut
Remediation
Revolut changed things. The changes are real, recent, mostly reactive, and in the clearest cases confined to one country.
CHI credits corrections in full when assessing the customer's present-day experience, regardless of what prompted them. External compulsion does not retroactively erase conduct, but it also does not reduce the value of a fix. What follows is every dated change the record supports, classified by what kind of evidence stands behind it.
In-app calling launched, framed as an anti-impersonation measure and initiated by Revolut; listed on pricing materials as a Premium, Metal and Ultra benefit, English only.
DemonstratedPolish consumer remedies committed under the UOKiK decision and recorded by the regulator: automatic refunds of overpaid fees, free plan upgrades and 0.2% cashback for two months. Delivery was not independently verified in this investigation.
DemonstratedLithuanian remediation agreed and settled following the €3.5m transaction-monitoring fine; deficiencies acknowledged, remediation plan adopted, no money laundering found.
DemonstratedCustomer service restructured. Recorded in the AGCM decision itself, which is why it is evidenced rather than asserted — but it falls after the conduct period that decision examined.
DemonstratedMiCA authorisation obtained via CySEC for the EEA crypto business.
DemonstratedItalian investment marketing rewritten, including explicit “whole or fractional shares” wording — before the decision landed, after the proceeding opened.
Italy-specificCall identification added, alongside Wealth Protection, spending delays, transaction limits and single-use virtual cards introduced across 2024–26.
DemonstratedUK banking restrictions lifted by the PRA after a twenty-month mobilisation; UK customers begin phased migration from safeguarded e-money to FSCS-protected deposits. Completion status is not public.
DemonstratedNew suspension clause takes effect. The umbrella clause permitting suspension for “reputational damage” or “other reasons” is removed from the EEA terms and replaced. This is the single clearest contractual improvement in the record on the domain at issue.
DemonstratedItalian terms add a reason and a clock. Revolut will tell the customer the reason for a suspension “if the law allows” and give an estimated time to resolve. On the evidence available this is Italy only; whether it has been rolled out to the UK or the rest of the EEA is unresolved.
Italy-specificFurther authorisations: Australian launch (21 July), French banking licence (10 August), and OCC preliminary conditional approval of a US national bank charter (2 September; FDIC and Federal Reserve approvals pending).
DemonstratedRisk and compliance headcount up 42% in a year when total headcount grew 10%; a third of the workforce stated to be in financial-crime prevention. Revolut's own figures, unaudited.
Company-statedWhat CHI will not claim in either direction. There is no evidence that these changes failed, and this page does not say they did. There is also no evidence that they solved the problem globally, and this page does not say that either. The two clearest contractual fixes are dated March and June 2026 — the first group-wide, the second, so far as the record shows, Italian. Both post-date a conduct period that ended in July 2025 and a proceeding opened in July 2025. That sequence is worth stating plainly, and it is not an accusation: a company that fixes something after a regulator arrives has still fixed it.
The case for Revolut
Why this verdict is only weakly supports.
On this page the counter-case is not a courtesy paragraph. It is the reason the verdict sits where it does, and it is built from evidence Revolut does not control: ombudsman adjudications, regulator datasets, consumer-body surveys and licensing decisions. A great deal of what Revolut has done to its customers has been straightforwardly good for them, and some of it has pressured incumbents into improving too.
An independent adjudicator does not overturn it more than its peers
On authorised push payment fraud in January–August 2025 the Financial Ombudsman upheld 30% of complaints against Revolut, against 41% for Monzo, 34% for HSBC and Lloyds and 31% for Barclays. Revolut's fraud-complaint volume is the highest in absolute terms; its error rate on escalated cases is not an outlier.
The scam metric a regulator publishes improved sharply
The Payment Systems Regulator measured the value of scam proceeds received per £1 million of transactions. Revolut's figure fell from £1,158 in 2022 to £756 in 2023 — a 35% reduction — with the per-transaction measure down 32% over the same year. Being a high receiver measures mule onboarding, not how a firm treats its own defrauded customers.
It is inside the mandatory reimbursement regime on standard terms
Since 7 October 2024 Revolut has operated the UK's mandatory APP reimbursement scheme on the standard published terms: the £100 excess, waived for vulnerable customers; five business days to reimburse, extendable to 35; a thirteen-month claim window. It was not a signatory to the earlier voluntary code — which was lawful, and shared by Monzo and every other e-money institution.
The pricing is genuinely cheap, and audited by the market
- 0% currency fee on the free tier to £1,000 a month, then 1%, against a high-street benchmark of roughly 3%. A weekend markup still applies on the free tier.
- Free cash withdrawals to five a month or £200, then 2% with a £1 minimum, against roughly 2.5% and a £3 minimum.
- International transfers at 0.15%, capped at £250 in local currency, with per-currency minimums and a £3–£5 flat charge for non-local currencies.
- Per-transaction rate and fee shown in-app before the customer confirms.
The fair caveat: Chase, Starling, Monzo and First Direct now match much of this. The advantage is against incumbents, not against modern peers.
Its advertising record is clean
In ten years the Advertising Standards Authority has never upheld a complaint against Revolut. There is one formal ruling, in June 2026, and it was not upheld; two older cases were resolved informally. Savings rates are marketed with the qualifying plan named in the headline, and referral-reward odds are published.
Four prudential regulators authorised it in 2026, and a fifth approved it conditionally
The PRA lifted the restrictions on its UK licence on 11 March 2026 after a twenty-month mobilisation; French and Australian authorisations followed in 2026; Mexico's banking launch came in January 2026; the OCC granted preliminary conditional approval of a US charter on 2 September 2026. The ECB has supervised Revolut Bank UAB directly since January 2025. A licence evidences that threshold conditions were met — it is not a regulator's verdict on customer treatment, and CHI does not present it as one.
Customers rate it above the high street
Which?'s September 2025 survey scored Revolut 83%, sixth of twenty-two providers and above every large high-street bank in the set, with five stars for its app. Which? withheld Recommended Provider status, citing the absence of FSCS cover at the time and concerns about fraud complaints; its “contacting customer services” sub-score was three stars. The Revolut sample was 121 respondents within a 6,665-person survey.
The investment business is unremarkable, in the good sense
No enforcement action, no gamification finding, seventeen investment complaints at the Ombudsman in a nine-month period, transfers out available, industry-standard fractional-share terms, expanding crypto withdrawal access, no order-flow payment in the UK or EEA, and MiCA authorisation since October 2025.
The controls are being tightened, and it has been fined for not tightening them sooner
The Bank of Lithuania's €3.5m fine in April 2025 was for identifying too few suspicious transactions, not too many. Revolut reports risk and compliance headcount up 42% in 2025 — its own unaudited figure. Revolut has added Wealth Protection, spending delays, transaction limits, call identification and single-use virtual cards. Scale and profitability — $2.3bn profit before tax in 2025 — reduce the financial pressure to under-fund controls, which is an inference rather than a finding.
What survives against Revolut
- A national consumer authority found the process around account blocks at one branch to be an aggressive and misleading practice, on the company's own documents and figures.
- On those figures, blocks averaged [51–100] days and [35–50]% ran past a month. [10,001–50,000] cases exceeded Revolut's internal standard and [51–100] customers were compensated on narrow eligibility — two separate figures in the decision, which does not link them.
- Group-wide, the enabling design facts are documented: no published retail review clock, no published restriction statistics, support that cannot lift restrictions, and, until 31 March 2026, EEA terms permitting suspension for “reputational damage” or “other reasons”.
- The recourse route that is universal is the one that cannot resolve the problem; the voice channel is listed as a paid-plan benefit and its true boundary is unresolved.
- The clearest contractual fix — a reason and an estimated resolution time — appears, on this record, in one country.
What answers it
- The finding is one jurisdiction, one conduct period, one first-instance decision, contested and under announced appeal.
- AGCM did not find the blocks unlawful, and the Bank of Italy's opinion accepted the controls' legality and the tipping-off constraint.
- On the same submission, [95–99]% of cases were completed within Revolut's internal standard.
- Independent outcome data — ombudsman uphold rates, the regulator's scam-receiving metric, advertising rulings, licensing — is ordinary to good, not hostile.
- Fraud refusal, pricing deception, investment conduct, dark patterns and payments reliability were all tested and none survived.
Neither column cancels the other, and that is the verdict. A page that printed only the left column would describe a company most Revolut customers would not recognise. A page that printed only the right column would erase a regulator's finding, made on the company's own numbers, about what happens to people whose money is held. Weakly supports is what is left when both are held at once.
The ledger
Every regulatory matter located, classified by what was actually found.
Cumulative monetary penalties located across all jurisdictions between 2015 and September 2026 total €15.27 million, of which €11.5m is contested and under announced appeal. No penalty by a UK, Irish, US, Australian or data-protection regulator was located.
| Date | Body | Type | Subject | Measure | Status |
|---|---|---|---|---|---|
| Feb 2019 | ASA → FCA, UK | Referral | Outdoor ads with user statistics Revolut admitted were made up | None public | No public outcome |
| May 2021 | FCA, UK | Sector letter | E-money firms holding themselves out as bank alternatives | Write to customers | Sector-wide, not Revolut-specific |
| Mar 2022 | Bank of Lithuania | Finding | AML internal controls: risk factors, PEP approvals, relationship purpose | €50k + €150k | Closed |
| Nov 2022 | Bank of Lithuania | Warning & fine | Complaints not handled in accordance with law; in-app chat complaints unregistered. Separately, late filing of financial statements | Warning + obligation + external consultant; €70k for the late filing | Closed; entity since merged |
| Mar 2023 | Companies House / BDO, UK | Qualified audit opinion | Scope limitation on £477m of £636m FY2021 revenue | None | Later years clean; not a misstatement finding |
| Jul 2024 | PRA / FCA, UK | Licence with restrictions | Mobilisation; £50,000 aggregate deposit cap | — | Restrictions lifted 11 Mar 2026 |
| Dec 2024 announced Jan 2025 | UOKiK, Poland | Commitment decision | 2020–23 fee changes not notified on a durable medium | Refunds, upgrades, cashback; no fine | Closed; no infringement finding |
| Jan 2025 | ECB / SSM | Supervisory status | Direct supervision of Revolut Bank UAB and Revolut Holdings Europe UAB | — | Ongoing |
| Apr 2025 | Bank of Lithuania | Finding | Transaction monitoring: suspicious operations not always properly identified. No money laundering found | €3.5m | Settled; remediated |
| Jul 2025 | AGCM, Italy | Proceeding opened | Three practices; inspection 8 July; commitments rejected 18 November | — | Led to the March 2026 decision |
| Mar 2026 | AGCM, Italy | Finding · contested | Aggressive and misleading practice in account suspension, limitation and blocking; misleading investment and IBAN claims | €11.5m total (€5m + €5m + €1.5m) | First instance; appeal announced, no filing located |
| Jun 2026 | ASA, UK | Ruling | Revolut Mobile “Unlimited” claim | Not upheld | Closed |
| Sep 2026 | OCC, US | Preliminary conditional approval | National bank charter; capital and product conditions | — | FDIC and Federal Reserve pending |
Scroll the table horizontally on narrow screens
Two matters are deliberately excluded from this ledger as findings. A 2019 press allegation that sanctions screening was switched off was denied, produced no regulatory finding and is not established. A June 2026 report that the ECB imposed supervisory restrictions in July 2025 rests on a single source, the ECB declined to comment, and the status is unknown — it is not an enforcement action and is not treated as one here. Two High Court judgments involving Revolut as a receiving institution decided preliminary and strike-out questions of law; neither is a finding of fault, and one is under appeal.
The verdict
Weakly supports — and the reasons it is not the verdict above or the verdict below are both worth stating.
Why not supports?
A supports verdict would require the restriction-process finding to be established in more than one jurisdiction, or by more than one adjudicator, or for the recourse deficit to rest on independent outcome data rather than design description.
Neither condition is met. The strongest operating finding is one contested first-instance decision, in one country, covering one conduct period, with no appeal yet heard. Outside Italy, what exists is architecture — documented, group-wide, and consistent with the Italian finding, but architecture is not outcome. No equivalent finding, dataset or complaint-outcome signature was located in the UK, Ireland, Lithuania, Australia or the United States, and the investigation looked.
The adversarial review is part of this answer. Of fifty-six proposed findings, thirty-four were defeated in the form proposed and sixteen were narrowed. A case in which six of fifty-six findings survived hostile review intact is not strong enough for supports.
Why not inconclusive?
Because a national regulator examined precisely the conduct at issue, using Revolut's own internal documents and figures, and found the process unfair — and because the design facts that make it possible are not confined to Italy.
Revolut's public guidance that support cannot lift restrictions is not an Italian statement, and neither is the absence of a published retail review clock or of any published restriction statistics. The chat-first architecture, the automated card line, the email route with no service level and the voice channel listed as a paid-plan benefit are group-level facts; until 31 March 2026 Revolut's EEA terms permitted suspension for “reputational damage” or “other reasons”. Those are descriptions of how the system is built, not measurements of how it performs outside Italy — which is exactly why the verdict is weakly supports rather than simply supports.
A €5 million finding on this exact question — one of three components within the €11.5m total — grounded in the company's own submissions and accompanied by a home-regulator process warning in 2022, is more than a lead. The defence review itself conceded that a neutral reviewer would not land on inconclusive.
2015 – c.2019
Thin and largely journalistic: two processor-driven card outages, a denied and never-adjudicated screening allegation, a culture report, an advertising referral with no published outcome. The first freeze reporting appears at the boundary. Nothing in this period reaches primary-source standard on the hypothesis.
Inconclusivec.2019 – 2023
The evidence hardens: the home regulator's 2022 complaints-handling warning, accumulating lock-out reports, the free currency allowance cut and the fee above it doubled, subscription prices raised, audit control failures, and fee changes in Poland notified invalidly. Improvements run alongside — crypto withdrawals opened, crypto fees cut.
Weakly supports2024 – September 2026
The regulatory exhibit arrives, contested, alongside an under-detection fine that cuts the other way. The recourse architecture is unchanged in kind though augmented. The same period contains the mandatory reimbursement scheme on standard terms, peer-level uphold rates, the UK banking licence and FSCS migration, and the first contractual promise anywhere of a reason and a resolution estimate.
Weakly supportsCustomer-friendly for the many; customer-unfriendly for some of the restricted few. One real hostility domain, narrowly evidenced; the rest of the hypothesis falsified on independent data.
Weakly supportsReserved
Revolut ↔ Robinhood
Both investigations captured comparator fields against the other, and both reached weakly supports by entirely different routes. The comparison is not published here. A head-to-head is a separate piece of work, and writing it inside a company page would smuggle in a framing neither investigation tested.
The constraints are recorded now so that the eventual comparison is built correctly rather than conveniently:
Binding non-comparability rules, carried forward
- Bank-first and brokerage-first are not the same severity scale. Losing access to a current account is losing access to the money that pays rent. Losing access to a brokerage position is different in kind. Severity is not comparable raw.
- Foreign exchange is Revolut's founding product and essentially absent from Robinhood's US offering. It must be scored not-applicable for Robinhood, never given a clean sheet.
- Order-flow payment differs by jurisdiction, not by virtue. It is prohibited in the UK and EU, which is why Revolut receives none there — while its US broker entity does receive a share.
- Restriction counts cannot be compared raw. Compare the quality of notice, explanation and recourse, normalised by product and jurisdiction — not the volumes.
- Three protection regimes, not one: FSCS deposits, the Lithuanian deposit-guarantee scheme and US pass-through insurance differ materially, and crypto sits outside all of them.
- Five tiers against one. Revolut's pricing is explicit fees plus spread across five plans; Robinhood's is one subscription and embedded compensation. A fee-percentage comparison flatters one firm; a disclosure comparison flatters the other.
Revolut built one of the best consumer banking products of the last decade, and then built the machinery that decides you are an exception — machinery this record describes far better than it measures.
For the customer who never trips a rule, it is cheaper, faster and better disclosed than the bank it replaced, and an independent survey puts it above every large high-street name. Nearly everything this investigation predicted would be wrong with it turned out not to be: it does not refuse fraud claims at unusual rates, it does not deceive on price, it does not gamify investing, and no regulator or consumer body has found a dark pattern in it.
What it does not publish is how long a review takes. What its own guidance concedes is that the support everyone can reach cannot lift a restriction. And where a consumer authority examined that process directly — in one country, on the company's own figures, in a decision the company disputes and will appeal — what it found was people locked out of their own money without adequate notice, without an effective right to reply and without adequate help.
That is enough to keep Revolut at weakly supports.
It is not enough to call Revolut hostile, and this page does not.
Reading this beside the rest of the index
Every company here is assessed against the same question. Revolut is the one where the answer depends on which customer you are.
Read it against the other financial-disruptor investigation, against the pattern that carries its secondary pricing finding, and against the methodology that explains why no number appears on this page.