Why banks reject high-risk businesses — and how to fix it before you reapply.

Most rejection notices are deliberately vague. Institutions have legal and reputational reasons not to fully explain a decline: it can expose them to a dispute, it can reveal underwriting criteria they would rather not publish, and in some cases it can create a paper trail an applicant could use against them. The result is that a business receives a one-line decline and is left to guess at the cause, which is precisely the situation that leads to a second, third and fourth rejection when the applicant resubmits an unchanged file elsewhere.

In our experience running files across a network of more than 120 institutions, the specific causes behind a generic decline cluster into a small, repeatable set: unclear or unverifiable beneficial ownership, a mismatch between the stated activity and the actual licence held, a weak or absent source-of-funds narrative, exposure to prohibited or high-risk jurisdictions, an inadequate or templated AML policy, unrealistic financial projections, a sloppy or non-compliant website and terms of service, and an undisclosed prior account closure. Almost every rejection we have helped a client work through traces to one or more of these eight causes.

This guide works through each cause in turn: what the underwriter is actually looking for, why the gap ends the application, and specifically how to remediate it before resubmitting. It closes with a pre-reapplication checklist that should be worked through in full before a second application goes anywhere, because a partial fix produces the same outcome as no fix at all.

One framing point before the detail: fixing the cause of a rejection is not the same as making the file look better. Underwriters read a large number of files and are trained to distinguish a genuine remediation from a cosmetic one — a business narrative rewritten to sound more confident without any underlying change will not clear a committee that declined the same business six weeks earlier for a documented reason. The remediation has to be real, evidenced and reflected consistently across every document in the file.

For the broader mechanics of how high-risk banking works — sectors affected, institution types, jurisdiction effects and layered account architecture — see our companion guide, High-Risk Business Banking: How to Get a Bank Account When Everyone Says No, which this guide assumes as background context.

Beneficial ownership is the single most common cause of stalled or declined applications, and it is rarely because ownership is being deliberately concealed. Far more often it is because the ownership chain has grown organically over years — a holding company added for an earlier tax or investment reason, a trust structure set up by a different adviser for a different purpose, a nominee shareholder retained from an early-stage arrangement that was never unwound — and nobody has produced a single document that shows the whole chain clearly.

What an underwriter needs is a beneficial ownership chart that traces, in one diagram, every layer from the applicant entity down to the natural persons who ultimately own or control it above the applicable threshold, typically 25% but sometimes lower depending on the institution's own policy. Every intermediate entity in that chain needs its own constitutional documents available, and every natural person at the bottom needs certified identification, proof of address, and screening against sanctions, politically exposed persons lists and adverse media.

The specific failure modes we see repeatedly: a nominee shareholder or director whose relationship to the beneficial owner is not documented anywhere in writing, meaning the underwriter cannot establish who actually controls the vote; a trust or foundation in the chain with no deed or letter of wishes provided, leaving the underwriter unable to determine who the effective controller is; bearer shares or an equivalent instrument that defeats verification entirely, which is close to an automatic decline at any institution with a competent compliance function; and ownership percentages that do not sum correctly across the chain, which reads as sloppy at best and evasive at worst.

Remediation is mechanical but has to be done properly rather than papered over. Unwind or fully document nominee arrangements — either replace the nominee with the actual beneficial owner on the register, or obtain and produce a formal nominee declaration that clearly states who the nominee acts for. Where a trust sits in the chain, obtain and be prepared to produce the trust deed, the schedule of beneficiaries, and identification for the settlor, trustees and any beneficiary with a vested interest. Convert any bearer share arrangement to registered shares before applying anywhere serious; no reputable institution will accept an unresolved bearer share structure in 2026. Finally, produce the single-page ownership chart yourself rather than leaving the underwriter to reconstruct it from a stack of certificates — this alone resolves a meaningful share of ownership-related delays.

Where the ownership structure genuinely is complex for legitimate commercial reasons — a multi-jurisdictional group with several operating entities and a holding structure built for real operational reasons — that complexity is manageable, but it needs to be explained in writing as part of the file rather than left for the underwriter to puzzle out unassisted. A complex structure with a clear explanation clears far more often than a simple structure with an unexplained gap.

A second recurring cause of decline is a gap between what the business's licence actually authorises and what the business narrative or transaction flow suggests it is actually doing. This mismatch is treated seriously because it is precisely the pattern regulators tell institutions to watch for: an entity operating beyond its authorised scope, whether deliberately or through simple scope creep as the business has grown.

Common examples: a payment institution licensed for money remittance whose transaction flow shows patterns consistent with acquiring or e-money issuance, activities its licence does not cover; a gaming operator licensed in one jurisdiction whose customer base, evident from IP or currency data in the transaction narrative, clearly includes customers in jurisdictions where that licence has no effect and a separate authorisation would be required; a virtual asset business licensed for exchange services whose described activity includes custody, which frequently falls under a different, more demanding authorisation category. In each case the underwriter is not necessarily accusing the applicant of bad faith — often the business has genuinely grown into adjacent activity without updating its licensing — but the mismatch itself is enough to end the application until it is resolved.

The remediation path depends on which side of the mismatch is wrong. If the licence is narrower than the actual activity, the business needs to either apply for the additional authorisation before reapplying for banking, or genuinely scale back the activity to what the existing licence covers and reflect that honestly in the narrative and transaction projections. If the described activity in the application is broader than what the business actually intends to do, the fix is simpler: narrow the narrative to match reality, because an overstated business description creates exactly the same red flag as an understated licence.

Where an activity is arguably outside any licensing perimeter at all — a genuinely unregulated holding company, a software vendor selling to licensed operators without itself touching regulated activity, a treasury entity that only manages the group's own funds — that position needs a written, reasoned legal analysis attached to the file, ideally from qualified counsel in the relevant jurisdiction, rather than an unsupported assertion in the business narrative. An underwriter who reads "we believe this falls outside the regulatory perimeter" with no supporting analysis will treat it as an unlicensed operator by default.

The broader lesson is sequencing: licensing decisions need to be finalised, or at minimum genuinely in progress with documentary evidence of the application, before a banking application goes out, rather than treated as something to sort out in parallel. A banking application submitted while a licensing gap is still open rarely survives contact with a competent compliance team, and resubmitting the same activity description after the licence is actually granted is a stronger second application than trying to argue the gap away.

Source of funds and source of wealth are related but distinct concepts that applicants routinely conflate, and the conflation itself is often what causes an underwriter to lose confidence in the file. Source of wealth explains how a beneficial owner's overall wealth was generated over time — a prior business sale, an inheritance, accumulated employment income, investment returns. Source of funds explains where the specific money funding this particular account or transaction came from right now. A file that documents one but not the other leaves an obvious gap a competent underwriter will always ask about.

The most common failure is an assertion without evidence: a statement that funds derive from "business income" or "personal savings" with no bank statement, no sale agreement, no tax filing, no brokerage record attached to support it. Compliance teams are trained specifically to be unpersuaded by assertions and to require documentary evidence, and a narrative that reads confidently but produces nothing verifiable when checked is treated more harshly than an honest gap that is flagged and being worked on.

For founders whose wealth originated in digital assets, private equity, or another less conventional source, the documentation challenge is real but solvable: historic exchange records and wallet statements showing acquisition and disposal history, fund administrator statements, share purchase and sale agreements, and where available a professional valuation or audit opinion covering the relevant period. The underlying principle does not change just because the asset class is unconventional — the underwriter needs a documented, traceable chain from an identifiable original source to the funds sitting in the account today.

A second common failure is a mismatch between the scale of funds being introduced and the documented wealth or income of the beneficial owner — an account being funded with an amount that is disproportionate to any evidenced source produces an immediate and difficult-to-resolve red flag, and no amount of narrative confidence overcomes a numerical inconsistency of this kind. Where a discrepancy of this sort genuinely exists for a legitimate reason — a recent liquidity event, a co-investment, a loan from a documented third party — that reason needs to be explained and evidenced explicitly rather than left for the underwriter to notice and query.

Remediation here is entirely about assembling documentation that already exists rather than manufacturing anything new: bank statements covering the relevant period, the sale or liquidity event documentation, tax filings consistent with the stated income, and a written narrative that ties the documents together into a coherent story an underwriter can follow without a follow-up call. Where genuine gaps exist because records were not kept at the time, a sworn statement or an accountant's letter reconstructing the position, while weaker than contemporaneous records, is materially better than silence.

Jurisdictional exposure ends applications regardless of how strong the rest of the file is, because it is frequently treated as an override factor in an institution's risk model rather than one input among several. This applies on both sides of the relationship: the jurisdiction where the applicant entity is incorporated or managed, and the jurisdictions where its underlying customers or counterparties are located.

The clearest and least negotiable category is exposure to sanctioned jurisdictions or sanctioned parties — any evidence of dealings, even indirect, with entities or individuals on an applicable sanctions list is close to an automatic and often permanent decline, and no amount of additional documentation elsewhere in the file overcomes it. The second category, Financial Action Task Force grey or black list jurisdictions, is more negotiable but still heavily weighted: an applicant incorporated in, or with material customer concentration in, a listed jurisdiction faces materially enhanced scrutiny and, at many institutions, an outright policy exclusion regardless of individual circumstances.

A third, less obvious category that catches many applicants is customer geography that has simply not been analysed. A business may be cleanly incorporated and licensed but have never actually reviewed where its customer base sits geographically, and an underwriter who requests this breakdown and finds meaningful concentration in a restricted jurisdiction will decline the file even where the applicant had no intention of specifically targeting that market — the exposure exists regardless of intent.

Remediation starts with an honest audit: map the actual jurisdictional footprint of customers, suppliers and counterparties, not the intended or assumed footprint. Where restricted-jurisdiction exposure is identified, the options are to genuinely exit that exposure — geoblocking, contractual restrictions, active enforcement rather than a policy that exists only on paper — or, where exit is not commercially realistic, to build and document a specific enhanced monitoring programme for that exposure and present it transparently as a managed risk rather than an unaddressed one. An institution is far more likely to accept a modest, honestly disclosed and actively managed exposure than to discover an undisclosed one during due diligence.

It is worth being direct about the limits here: some jurisdictional exposure is simply not fixable in a way that will satisfy a mainstream institution's committee, and in those cases the honest advice is to restructure the customer base or accept a narrower set of institutions willing to underwrite that specific exposure, rather than to continue reapplying broadly with the same unresolved geography.

An anti-money-laundering policy that has clearly been downloaded, lightly edited and submitted without genuine adaptation to the business is one of the fastest ways to lose credibility with a compliance reviewer, because it signals that the applicant has not actually thought through how illicit activity could occur within its own specific business model — which is precisely what the policy is supposed to demonstrate.

The tell-tale signs of a templated policy are easy for an experienced reviewer to spot: generic risk factors that do not correspond to the business's actual products or customer base, thresholds and procedures copied from a different jurisdiction's regulatory framework, no named individual actually responsible for compliance, and no evidence that the policy has ever been applied — no sample suspicious activity reports, no training records, no evidence of periodic review or board approval.

A policy that will actually satisfy an underwriter needs to be built around the business's real risk profile: specific customer types and how each is verified at onboarding and re-verified periodically, the specific transaction monitoring rules and thresholds in place with the actual tooling named, a defined escalation path from a monitoring alert to a suspicious activity report with the specific individual responsible named and their qualifications stated, a sanctions and PEP screening process with the actual screening provider identified, and a record of staff training that has genuinely taken place rather than a training plan that exists only on paper.

For businesses without an internal compliance function capable of producing this independently, engaging a qualified compliance consultant to build a policy specific to the business, rather than adapting a template, is worth the cost relative to the delay and reputational cost of a decline. The policy does not need to be lengthy or elaborate — a focused, accurate ten-page document that clearly reflects the actual business is far more persuasive than a generic fifty-page document that reads as boilerplate.

Evidence of the policy being lived, not merely written, matters as much as the document itself. Institutions increasingly ask for sample outputs — an anonymised example of a monitoring alert and how it was resolved, a record of a recent training session, minutes of a compliance committee meeting — and an applicant able to produce these on request demonstrates a functioning compliance culture in a way no policy document alone can.

Financial and transaction projections are read by underwriters as a signal of both credibility and future monitoring risk, and applicants routinely damage their own files by submitting projections built to impress rather than projections built to be accurate. An aggressive projection that the business is unlikely to meet creates two separate problems: it undermines the underwriter's confidence in the applicant's judgement at the outset, and it sets up an inevitable mismatch between projected and actual volumes in the months after onboarding, which is itself a standard trigger for enhanced review or account closure.

The specific patterns that raise concern: monthly transaction volumes that imply an implausible customer acquisition rate for a business with no operating history to support it; average transaction sizes inconsistent with the stated customer base, such as a retail consumer product projecting institutional-scale ticket sizes; a currency or counterparty mix in the projection that does not match the geography described elsewhere in the file; and projections presented with no supporting methodology, as a bare set of numbers with no explanation of how they were derived.

A projection that clears committee well is conservative, methodologically transparent, and internally consistent with every other document in the file. It should show the assumptions behind the numbers — customer acquisition assumptions, average order value, expected growth rate — rather than presenting a bare table, and it should be phased realistically, typically showing a ramp-up period rather than immediate full-scale volume from month one, which reflects how most businesses actually operate and is more credible to an underwriter who reviews these files professionally.

Where a business genuinely does expect rapid scale — a proven model expanding into a new market, for instance — that expectation needs to be substantiated with evidence from the existing business (actual historical volumes, existing customer contracts, a signed distribution agreement) rather than presented as an unsupported forecast. Evidence-backed ambition is credible; unsupported ambition is not.

Finally, projections submitted at onboarding are not a one-time formality — they become the baseline against which the institution monitors the live account. A business that keeps the institution informed as actual performance diverges from the original projection, proactively rather than only when asked, experiences materially fewer monitoring holds and a much lower risk of an unexpected review triggering a relationship-ending question.

It is common for applicants to underestimate how closely a compliance reviewer examines the customer-facing side of the business — the website, the app, the terms and conditions, the pricing page — treating the corporate documentation as the entire file when in practice the public-facing presentation is reviewed just as carefully, because it is the clearest evidence of what customers are actually being offered and how the business actually represents itself in the market.

Specific issues that trigger concern: a website that describes services the licence does not cover, or that markets to jurisdictions the business claims not to serve; pricing that is unclear, hidden, or inconsistent between the website and the terms and conditions; terms and conditions that are generic, clearly copied from an unrelated business, or missing entirely — a business processing customer payments with no visible refund, cancellation or dispute resolution policy is an immediate concern for chargeback-sensitive sectors; marketing claims that are unsubstantiated or potentially misleading, particularly in sectors like nutraceuticals, forex or crypto where regulators actively police promotional claims; and no visible compliance information at all — no registered company details, no licence number displayed where required, no data protection or privacy policy.

The remediation here is often the fastest and cheapest fix in this entire guide, because it does not require new licensing or new documentation from third parties — it requires the business to actually review and correct its own public materials. This means auditing the website and app against the licence actually held, ensuring pricing is transparent and consistent across every channel, publishing complete and specific terms and conditions rather than a generic template, removing any marketing claim that cannot be substantiated, and displaying the licence number, regulator and registered entity details clearly where the relevant regime requires it.

This step is worth completing before any banking application is submitted, not after a decline, because it is one of the few elements of the file an underwriter can verify independently and immediately, without waiting for a document request to be answered. A business whose public-facing presentation does not match its stated licensing and activity creates doubt about the rest of the file before the underwriter has even reached the corporate documents.

A related point: screenshots or a short written summary of the customer journey — how a customer signs up, verifies their identity, funds their account and can request a refund — proactively included in the application file demonstrates operational maturity that a bare set of corporate documents does not convey, and pre-empts a request that would otherwise slow the review down.

Failing to disclose a previous account closure is, in our experience, the single most damaging thing an applicant can do to a banking application, and it is damaging regardless of whether the original closure had a legitimate, non-sinister explanation. Institutions increasingly run background checks and share informal intelligence within the sector, and a closure that surfaces during due diligence rather than being disclosed upfront is treated as evidence of concealment, which converts a manageable fact into a trust problem that is far harder to repair.

The pattern we see repeatedly: a business was offboarded by a previous bank or EMI, often for a portfolio-level reason unrelated to the specific business — a sector-wide de-risking decision, a correspondent bank pulling back from a category — and the applicant, embarrassed or worried it will count against them, omits it from subsequent applications on the assumption it will not be discovered. It is discovered far more often than applicants expect, because reference checks, adverse media screening and informal relationship-manager networks routinely surface prior banking history.

The correct approach is full disclosure paired with a clear, factual explanation and, where relevant, evidence of remediation. State which institution, roughly when, and the reason given at the time, even if that reason was vague. Where the closure was a portfolio-level decision unrelated to conduct, say so plainly — this is a common and entirely understandable outcome that underwriters are used to seeing and do not hold against an applicant nearly as harshly as an undisclosed gap. Where the closure followed a specific concern, whatever it was, explain what has changed since: a licensing gap closed, a compliance policy rebuilt, a problematic counterparty relationship terminated.

Preserving the paper trail from the original relationship — statements, the closure notice itself, and any correspondence explaining the reason — is valuable, because a well-documented account of what happened and what was fixed is considerably more persuasive than a verbal assurance, and the next institution will very likely ask for exactly this material during due diligence regardless.

It bears repeating because it is so consistently underestimated: disclosed and explained is a manageable fact that experienced underwriters see regularly and know how to assess fairly. Discovered and undisclosed is, at most serious institutions, treated as close to a permanent disqualifier, both for that specific application and, informally, for future applications where the same reviewer or their contacts are involved.

Before resubmitting any application, work through this list in full and honestly, rather than addressing only the single issue most likely to have caused the original decline. Underwriters frequently find a second issue once the first is fixed, and a file that has been comprehensively reviewed clears far more consistently than one that has been patched at a single point.

Ownership and structure: produce a single-page beneficial ownership chart tracing every layer to natural persons; resolve or fully document any nominee arrangement; unwind any bearer share instrument; obtain trust deeds and beneficiary schedules for any trust in the chain; confirm every ownership percentage sums correctly across the structure.

Licensing and activity: confirm the licence held genuinely covers the activity described in the file and the activity actually being conducted; complete or evidence in-progress any additional licensing the activity requires; obtain a written legal opinion for any activity claimed to sit outside a regulatory perimeter; ensure the business narrative describes activity consistent with the licence, neither broader nor narrower.

Funds and financials: assemble documented, evidenced source of wealth for every controlling beneficial owner; assemble documented source of funds for the specific capital funding the account; reconcile the scale of funds being introduced against the evidenced wealth of the owners; rebuild financial and transaction projections conservatively, with stated assumptions and a realistic ramp-up phase, consistent with every other document in the file.

Geography, compliance and presentation: map and honestly disclose the actual jurisdictional footprint of customers and counterparties; document any restricted-jurisdiction exposure and the specific controls managing it, or genuinely exit that exposure; rebuild the AML policy around the business's real risk profile with a named compliance lead and evidence it has actually been applied; audit the website, app, pricing and terms and conditions against the licence and correct any mismatch; disclose any prior account closure fully, with the reason given and the remediation taken, and assemble the supporting paper trail. Only once every item on this list has been genuinely addressed, not merely asserted, is a file ready to go back out — and at that point it is generally worth engaging a specialist adviser to confirm the file is complete and to match it against institutions whose current appetite fits the specific, now-corrected profile, rather than resubmitting speculatively to whichever institution is easiest to reach.