Equitystack Pro.
Security

Vetting a Fintech Vendor's Security: A 30-Point Due Diligence Checklist for Banks

Thirty questions to put to any fintech vendor before you sign, mapped to what supervisors in the EU, UK, US, India, Singapore and Australia expect, plus the answers that should end the deal.

Equitystack Pro editorial team · November 2026 · 10 min read

Tier the vendor before you send a single questionnaire

The first decision in any vendor security assessment for banks is how much assessment the vendor needs. A marketing analytics tool and a card processor shouldn't get the same 300-question spreadsheet. If they do, critical vendors get skimmed and trivial ones hold up procurement for weeks.

Supervisors in different regions have landed on a similar model for third-party risk management: classify the vendor by criticality, do due diligence before signing, write specific rights into the contract, monitor during the term, and keep a working exit plan. The details vary by jurisdiction:

RegimeWho it coversWhat it adds to your review
DORA (EU) and EBA outsourcing guidelinesEU banks, payment and e-money institutionsMandatory contract clauses for ICT providers, a register of information on ICT arrangements, exit strategies for critical or important functions
PRA SS2/21 and the critical third parties regime (UK)UK banks and insurersMateriality assessment, audit and access rights, stressed exit planning
Interagency Guidance on Third-Party Relationships (US)Banks supervised by the Fed, OCC and FDICDue diligence scaled to risk across the full relationship lifecycle
RBI Master Direction on IT outsourcing (India)Indian banks and NBFCsRegulator access, concentration review; separate payment data storage rules apply
MAS outsourcing and technology risk guidelines (Singapore); APRA CPS 230 and CPS 234 (Australia)Licensed financial institutionsMaterial arrangement notifications, short incident reporting windows, resilience testing

Use three tiers. A critical vendor supports a critical or important function, holds customer data at scale, or would halt payments or lending if it failed. It gets all 30 points. A high vendor gets points 1 to 24. A low vendor gets points 1 to 6 plus the data-location questions. Write down the tiering rationale, because examiners ask for it.

Advertisement

Certifications: read the scope, not the logo (points 1–6)

The SOC 2 versus ISO 27001 question comes up in most reviews, and buyers often get it wrong. A SOC 2 report is an auditor's attestation on whether specific controls operated over a defined period, with every exception listed. An ISO 27001 certificate shows an accredited body audited the vendor's information security management system. It confirms a working system exists but gives you little control-level detail. US vendors usually lead with SOC 2; European, Indian and Asian vendors more often hold ISO 27001. Either can work for a critical vendor, but you still need the detail behind it.

  • 1. Type II, recent. Ask for a SOC 2 Type II report covering at least six months, ideally twelve, with a period end inside the last six months. If it's older, ask for a bridge letter. A Type I report shows only that controls were designed, at a single point in time.
  • 2. The right system in scope. Check that the product you're buying appears in the system description, and which trust services criteria are covered. A Security-only report says nothing about availability or processing integrity.
  • 3. A current ISO certificate. It should be ISO/IEC 27001:2022; the transition period for the 2013 version has ended. Verify the certificate with the issuing body, and read the scope statement and the Statement of Applicability.
  • 4. Exceptions and opinion. Read the exceptions and management responses. A qualified opinion on access or change management needs an explanation before you go further.
  • 5. CUECs and carve-outs. Complementary user entity controls are controls you must operate yourself. Carved-out subservice organizations, often the hyperscaler, aren't covered by the report at all.
  • 6. Sector attestations. If the vendor touches card data, ask for a PCI DSS Attestation of Compliance against the current version. Where relevant, ask for a SWIFT CSP attestation and CSA STAR for cloud services.

Penetration test evidence and technical controls (points 7–12)

Vendors often say "we pen test annually" without saying what was tested or what happened to the findings. Ask for documents, not assurances.

  • 7. Independent test within 12 months. It should cover the in-scope application and its APIs and come from a qualified third party, such as a CREST-accredited firm or equivalent. An executive summary with scope and methodology is the minimum. Many vendors share full reports under NDA.
  • 8. Remediation evidence. Get critical and high findings with fix dates and retest confirmation. A critical finding still open after 30 days is a serious warning sign.
  • 9. Your right to test. Secure the right to commission your own test, or to include the vendor in threat-led testing under schemes such as TIBER-EU, DORA TLPT, CBEST or Hong Kong's iCAST.
  • 10. Encryption and keys. Expect strong encryption at rest (AES-256 is the norm) and TLS 1.2 or higher in transit. Ask whether customer-managed or HSM-backed keys are available, and who at the vendor can reach them.
  • 11. Privileged access. Is MFA enforced for all administrative access? Is there a privileged access management tool? How do vendor staff reach production data? How long are logs kept, and can you get them?
  • 12. Secure development. Ask about dependency scanning, whether they can supply a software bill of materials, and their patch SLAs. For critical vulnerabilities, days rather than weeks is the reasonable expectation.

Data residency and the sub-processor chain (points 13–18)

Cross-border banks most often find surprises here. A vendor can host data in Frankfurt and still give a support team in another country full read access, and many regulators treat that access as processing.

  • 13. Storage, processing and access locations. Get all three in writing. Some rules impose strict localization. India's payment data storage requirement is one example, and several Gulf and Southeast Asian regulators restrict offshore processing of banking data.
  • 14. Transfer mechanism. For personal data, identify the legal basis for each cross-border flow: EU standard contractual clauses, the UK IDTA, or the transfer provisions in India's DPDP Act or Brazil's LGPD, for example. Have privacy counsel confirm it, since requirements differ by country.
  • 15. A complete sub-processor list. For each, get the name, function, location and the data it touches. "Available on request" isn't good enough for a critical service.
  • 16. Change notice and objection. Expect 30 to 90 days' notice of new sub-processors, plus the right to object or terminate if a change breaches your regulatory constraints.
  • 17. Flow-down. Audit, confidentiality and incident-notification obligations should pass to material subcontractors. DORA requires this for subcontracting that supports critical functions.
  • 18. Fourth-party hosting. Ask which cloud provider and regions they use, and whether the service fails over between regions. A single-region deployment on one hyperscaler is a resilience risk, not a technical footnote.

Incident notification windows and resilience (points 19–24)

Your vendor's notification deadline must be shorter than your own deadline to the regulator. If it isn't, you'll miss your filing whenever the vendor uses its full allowance. Typical regulatory windows (confirm current rules with counsel, as thresholds and definitions change):

JurisdictionBank's reporting window (general)Practical vendor clause to seek
EU (DORA)Initial notice within hours of classifying an incident as major, with an outer limit of about 24 hours from awarenessNotice within 1–4 hours of detection
US banking agenciesWithin 36 hours of determining a notification incident; bank service providers must notify affected banks as soon as possibleNotice within 4–24 hours
India (CERT-In)Within 6 hours of noticing certain cyber incidentsNotice within 1–2 hours
Singapore (MAS)Within 1 hour of discovering a relevant incidentImmediate notice, with a named 24/7 contact
Australia (APRA CPS 234)Within 72 hours for material incidentsNotice within 24 hours
  • 19. The clock starts at detection. Not at "confirmation" or "determination," because the vendor controls when those happen.
  • 20. Notice content and cooperation. The contract should specify what each notice contains, how often updates arrive, and your access to forensic findings and logs.
  • 21. Tested RTO and RPO. Get recovery time and recovery point objectives for your service specifically, with disaster recovery test results from the last 12 months.
  • 22. Ransomware recovery. Ask for immutable or offline backups and evidence of a full restore test.
  • 23. Cyber insurance. Ask for the coverage limit and confirm third-party liability is included. Insurance cover and payouts vary by country, so have your broker review the policy.
  • 24. Incident and enforcement history. Ask about material breaches and regulatory actions over the last three years, and what changed afterward.

Concentration risk, audit rights and exit (points 25–30)

Outsourcing due diligence checklists tend to be weakest here, and supervisors have pushed hardest on it since DORA and the UK critical third parties regime took effect.

  • 25. Concentration. Map how many of your critical services already depend on this vendor or its hyperscaler. If three payment services run on one cloud region, a fourth deepens a concentration you already have.
  • 26. Audit and access rights. You need unrestricted rights for your own staff, your auditors and your supervisor. Pooled audits with other client banks are a reasonable way to deliver this. Refusing supervisor access is not.
  • 27. Financial viability. Ask for audited accounts. For venture-backed fintechs, ask about cash runway too. If the code is proprietary and critical to you, consider source code escrow.
  • 28. Termination triggers. You should be able to terminate without penalty on a regulatory direction, a material security breach, a change of control, or an unapproved sub-processor change.
  • 29. Data return and transition. Agree the export format, return timeline, a deletion certificate, and transition assistance for roughly 6 to 18 months at pre-agreed rates.
  • 30. A documented exit plan. Write down how you'd move the service elsewhere or bring it in-house, and test it on paper at least once a year. DORA requires exit strategies for ICT services supporting critical or important functions.

Answers that should stop the deal, and what the review costs

Not every gap should kill a contract. Some can be fixed with contract terms or remediation deadlines. Others mean the vendor can't legally or safely serve a regulated bank. Sort them early so procurement knows where it stands.

Vendor answerVerdict
Refuses audit or access rights for your supervisorStop. Most regimes above treat this right as non-negotiable for material outsourcing.
Won't disclose sub-processors or their locationsStop. You can't complete your register or assess residency.
Data processed or accessed somewhere your regulator prohibitsStop, unless the architecture can change before go-live.
No independent pen test in 12 monthsPause. Require a test and remediation before signing.
SOC 2 Type I onlyConditional. Acceptable for a young vendor if the contract requires a Type II report within 12 months.
Notification window longer than your regulator'sNegotiate. Don't sign until the vendor's window is shorter than yours.
No exit assistance or proprietary-only data exportNegotiate. If the vendor won't move, escalate to the risk committee.

Costs depend on vendor tier and how much work you do in-house. As typical 2026 market ranges that vary by contract: a full internal review of a critical vendor often takes several dozen to a hundred-plus staff hours. External assessors commonly charge from roughly $10,000 to $60,000 or more per critical vendor. Third-party risk platforms such as OneTrust, ProcessUnity or Mitratech's Prevalent typically run from tens of thousands to a few hundred thousand dollars a year, depending on vendor count and modules. Security ratings services such as SecurityScorecard or BitSight usually add somewhere in the tens of thousands. Local pricing differs, so confirm figures with providers.

Hypothetical example: Say a mid-sized bank is evaluating a payments fintech for real-time transfers. That's a critical vendor, so all 30 points apply. The vendor holds ISO 27001:2022 but no SOC 2. Its pen test is 14 months old, and its support team accesses data from a third country. The bank requires a new pen test before signing, has privacy counsel confirm the legal basis for the support-access transfer, and asks for ISO audit findings in place of a SOC 2 report. The deal closes several weeks later than planned, with a better contract.

Decision rule: any "stop" answer from a critical vendor goes to the risk committee before anyone negotiates price. Every "negotiate" item belongs in the signed contract, not in a side email.

Common questions

Should a bank accept ISO 27001 instead of a SOC 2 Type II report?

Often, yes, particularly from vendors outside the US where ISO 27001 is the norm. The certificate tells you less about how individual controls perform, so ask for the Statement of Applicability, the scope statement and recent audit findings to fill the gap. For critical vendors, many banks also send a targeted control questionnaire.

How often should a critical fintech vendor be reassessed?

Most banks fully reassess critical vendors every year, with monitoring in between through security ratings, new SOC reports and incident notices. High-tier vendors are often reviewed every 18 to 24 months. A breach, an acquisition or a major new sub-processor should trigger an immediate review whatever the schedule says.

Can we rely on a pooled or shared assessment instead of our own audit?

Many supervisors accept pooled audits and industry shared assessments if the scope covers your services and you can see the results. You still own the risk decision, and the contract must preserve your own audit rights and your regulator's. Treat shared assessments as evidence, not a replacement for your judgment.

What incident notification window should we write into the contract?

Start from your shortest regulatory deadline and work backward, leaving time for internal triage before you file. For EU, Indian or Singapore banks, that usually means vendor notice within one to four hours of detection. Have counsel confirm the window for each jurisdiction you operate in.

Do these requirements apply to small fintech vendors that aren't regulated themselves?

Yes. The obligations sit with the bank, so it must secure the rights and evidence through the contract whether or not the vendor is regulated. Smaller vendors may need time to produce a SOC 2 Type II report or a tested exit plan. Contractual milestones with termination rights are the usual way to bridge that gap.

Written by the Equitystack Pro editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.