On this page
In brief
- Core banking platforms maintain customer accounts, balances, transaction records and product rules, while payment connectivity and regulated services may sit with separate providers.
- Compare Mambu, Thought Machine Vault Core, Temenos and Engine by Starling by their capabilities, deployment options, customisation needs and UK references.
- Validate ledger accuracy, APIs, integrations and Faster Payments, Bacs and CHAPS connectivity through realistic tests, including failure and reconciliation scenarios.
- Build a three-to-five-year GBP cost model that includes implementation, migration, third-party services, staffing, support and exit costs, using dated quotes or clearly labelled assumptions.
- Assess regulatory responsibilities, safeguarding, operational resilience and data portability, then select a delivery partner and migration plan using evidence, delivery gates and a weighted scorecard.
This guide compares Mambu, Thought Machine Vault Core, Temenos and Engine by Starling, and explains how to assess costs, UK payment connectivity, regulatory responsibilities and implementation risks before choosing.
1.1 What Does a Core Banking Platform Do?
A core banking platform is the system of record that maintains customer accounts, balances and transaction histories. At its centre is the banking ledger: the controlled record of debits, credits and account positions used to determine what each customer owns, owes or can spend. Depending on the product scope, the banking core also applies product rules for fees, limits, interest accrual, account status and transaction posting, while producing the data required for reconciliation, finance and regulatory reporting.
For current accounts, wallets and stored-value products, core banking services typically cover account creation and lifecycle management, booked and available balances, transaction records, fees and interest calculations. A broader core may also support deposits, overdrafts and lending by maintaining principal balances, repayment schedules, accrued interest and other product-level obligations. The precise boundary varies by platform: a core designed for a digital bank may contain substantially more product logic than a ledger designed primarily to record payment movements.
The banking core should not be confused with the systems connected to it. A payment gateway or payment orchestration layer connects the product to payment methods, processors or schemes and manages the movement or routing of payment instructions. A customer interface provides the web or mobile experience through which users view accounts and initiate transactions. KYC and AML systems support identity verification, customer due diligence, sanctions screening and transaction monitoring. These systems can initiate, enrich or assess activity, but the core generally remains the authoritative source for accounts, postings and balances.
System | Primary role |
Core banking platform | Maintains accounts, ledger postings, balances, transaction records and product rules such as fees and interest. |
Payment gateway or orchestration layer | Routes payment instructions and connects to processors, banks or payment schemes. |
Customer interface | Provides the web or mobile experience for account access and transaction initiation. |
KYC/AML tools | Support identity verification, customer due diligence, sanctions screening and transaction monitoring. |
This distinction matters when assessing UK payment connectivity. Access to Faster Payments, Bacs or CHAPS is not automatically a feature of the ledger itself. A core banking platform may expose APIs and transaction workflows that support those payment types, while scheme access, sponsor-bank arrangements, payment processing and settlement are provided through separate infrastructure. Buyers should therefore distinguish between ledger capability and the external connectivity required to execute and settle payments.
Buying or licensing core banking software also differs materially from using Banking-as-a-Service, or BaaS. With core software, the vendor normally supplies technology rather than regulated banking services. The fintech, bank, electronic money institution or payment institution operating the platform remains responsible for its customer proposition, regulatory permissions and regulated activities, subject to the structure of its business and any outsourced arrangements.
Under a BaaS model, a regulated provider supplies some combination of accounts, payment services, e-money issuance, safeguarding arrangements, payment-rail access and compliance infrastructure under its own regulatory permissions. The BaaS provider is accountable for the regulated activities it performs, but this does not mean that every compliance obligation or aspect of the customer relationship automatically transfers to it. The contractual and regulatory model determines which entity is the legal service provider, which entity contracts with the end customer, who performs customer due diligence and monitoring, and how complaints, safeguarding, disclosures and other regulatory obligations are allocated. A fintech may still control the brand and customer interface even where a regulated provider supplies the underlying financial service.
The practical choice therefore starts with the required system of record. A full core banking platform is usually justified where the business needs to manage multiple account or lending products, maintain complex product rules, calculate interest or fees, support high transaction volumes and provide a reliable ledger across a substantial financial-services operation. It is particularly relevant where the business must control product configuration, account lifecycle and transaction processing rather than simply pass payment instructions to another regulated provider.
A narrower ledger or payments platform may be sufficient where the product has limited account logic, relies on a regulated partner for underlying accounts or safeguarding, or mainly needs to record balances and orchestrate payments. In those cases, introducing a complete banking core can add implementation, integration and operating complexity without creating corresponding value. The critical requirement is that the chosen architecture provides an accurate and auditable financial record while matching the actual product, regulatory and operational responsibilities of the business.
1.2 Compare Platforms, Deployment Models and Custom Development
Compare the platforms against the products and operating model you need, not just their feature lists. A current account, savings account, loan and payment account can place different demands on account servicing, transaction processing, product rules and reporting. Also check what is included in the proposed licence or service: a feature described by a vendor may depend on a separately licensed module, a partner integration or development by your team or implementation partner.
Capability | Mambu | Thought Machine Vault Core | Temenos | Engine by Starling |
Account management | Core: deposit, savings and loan accounts, client records and account-management APIs. Confirm which account types and lifecycle features are included in the proposed service. | Core: current and transaction accounts, linked accounts and real-time balance management. Confirm the product scope and account features required for your use case. | Core: broad retail and corporate banking functionality, with APIs for account servicing. Confirm the selected Core components, local market configuration and licence scope. | Platform capability: account, balance and customer-data management across spending, saving, borrowing and investing propositions. Confirm which products are in the proposed implementation. |
Transaction processing | Core: account transactions and ledger journal entries. Mambu Payments provides payment capabilities, but UK scheme access, sponsor arrangements and settlement responsibilities must be verified separately. | Core: real-time ledger postings through Vault Core APIs. Thought Machine also offers Vault Payments; confirm whether it is in scope and how payment access and settlement will be provided. | Core: banking transaction processing and APIs. Temenos Payments is a separate capability in its broader portfolio; confirm whether it is included and which payment integrations are required. | Platform capability: Engine describes payment and card processing as part of its platform. Verify which UK payment types, gateways, sponsorship and settlement services are included in the proposal. |
Product configuration | Core: configurable deposit and lending products, with APIs to manage product configuration. Configuration-as-code workflows are documented; confirm tooling and governance for the chosen product scope. | Core: configurable product logic, including limits, interest rates and fees. Confirm how product changes are designed, tested and deployed, and what specialist development support is required. | Core and market configuration: broad product functionality and UK Model Bank capabilities are available. Confirm which model-bank components and other modules are licensed for the implementation. | Platform capability: configurable products across several banking propositions, with products that can be enabled and launched in stages. Confirm the degree of configuration available for the target products. |
Reporting and data | Core: accounting reports include balance sheet, profit and loss, and trial balance. Mambu Insights/data-lake capabilities and export scope should be confirmed separately. | Core platform data: ledger information can be accessed through APIs and event streaming. Confirm which operational, finance and regulatory reports are included and which require a data platform or integration. | Scope-dependent: reporting depends on the selected Core components and related products. Request the report catalogue, sample outputs, export formats and any separate data or analytics requirements. | Platform capability: Engine describes standard reports, dashboards and a Data Feed API for analytics, finance and regulatory reporting. Confirm report content, access and export entitlements. |
Extensibility | Core: REST APIs support account management, transactions and configuration. The Streaming API is documented as an enterprise feature; confirm its commercial inclusion. | Core platform: APIs and streaming support integrations and custom product logic. Vault Payments and other connected services should be treated as separate scope unless the proposal says otherwise. | Core platform: APIs support integration and modular deployment. Confirm which components, APIs and integration services are covered by the licence and implementation proposal. | Platform capability: Engine describes an API-based, modular platform with configurable workflows and customer journeys. Confirm any custom development, integration work and ongoing support charges. |
Published UK references | Mambu publishes a UK customer case for Chetwood Bank. Validate that the reference is comparable in product scope and delivery model to your proposed use case. | Thought Machine reports that Shawbrook went live with Vault Core, launching a buy-to-let mortgage product on the platform in 2025. | Temenos reported that Kingdom Bank went live on Temenos SaaS in October 2026; Bank of Beirut UK also reported a Core and Payments go-live in 2026. | Engine powers Starling Bank in the UK. Its published external client examples include Salt Bank in Romania and AMP Bank in Australia; the cited public materials do not establish a separate UK customer implementation. |
Treat these as published platform capabilities, not a guarantee that every feature is included in a specific proposal. Ask each vendor to identify what is standard, separately licensed, provided by a partner or requires custom development, and confirm the implementation and support costs in writing.
Deployment model and buy-or-build are separate decisions. SaaS means the provider operates the service under a managed arrangement; it does not mean the bank transfers its own regulatory responsibilities. Self-hosted cloud puts the platform in a cloud environment selected or controlled by the institution, while on-premise places infrastructure within its own facilities. These choices affect operational control, data and infrastructure responsibilities, upgrade planning and the engineering skills needed to run the service. They do not by themselves determine how much product logic or customer-facing functionality the institution must develop.
Based on public vendor documentation, Mambu and Engine are offered as SaaS platforms; Thought Machine documents SaaS, customer-cloud, hybrid and on-premise options; and Temenos documents SaaS, cloud and on-premise options. Ask each supplier to confirm the actual deployment choices, regions, service responsibilities and commercial terms available for your shortlisted product. A customer-managed deployment may offer more infrastructure control, but the institution takes on more responsibility for operations, security, resilience and upgrades. SaaS can reduce that operational burden and support provider-managed updates, but it increases reliance on the vendor’s service model, release process and contractual exit arrangements.
Buying a core platform also does not remove the need for engineering. Teams still have to configure products, connect the core to channels and third-party services, manage data and test changes. A platform with extensive configuration or extension options can provide room to differentiate, but may require specialist developers and careful release governance. Establish which capabilities are standard, which require additional modules or partner integrations, and which are proposed development. For separately licensed capabilities, confirm the price, implementation effort, support owner and upgrade implications in the proposal.
Building a ledger and core banking service from scratch offers control over architecture and product behaviour, but creates a continuing obligation to maintain reliable account balances, transaction records, controls, reporting and operational processes as products and requirements change. The work continues after launch: engineers must support incidents, test releases, manage security and resilience, and preserve consistent records as the business adds features or enters new markets. For many firms, custom development is more practical around a purchased core—for customer journeys, integrations or specific workflows—than as a replacement for the ledger itself. The choice depends on whether the expected differentiation justifies the long-term cost of owning and staffing that technology.
For firms that need to launch branded accounts, wallets, cards or payment flows without building every component themselves, modular financial infrastructure is another option to assess. Framnex’s solutions include white-label banking, embedded finance, Wallet-as-a-Service and payment orchestration, alongside core banking, ledger and API capabilities. Framnex describes itself as a B2B software and infrastructure provider, with regulated activities performed by applicable licensed third-party partners. Buyers should therefore check which ledger and account services are included, how partner responsibilities are allocated, and whether the model meets their regulatory and data-portability requirements.
Vendor sources: Mambu Documentation Hub, Mambu developer documentation, Thought Machine Vault Core, Thought Machine and Shawbrook, Temenos Core Banking, Temenos APIs, Reliance Bank and Temenos, Engine platform, Engine SaaS service and Engine and Starling.
1.3 Validate Ledger Accuracy, APIs and UK Payment Connectivity
Validate a core banking platform with evidence from end-to-end tests, not a feature list or demonstration using ideal transactions. The ledger, API layer, payment connections and operational processes must work together: an accepted payment instruction should create the expected account and ledger entries, produce accurate customer and finance records, and reach the correct status when processing succeeds, fails or requires repair.
Define ledger acceptance tests
Agree expected results before the proof of concept (PoC), including balances, journal entries, transaction states, audit records and reconciliation outcomes. Run at least the following cases:
- Double-entry bookkeeping: For each completed transaction, verify that debits and credits balance in the correct accounts, currency, value date and accounting period. Check that account balances, control accounts and general ledger totals agree, including any fees, interest, tax or FX entries in scope.
- Ordering and concurrency: Submit transactions against the same account sequentially and concurrently. Confirm the platform applies its documented ordering rules, enforces available-balance and limit controls, and returns consistent balances and transaction statuses to all channels.
- Idempotency and retries: Replay the same request after a timeout or lost response, using the same idempotency key where supported. Verify the retry does not create a duplicate debit, credit or payment instruction, and establish the key's scope and retention period.
- Reversals and corrections: Test declined, returned, cancelled and reversed transactions. Confirm corrections are represented by traceable compensating entries, preserve the original record and reason, and update customer balances and downstream reports correctly.
- Reconciliation and exceptions: Compare ledger entries with payment files, gateway or sponsor reports, settlement records and bank statements. Introduce missing, duplicate, late and mismatched items; confirm they are flagged, assigned, investigated and resolved with an audit trail rather than silently forced to balance.
Exercise APIs, webhooks and data outputs
Test the interfaces the production design will depend on, including account and transaction APIs, authentication and permissions, rate limits, error responses, timeouts, pagination and version changes. Confirm whether responses are synchronous or asynchronous and how clients should determine the authoritative transaction state. For webhooks, simulate slow or unavailable receivers, duplicate and out-of-order notifications, retries and recovery after an outage. Verify that consumers can identify events, avoid duplicate processing and reconcile missed notifications against a durable source.
Run the agreed peak-volume profile, including bursts, concurrent account activity, end-of-day processing and recovery after a backlog. Measure end-to-end latency and throughput against business requirements, and check that degradation or a dependency outage does not create duplicate postings or uncertain payment states. Test reporting exports using realistic volumes and periods: verify completeness, field definitions, timestamps, currency precision, control totals, access permissions and the ability to reproduce a finance or operational reconciliation.
Connect the PoC to representative existing infrastructure, such as customer channels, finance and data platforms, identity services, fraud monitoring and KYC/AML systems. Test both successful and adverse outcomes, including a customer or payment requiring review, an unavailable screening provider, a rejected response and a corrected record. Agree which system owns each decision and record, how exceptions are escalated, and how data is protected in transit and at rest.
Map UK payment connectivity and responsibilities
For each payment type, draw the full path from the customer instruction through the core, payment gateway or service provider, scheme and settlement account to the receiving institution. Identify the party that creates and validates the payment message, submits it, monitors its status, handles returns and exceptions, and reconciles settlement. Confirm whether the core has a native connector, uses a separately supplied module, or depends on a partner or sponsor integration.
- Faster Payments: Establish whether the proposed design uses direct settling participation, direct connection with a sponsor settling on the firm's behalf, or indirect access through a sponsor. These models allocate connection, settlement and operational responsibilities differently. Confirm supported payment types, limits, status reporting, returns and contingency handling against the selected route.
- Bacs: Confirm how Direct Credit and Direct Debit submissions and outputs will be prepared, validated and processed, including service-user, bureau or sponsor arrangements where applicable. Test file acknowledgements, rejects, amendments, returns and the three-processing-day cycle, including non-processing days and the posting of settlement entries.
- CHAPS: Confirm whether the firm will be a direct or indirect participant and who provides the SWIFT connectivity, payment messaging, settlement account and operational monitoring. For direct participation, the Bank of England's published criteria include access to a settlement account and ongoing technical and operational compliance. Test high-value payment authorisation, status, rejection, repair and reconciliation paths.
Technical connectivity is not the same as permission or readiness to use a payment scheme. A platform or gateway may be able to format and transmit messages, but the firm must still establish its eligibility and participation route, complete scheme and sponsor onboarding, meet applicable contractual, operational and technical requirements, and put settlement and exception processes in place. In the UK, Pay.UK describes direct and indirect access routes for Bacs and Faster Payments; CHAPS participation and settlement requirements are set out by the Bank of England. Confirm the firm's own obligations with the relevant scheme operator, sponsor and advisers.
Require proof from a realistic proof of concept
Make PoC acceptance depend on reproducible evidence: test scripts, input data, expected and actual ledger entries, API and webhook logs, payment acknowledgements, settlement records, exports, exception queues and reconciliation results. Include normal transactions, peak-load batches, timeouts, duplicate requests, late or rejected files, unavailable dependencies, returns and recovery after an interruption. Ask the vendor and delivery partner to show who investigates each exception and what operational evidence the firm can access. Record gaps as a specific configuration, licensed capability, partner dependency or proposed development, and do not treat an untested roadmap item as proven functionality.
For current access and settlement requirements, see Pay.UK's Faster Payment participation options, Pay.UK's Bacs direct participation guidance, Pay.UK's access overview, the Bank of England guide for non-bank payment service providers and the Bank of England's CHAPS requirements. For API test design, consult the relevant platform's current API documentation; for example, Mambu's webhook best practices and Mambu's idempotency documentation.
1.4 Estimate Total Cost of Ownership and Implementation Time
Compare core banking proposals over the same three-to-five-year period and against the same business volumes. A low subscription price can give a misleading picture if payment access, partner services, migration, infrastructure or internal support sit outside the quote. Build a GBP model that separates one-off implementation costs from recurring and usage-based costs, and record whether each figure is a supplier quote, a partner estimate or an internal planning assumption.
Set a common modelling basis
The following is an illustrative five-year planning scenario, not a forecast or a vendor benchmark. Replace it with your business plan and written supplier quotes. For a consistent comparison, assume 100,000 active accounts in year one and four financial products. If the 2 million monthly transactions are end-to-end financial transactions, not individual debit and credit ledger entries, and are spread evenly across a 30-day month, average throughput is about 0.8 transactions per second. For illustration, model a peak of 2.3 transactions per second, sustained for 15 minutes. Use the same forecast mix for each vendor: ledger postings, incoming and outgoing payments, reversals, and batch or end-of-day jobs. Model account and balance enquiries separately as API reads, using the expected request rate, and clarify whether API reads count towards each vendor’s pricing meter. Treat this as a planning assumption, not a production capacity requirement, and replace it with forecast peak volumes before final pricing and performance testing. Model account and transaction growth consistently across every shortlisted platform; for example, apply the same annual growth assumption to all options. Record account definitions, transaction types, environments and peak-volume assumptions, since vendors may meter these differently.
Use GBP, state whether amounts include VAT, and apply the same inflation, foreign exchange and discounting assumptions to every option. Ask vendors to price the same term, usage bands, minimum commitments, service levels, environments and support coverage. Where a quote is not available, mark the amount as "TBC" rather than substituting an industry average. As checked on 2 November 2026, vendor materials describe service and pricing approaches but do not provide a directly comparable rate card for these shortlisted core platforms.
Build the three-to-five-year cost model
Use the following cost categories to build the same five-year GBP comparison for each shortlisted platform. Record whether each amount is a supplier quote, partner estimate or internal assumption.
Cost item | What to include | How to estimate and compare |
Core platform licence or subscription | Recurring charges for the core software or SaaS service | Request a dated GBP quote for the same contract term, products, environments and service levels. |
Minimum commitments | Contractual minimum spend, if applicable | Confirm the minimum amount and how it applies if actual usage is below the committed level. |
Usage fees | Charges based on accounts, transactions, products, API calls or other vendor meters | Ask each supplier to define its pricing units and model fees against the same annual volumes and growth assumptions. |
Implementation and configuration | Discovery, platform setup, product configuration, testing and supplier or partner delivery | Request a scoped estimate by phase, with roles, day rates, assumptions, exclusions and change-control rates. |
Integration build and change costs | Development and configuration of connections to customer channels, finance, KYC/AML, data systems and other infrastructure | Separate one-off build work from later changes. Request estimates for each integration and identify who will maintain it. |
Migration and transition from the existing core | Data profiling, remediation, migration rehearsals, opening-balance reconciliation, dual running and archives | Estimate after data discovery. For a replacement, include legacy-data complexity, unresolved exceptions and customer-continuity work. |
Infrastructure and environments | Cloud, hosting, test environments, monitoring, storage and backup where charged separately | Confirm what the SaaS or managed-service fee includes. For customer-managed deployments, estimate infrastructure and operating costs separately. |
Support, maintenance and upgrades | Ongoing vendor support, maintenance, release work and enhanced service coverage | Compare support hours, response targets, included upgrade work and charges for additional support or custom changes. |
Internal staffing and operations | Product, engineering, finance, security, compliance and operational staff needed to run the platform | Estimate planned FTE by role and year using the business’s fully loaded employment cost. Include staff retained after go-live. |
Payment access, gateway, sponsor and settlement fees | Recurring and transaction-based charges from payment gateways, scheme access providers, sponsors and settlement providers | Request separate quotes and identify transaction charges, minimum commitments and the party responsible for each service. Do not include integration build costs in this row. |
BaaS and other non-payment third-party services | BaaS, data providers and other external services not already listed above | Request a breakdown of bundled and pass-through charges, transaction limits, minimum commitments and contracted responsibilities. Avoid counting gateway or sponsor fees twice. |
Exit and data portability | Data extraction, termination assistance, replacement-provider migration and continued access to records | Review the contract for included assistance, time limits and charges; cost a realistic transition plan rather than assuming export is free. |
For each year, add platform charges, implementation and migration, integration work, payment and other third-party services, infrastructure and operations, internal staffing, support and maintenance, and any exit costs planned for that year. Show one-off, fixed recurring and variable charges as separate subtotals. Compare suppliers over the same three-to-five-year period and volume assumptions. A cost-per-active-account or cost-per-transaction measure can help comparison, but should not replace the total because suppliers may include different services in those measures.
Separate core costs from payment access and BaaS
A core platform may maintain accounts and process ledger transactions without providing the complete scheme access, sponsorship, settlement or regulated banking service your proposition requires. Keep platform charges distinct from Faster Payments, Bacs or CHAPS access; sponsor and gateway fees; and any BaaS provider charges. Record which entity contracts with each provider, receives settlement, handles returns and bears operational responsibilities. If a BaaS arrangement bundles several services, request a breakdown of included capabilities, transaction limits, pass-through charges and minimum commitments so the bundled quote can still be compared fairly.
Model greenfield and replacement separately
A greenfield launch avoids conversion of a live book, but still requires product configuration, channels, integrations, data design, testing, operational procedures and payment onboarding. A replacement project also has to map legacy data and business rules, cleanse records, reconcile opening balances, migrate transaction history or establish archive access, and plan dual running and customer continuity. The estimate can move materially with data quality, bespoke legacy customisations, unresolved exceptions, integration complexity, availability of experienced delivery partners and the number of products included in each release.
For an early planning model, label timeline ranges as assumptions rather than supplier commitments. For example, a team could use 9-15 months from mobilisation to technical go-live for a defined greenfield scope and 18-36 months for a complex replacement as provisional planning ranges, then re-estimate after discovery, data profiling and a partner-backed delivery plan. These are illustrative planning assumptions only; they are not published delivery commitments for Mambu, Thought Machine, Temenos or Engine by Starling. Ask each bidder to give a dated, dependency-based schedule covering design, configuration, integration, test cycles, migration rehearsals, operational readiness and cutover, with named customer and supplier responsibilities.
Distinguish technical go-live from the date the business can lawfully launch or expand the service. A system can pass technical acceptance while regulatory permissions, scheme access, sponsor onboarding or other required approvals remain outstanding. Keep those external dependencies visible in the programme plan and identify which party owns each application, evidence pack and operational readiness action. Do not treat a vendor implementation estimate as a regulatory approval timetable.
For current commercial scope, use supplier documentation as a starting point and rely on a dated proposal for actual pricing. Mambu describes usage-scaled pricing on its composable banking page; Engine's 2025 annual report describes SaaS pricing based on account number and size, and service fees for implementation and development; Thought Machine describes the Vault Core platform; and Temenos describes its managed service scope on its SaaS page. These disclosures help identify potential charging components, but they are not comparable customer quotes.
1.5 Assess UK Regulatory Fit, Resilience and Exit Risk
Choose a core banking platform that supports your regulatory obligations and operating model, but do not treat the platform or its provider as a substitute for the firm's own governance. A bank, an authorised payment or e-money institution, and a business distributing services through a regulated partner may have different responsibilities. Identify which entity contracts with customers, holds or receives funds, executes payments, makes regulated decisions and keeps required records. Outsourcing technology or using a partner does not, by itself, transfer the firm's regulatory accountability.
Map responsibilities to the business model
- PRA and FCA-regulated banks: Banks are subject to prudential and conduct regulation. Assess how the core supports the bank's controls, reporting, record keeping and oversight of material outsourcing and third-party dependencies. The bank should be able to understand and evidence how the service operates, including where it relies on a vendor or cloud provider.
- FCA-authorised payment and e-money firms: Determine how the system supports payment-services or e-money records and any applicable safeguarding, reconciliation and reporting duties. FCA rules require relevant firms to keep accurate records that distinguish relevant funds from other funds and to reconcile them as required. Safeguarded funds are not the same as bank deposits: the legal basis, treatment and customer protections differ.
- Businesses distributing through a partner: Document which regulated entity provides each service and which party performs onboarding, customer communications, payments, record keeping, complaints handling and safeguarding, where applicable. Contractual allocation helps define delivery, but does not determine whether your own activities require authorisation or create regulatory obligations. Confirm the perimeter with qualified advisers and the relevant regulator.
Check outsourcing, reporting and operational resilience
For a bank, review the PRA's outsourcing and third-party risk expectations, including due diligence, access and audit rights, business continuity and exit planning. FCA-regulated firms should assess the FCA requirements and guidance applicable to their firm type and arrangement. Until 18 March 2027, firms should continue to meet existing FCA notification requirements for material outsourcing arrangements. From 18 March 2027, new reporting rules will apply to firms in scope, including banks and authorised payment and e-money institutions. These rules require firms to notify the FCA of new or significantly changed material third-party arrangements and to submit an annual register. Confirm whether your firm is in scope and check the applicable reporting timetable.
Where FCA operational resilience rules apply, firms must identify important business services, set impact tolerances, map the people, processes, technology, facilities and information they depend on, and test their ability to remain within tolerances during severe but plausible disruption. Use those requirements to frame platform due diligence. Ask for evidence of recovery objectives, tested failover, backup restoration, incident exercises and remediation of test findings. Compare contractual recovery time objectives (RTOs) and recovery point objectives (RPOs) with the firm's own impact tolerances and service needs; a vendor target alone is not proof that the end-to-end service can meet them.
- Review role-based access, privileged account controls, segregation of duties, multi-factor authentication and timely access removal.
- Check encryption in transit and at rest, key management, security monitoring, audit logs and the ability to retrieve records for investigations and supervisory requests.
- Agree how quickly the provider must notify the firm of incidents, what information it must supply, how evidence is preserved and how the firm can meet its own notification and regulatory reporting duties.
- Map dependencies beyond the core provider, including cloud infrastructure, identity services, payment gateways, scheme sponsors, data processors and implementation partners. Assess concentration risk and any single points of failure.
- Review the scope and results of resilience and failover tests, including customer, ledger, API, payment, reconciliation and support processes, rather than reviewing infrastructure availability alone.
Review data protection and location terms
Under UK GDPR, the firm should establish the provider's role as controller or processor, the permitted processing, security measures, retention and deletion, breach assistance, subprocessors and support for data-subject rights. Ask where data is stored, backed up and accessed, including by support teams and subcontractors. Data location may be a contractual or risk requirement, but a UK hosting location alone does not settle every data protection question. Access or transfers involving recipients outside the UK may engage the international transfer rules. Depending on the route used, the firm may need to rely on UK adequacy regulations, an appropriate safeguard or an exception; where a safeguard is used, consult current ICO guidance on the required data protection test and additional protections.
Do not infer regulatory compliance from a vendor's security certification, audit report or cloud-provider assurance. These can support due diligence, but the firm must determine whether the controls, scope, service configuration and contractual protections meet its own legal and regulatory obligations.
Test safeguarding records and reconciliation
For a payment or e-money firm with safeguarding duties, confirm that the platform and surrounding processes can produce accurate, timely records of relevant funds, distinguish them from the firm's own money and support the required safeguarding calculations and reconciliations. Under CASS 15, safeguarding institutions must perform internal and external safeguarding reconciliations as frequently as necessary and at least once each reconciliation day, subject to the applicable rules and exceptions. The FCA also requires monthly REP027 safeguarding returns from authorised payment institutions, authorised e-money institutions and small e-money institutions, and from small payment institutions that opt to safeguard. Returns are due within 15 business days after month-end. Test exception handling, record corrections, access to underlying transaction evidence and the ability to provide information for audit, resolution and regulatory reporting. Where the firm uses multiple pools or provides different services, ensure the data and reconciliation design can keep the relevant balances distinct. A ledger balance by itself does not demonstrate that funds have been safeguarded correctly.
Make exit rights operational
Review the contract before relying on the platform. Check access and audit rights for the firm, its auditors and regulators; subcontractor disclosure and change controls; data ownership and permitted use; service and incident reporting; and the provider's responsibilities at termination. Define exports in usable formats with data dictionaries, identifiers, timestamps, transaction and journal history, customer records, configuration and audit evidence. Confirm whether the firm can retrieve data directly, how long it remains available, and what assistance and charges apply during transition.
Require an exit plan that names owners, dependencies and decision points, and test it with sample exports or a migration rehearsal. Establish how the business would maintain customer service, reconcile balances, preserve required records and move to a replacement provider if the core, cloud host, payment partner or sponsor became unavailable or the contract ended. A contractual right to export data is useful only if the firm can retrieve, interpret and reconcile it in time to execute a transition.
UK regulatory sources: FCA Handbook SYSC 15A on operational resilience, SYSC 15A mapping requirements, PRA Supervisory Statement SS2/21 on outsourcing and third-party risk management, FCA Handbook CASS 15.8 on safeguarding records and reconciliations and ICO guidance on international transfers.
1.6 Choose a Delivery Partner and Plan Migration
A delivery partner can help select a platform, configure products, build integrations, migrate data and support the service after launch. Assess the named team and its evidence for your specific platform and project, rather than assuming a large consultancy or a vendor partnership guarantees the right expertise. Ask for named practitioners, recent comparable deliveries, client references you can contact, and evidence of who will do the work in the UK time zone. Separate an implemented production system from a partnership announcement, accelerator, staff capability or project that is still under way.
Assess GFT and Accenture against the shortlisted platform
Public information provides starting points for due diligence, not a substitute for project-specific references. GFT lists work with Mambu and Engine by Starling and has a strategic partnership with Thought Machine. Its published work includes the Engine collaboration and Salt Bank delivery with Engine, which is a Romanian example rather than a UK migration. Confirm the exact products, modules, roles and delivery outcomes relevant to your programme. Public information reviewed does not establish equivalent GFT delivery evidence for Temenos, so request it directly if Temenos is shortlisted.
Accenture lists Mambu, Thought Machine and Temenos among its core banking ecosystem partners. Its public references include work with Thought Machine on Arvest Bank in the United States, a Temenos programme for Nordea across the Nordic region, and UK delivery of the Bank of England's RTGS renewal. These demonstrate relevant capabilities, but they are not all evidence of a current UK retail-core implementation on the exact shortlisted platform. Ask for platform-specific UK references and named team members. Accenture's public materials reviewed did not establish a delivery reference with Engine by Starling.
Compare both firms on platform selection, custom development, integration, migration and ongoing support. Establish whether the partner is independent in its recommendations or receives referral fees, implementation incentives or other commercial benefits from a vendor or cloud provider. Ask for a responsibility matrix covering the financial institution, platform provider, implementation partner, payment sponsor and other key suppliers. The institution should retain clear ownership of business decisions, regulatory oversight, acceptance criteria, data quality and go-live approval, even when delivery work is outsourced.
Set delivery gates and evidence
1. Discovery and scope: Agree business outcomes, products, customer journeys, target architecture, regulatory and operational constraints, data sources, volumes, dependencies and exclusions. Baseline the delivery plan, risks, assumptions and acceptance criteria.
2. Proof of concept: Demonstrate the highest-risk product rules, ledger behaviour, APIs, external integrations and operational workflows using representative data. Record unresolved gaps as configuration, licensed modules, partner services or custom development.
3. Configuration and build: Approve product specifications and interface contracts before implementation. Require version control, traceability from requirements to configuration or code, peer review and controlled release practices.
4. Testing: Complete functional, integration, security, performance, operational and user acceptance testing. Include failure, retry, duplicate, reversal, reconciliation and recovery scenarios, not only successful processing.
5. Migration rehearsals: Profile and cleanse source data, map fields and business rules, run repeatable test migrations, investigate rejected records and reconcile account counts, transaction history and financial balances. Obtain finance and business sign-off on opening balances and control totals.
6. Cutover approval: Confirm operational readiness, customer and staff communications, payment and sponsor readiness, support coverage, monitoring, incident ownership, recovery arrangements and a documented go/no-go decision.
7. Controlled launch and stabilisation: Use agreed release waves, transaction or customer cohorts, and heightened monitoring. Compare core balances, payment statuses, exceptions and financial reports against agreed thresholds before expanding the book or retiring legacy processes.
Make migration and rollback controls explicit
Agree which historical records will move to the new core and which will remain in a searchable archive. For every migration, retain source-to-target mappings, validation results, exception decisions and evidence of approval. Reconcile opening balances by account, currency and relevant ledger or safeguarding category, and agree how pending payments, accrued interest, fees, dormant accounts, complaints and legal holds will be treated. Do not accept a balancing total that hides unmatched records or unresolved differences.
Define rollback criteria before cutover, including measurable thresholds for balance discrepancies, transaction failures, payment backlogs, service availability and unresolved high-severity defects. Document who can call a rollback, the last safe point for reversing the cutover, how transactions accepted during the launch window will be reconciled, and how duplicate processing will be prevented if traffic returns to the legacy system. Rehearse the decision and data steps; a backup alone is not a rollback plan.
After launch, monitor availability, latency, posting and payment exceptions, reconciliation breaks, data quality, customer-impacting incidents and support queues against agreed thresholds. Assign each metric an owner, escalation path and reporting cadence. Keep the legacy service and its records available for the agreed transition period, subject to contractual, record-retention and operational requirements.
Use an RFP checklist and weighted scorecard
Before assigning scores, apply pass/fail gates for mandatory product functionality, security, regulatory and data requirements, payment connectivity, resilience, auditability and usable data export. A bidder that fails a mandatory requirement should not compensate with a high score elsewhere. For bidders that pass, use the same evidence request and scoring scale, such as 0 for no evidence, 1 for a material gap, 3 for meets requirement and 5 for strong, independently verifiable evidence. Weighted score equals the criterion score divided by five, multiplied by its weight.
Scored criterion | Weight | Evidence to score |
Functional fit and product configuration | 25% | Demonstrated support for required account types, transaction processing, product rules, reporting and servicing workflows. |
Regulatory, security and resilience fit | 20% | Evidence for required controls, auditability, data protection, operational resilience and support for the firm’s obligations. |
Three-to-five-year total cost of ownership | 15% | Dated GBP proposal covering licence, minimum commitments, usage, implementation, support, third parties and exit costs. |
Delivery and migration confidence | 15% | Dependency-based plan, named team, comparable references, migration rehearsals, testing and cutover approach. |
Scalability and performance | 10% | Evidence against forecast account volumes, transaction peaks, latency and recovery requirements. |
Integration and UK payment connectivity | 10% | Demonstrated fit with existing systems, KYC/AML tools, payment gateways, sponsors and required UK payment routes. |
Exit portability and supplier dependency | 5% | Usable data exports, documented exit assistance, tested portability and visibility of subcontractor dependencies. |
Total | 100% |
- Require a solution design that identifies standard platform capability, separately licensed modules, partner integrations and proposed custom development.
- Request a dated commercial proposal and three-to-five-year GBP cost schedule with assumptions, minimum commitments, usage bands, implementation scope, exclusions, support and exit charges.
- Request a dependency-based delivery plan, named roles and allocation, UK references, risk register, testing approach and migration rehearsal plan.
- Ask for sample ledger and reconciliation evidence, API and failure-handling demonstrations, security and resilience materials, subcontractor lists, data-location details and contract terms covering audit and exit.
- Identify delivery and commercial incentives, change-control rates, assumptions about client staffing, and the partner's responsibilities after go-live.
Public references to verify include GFT and Mambu, GFT and Engine by Starling, GFT and Thought Machine, Accenture's core banking practice, Accenture, Thought Machine and Arvest, Accenture and Temenos with Nordea, and Accenture and the Bank of England's RTGS renewal. These are starting points for validation; request evidence for the exact platform, scope, geography and delivery team proposed for your programme.
Independent research
Ask AI whether Framnex fits your business.
Research Framnex with your preferred AI assistant and see where the platform may fit your client base, product and launch model.
The prompt will be prefilled. Review it before sending.