Buy or configure the parts of digital banking that every bank needs, and build only what makes your bank different. For most banks that means a hybrid: a packaged digital banking platform for the plumbing, with the differentiating layer built on top by your own team or a partner. A full in-house build makes sense only when the channel is your strategy and you already run a product engineering organisation; a pure off-the-shelf purchase makes sense when speed matters more than difference.
The label matters less than what the choice commits you to: a team, a timeline, a risk profile, a degree of control and an exit. This guide sets out those commitments for each option, the factors that should decide between them, how each one fails, and the questions to settle before you sign anything.
What does each option really commit the bank to?
Building in-house, or bespoke with an engineering partner, commits the bank to being a software company for this product. You fund a standing team (product, design, mobile, back end, security, quality, operations) for the life of the channel, not for the project. You own the roadmap and the intellectual property, and you carry the delivery risk and every security patch. Exit is easy on paper, because the code is yours, and hard in practice if the knowledge sits with a few people.
Buying a packaged platform commits the bank to a vendor’s product and release cycle. You get working capability early and a shared cost of maintenance, and you accept that changes outside the configuration surface wait for the vendor’s roadmap or become paid customisations. Your team shrinks to configuration, integration and vendor management. Exit means a migration, so its cost is set by how much of your data, journeys and integrations can leave with you.
Hybrid means licensing a platform for the undifferentiated parts (authentication, payments plumbing, card controls, audit, notifications) and building the journeys and propositions that carry your brand on top of it, through documented extension points and ideally with source-code access. It commits you to two disciplines at once: managing a vendor and running a smaller engineering team. Done well, it gives most of the control of a build without most of the plumbing; done badly, it leaves you with a fork nobody can upgrade.
Which factors should decide between build, buy and hybrid?
Six factors carry the decision. Score each one honestly before you look at any product.
- How differentiating the channel is. If your app must behave unlike your competitors’ apps, weight control; if parity is enough, weight speed.
- In-house engineering capacity. Count the people you can keep for years, not the people you could hire for the launch.
- Time pressure. A licence condition, a competitor or a core programme may fix the date before you choose the method.
- Core constraints. What your core and middleware can expose in real time limits all three options equally; no sourcing choice fixes a slow core.
- Regulatory burden. Outsourcing, security and resilience obligations apply whichever way you source, as set out below.
- Long-run cost shape. Build is payroll-heavy and permanent, buy is licence-heavy and recurring, and hybrid is both in smaller amounts.
How do build, buy and hybrid compare?
The table summarises the three options across those factors. Treat it as a rule of thumb, not as data: a strong team or a weak product moves every cell.
| Factor | Build | Buy | Hybrid |
|---|---|---|---|
| Time to first release | Slowest | Fastest | Fast, then iterative |
| Standing team | Full product team | Configuration and vendor management | Smaller engineering team |
| Roadmap control | Full | Vendor-led | Shared: yours on top |
| Delivery risk | Carried by the bank | Mostly the vendor’s | Split by layer |
| Differentiation | Unlimited | Within configuration | High where it matters |
| Upgrade path | Yours to fund | Vendor releases | Vendor releases, if not forked |
| Exit | Own code, key-person risk | Migration | Keep your layer, replace the base |
| Cost shape | Payroll, permanent | Licence, recurring | Licence plus smaller payroll |
| Regulatory work | Security and resilience in-house | Outsourcing and vendor oversight | Both |
What regulation applies whichever option you choose?
No sourcing option moves accountability out of the bank. The Basel Committee’s Principles for Operational Resilience (2021) ask banks to map the interconnections and interdependencies behind their critical operations, and to manage their dependencies on third parties and intragroup entities, with due diligence before entering an arrangement and exit strategies that assess substitutability, including bringing a service back in-house. The same principles ask for resilient ICT, including cyber security, that is regularly tested, which applies to an in-house build exactly as it does to a purchase.
Regulators turn that into rules. The EBA’s 2019 Guidelines on outsourcing arrangements state that outsourcing cannot result in the delegation of the management body’s responsibilities, and call for a documented exit strategy when critical or important functions are outsourced; the EBA has since published successor guidelines on third-party risk that will repeal them once applicable. In Saudi Arabia, SAMA’s Rules on Outsourcing leave ultimate responsibility for the outsourcing policy with the bank’s board. In the UAE, the CBUAE announced in June 2021 an Outsourcing Regulation that requires board-approved outsourcing policies and a notice of non-objection before a bank outsources any material activity.
Whether your arrangement counts as outsourcing depends on who operates the software, not on what you call the project. The EBA’s final report notes that purchases of goods, including software licences, are not considered outsourcing arrangements, and a platform that a vendor hosts and runs for you is a different matter. Confirm the classification with your own regulator early, because it changes the contract. On secure development, NIST’s Secure Software Development Framework (SP 800-218) works in both directions: it is a practice set to hold your own team to if you build, and NIST notes that software purchasers can use it to communicate with suppliers during acquisition.
Which bank profile suits each option?
Build suits a bank or fintech whose digital channel is the business, that already employs product and engineering leadership, and that can fund the team for years. Digital-only challengers with an unusual proposition often sit here.
Buy suits a bank that needs credible mobile and internet banking quickly, competes on rates, relationships or network more than on app experience, and has a small technology team. It also suits a first release under a hard deadline, provided the contract leaves a route to extend later.
Hybrid suits the broad middle: established banks with a capable but finite technology team, several segments to serve, a core they are not replacing soon, and a need to look distinct in a market where competitors license similar products.
How does each option fail?
Each option has a characteristic way of going wrong, and each is visible in advance in the contract and the architecture.
- Build: the programme stalls on undifferentiated plumbing. Months go into authentication, session handling, limits, audit trails and app-store releases before the first distinctive feature ships, and sponsors lose patience.
- Build: the team that wrote it leaves. Without documentation and succession, owned code becomes code nobody dares to change.
- Buy: the platform locks the roadmap. Every request outside configuration joins the vendor’s queue, and the bank’s app converges on its competitors’ apps.
- Buy: customisation without limits. Enough paid change requests turn a product into a one-off that no longer receives clean upgrades.
- Hybrid: the seam is not respected. Teams modify the base product instead of extending it, the fork diverges, and upgrades stop.
- Hybrid: nobody owns the whole. Vendor and bank each assume the other is testing the integration between the layers.
What should you settle before deciding?
Put these questions to every candidate, including your own engineering team, and write the answers into the contract or the internal mandate as requirements, not assurances.
- Intellectual property: who owns the code, designs and configuration created for the bank, and what licence do you hold over the base product?
- Source code: do you receive it and can you build and deploy from it without the vendor, or is it held in escrow, and under what release conditions?
- Who can change what: which changes can business teams make by configuration, which need your developers, and which need the vendor?
- Upgrade path: how often are releases issued, how long is a version supported, and are your extensions tested against each release?
- Exit: in what formats can data, journeys and rules be exported, and what transition help is contracted?
- Security evidence: what secure development practices, testing and vulnerability handling can the supplier, or your own team, demonstrate?
- Core integration: who builds, owns and maintains the connectors to your core and middleware?
How do you make “we can change it later” testable?
Most build versus buy regret comes down to one promise: that the software can be changed or replaced later. ISO/IEC 25010, the international product quality model, gives the vocabulary to test that promise. It names maintainability and flexibility among its quality characteristics, which gives you the terms to ask how modular, modifiable and replaceable the software really is.
Turn each into an acceptance criterion. For modifiability, name a change your business expects (a new onboarding step, a new fee rule) and ask who makes it and how it reaches production. For replaceability, ask what it would take to swap one component, such as the authentication service, without touching the others.
How Eurisko fits
Eurisko offers the hybrid route and the build route. Eurisko DXP is a licensed, versioned digital banking platform that deploys inside the bank’s own estate and runs in front of any core; your teams change journeys, products and rules through its Journey Builder, and what they create on the platform belongs to the bank and can be exported in open formats. Around it, our engineers (more than 200, delivering banking software since 2009) build the differentiating layer or a fully bespoke product, with an optional build-operate-transfer track for banks that want to take the work in-house. Licence, source-code and escrow terms are part of that conversation, and we prefer to have it early.
We are not the right choice if you intend to own and staff the entire stack yourself, with no licensed product and no partner in it, or if the channel module bundled with your core already matches your ambition.