Running a Core Banking RFP: A 14-Step Vendor Selection Playbook
Most core banking RFPs end up picking the vendor with the best demo team, not the best system. This 14-step playbook builds each stage on evidence: weighted scoring, scripted demos, a sandbox proof of concept and reference calls you choose yourself.
Why core banking selections reward sales polish, and how to stop it
You're picking a system your institution will probably run for 10 to 15 years. Yet a typical core banking RFP goes wrong in the same three ways. The requirements list runs to 1,500 lines and every vendor answers "yes" to all of them. The demos run on the vendor's own data, presented by its best presales engineers. And the reference clients are ones the vendor picked. The shortlist that comes out measures who sells well. It says little about who fits.
The 14 steps below are ordered so each stage removes vendors on evidence. Each step also produces something you can show your board, and later your regulator, to explain the choice. The timeline is indicative for a mid-sized bank, lender or wealth manager. Smaller digital banks often finish in four to six months. Tier-one banks with several countries in scope can take more than a year.
| Phase | Steps | Typical duration | Output |
|---|---|---|---|
| Mandate | 1–3 | 4–8 weeks | Signed charter, target architecture |
| Market scan | 4–5 | 4–6 weeks | Longlist of 8–12 cut to 4–6 |
| Requirements and scoring | 6–9 | 6–8 weeks | RFP issued, scoring model locked |
| Evaluation | 10–11 | 8–12 weeks | Demo scores, PoC results for 2–3 vendors |
| Diligence and decision | 12–14 | 6–10 weeks | Preferred vendor, negotiated contract |
Settle the mandate before any vendor hears from you
Step 1: Write the business case in outcomes, not features
State what has to change and put a number on it: launch a new deposit product in weeks rather than quarters, retire two legacy cores, shorten batch windows, or open a new market. If the business case reads like a feature list, vendors will write the rest of your RFP for you.
Step 2: Assign decision rights
Name a steering committee of no more than seven people from business, technology, risk, compliance, finance and operations. Agree up front who holds the casting vote and who can veto, and limit the veto to security and regulatory knock-outs. A selection where anyone can block anything ends up choosing the least controversial vendor.
Step 3: Choose the target architecture
Decide whether you're doing a full replacement, a progressive "hollow-out" one product line at a time, or a new core alongside the old one for a digital brand. That choice changes the shortlist. Cloud-native platforms such as Thought Machine, Mambu or 10x usually suit progressive or greenfield programs. Broad suites such as Temenos, Oracle FLEXCUBE, Infosys Finacle, TCS BaNCS or Finastra are more often bought as full replacements. Most of these vendors sell both ways, so treat this as a tendency, not a rule.
Build a longlist that reflects your jurisdiction, then cut it fast
Step 4: Longlist 8 to 12 vendors on four filters
- Live clients in your region and segment. A core that runs well for European retail banks may have no localization for Islamic finance in the Gulf, or for Indonesian or Brazilian regulatory reporting.
- Deployment model. Options include SaaS, vendor-hosted, public cloud in your own tenancy, or on-premises. Check each against data residency and outsourcing rules. Indian payment data localization, Gulf regulators' cloud frameworks and EU requirements such as DORA can each rule out particular hosting setups.
- Delivery capacity. Look for local implementation partners who have completed a go-live with this product, not just signed a partnership agreement.
- Financial health. You're taking on decade-long counterparty risk.
Vendors focused on one market, such as Fiserv, FIS and Jack Henry for US community banks, belong on the list only if they actually serve your country.
Step 5: Use a short RFI to cut to four to six
Send a 30- to 50-question RFI covering architecture, regional clients, typical program length, pricing model and implementation approach. Give vendors two to three weeks. Remove anyone who can't name a live client of similar size in a comparable regulatory regime.
Write core banking requirements that separate vendors
Step 6: Replace checklists with scenarios
Checklists fail because vendors can answer "yes" without proving anything. Rewrite requirements as scenarios with a measurable result. Weak: "supports tiered interest." Strong: "configure a savings product with three balance tiers, a promotional rate for 90 days and a penalty on early withdrawal, without code changes; show the accrual and the posting." Keep it to 150 to 300 must-have scenarios and put the rest in an appendix.
Ask vendors to classify every response:
- available today in the standard product
- configuration only
- extension using supported tooling
- custom development
- on the roadmap, with a date
Then make that answer contractual.
Step 7: Specify the nonfunctional and regulatory requirements in full
Write down peak transaction volumes, the end-of-day window, recovery time and recovery point objectives, and API coverage, including ISO 20022 messaging for payments. Add audit trails and local reporting for every country in scope. Include your own regulator's outsourcing and exit-planning expectations. They differ between the EU, the UK, India, Singapore, the UAE and elsewhere, so have compliance confirm the list rather than copying another bank's template.
Lock the core banking evaluation criteria before responses arrive
Step 8: Weight the criteria and publish the method internally
Fix the weights before the first proposal arrives. Otherwise scores drift toward whichever vendor the committee already likes. Here is a hypothetical starting point for a mid-sized bank doing a full replacement:
| Criterion | Weight | Evidence source |
|---|---|---|
| Functional fit (scenarios) | 25% | RFP responses, scripted demo |
| Architecture, APIs, scalability | 20% | Technical workshop, PoC |
| Regulatory and security fit | 15% | Compliance review, certifications |
| Implementation approach and partner | 15% | Plan review, references |
| Total cost of ownership (10 years) | 15% | Commercial model |
| Vendor viability and roadmap | 10% | Financials, references |
A greenfield digital bank would probably shift weight from functional breadth to architecture and speed of product configuration.
Step 9: Define the scale and the knock-outs
Use a 0 to 5 scale with written definitions. For example, 5 means shown working in the standard product, and 2 means it needs custom development. Scorers mark independently before any group discussion. Knock-out criteria sit outside the weighting: a vendor that fails data residency or a mandatory security control is out, however good its other scores are.
Run scripted demos and a sandbox proof of concept on your terms
Step 10: Script the demos
Send scripts two to three weeks ahead, built from your highest-weighted scenarios and populated with anonymized sample data. Every vendor gets the same script, order and time limit, usually one to two days. Score each scenario live. Red flags:
- switching to slides when a scenario gets hard
- "we'd normally show that in phase two"
- an environment that is obviously staged for the demo
Step 11: Put two or three finalists through a core banking proof of concept
A real proof of concept means your people working in a sandbox. Their tasks:
- configure a product
- run interest accrual and an end-of-day cycle
- call the APIs from your digital channel
- load a sample of migrated data
Watching vendor staff do these things tells you little. What matters is what your team can do with the platform. Plan four to eight weeks. Paid PoCs are common; typical 2026 market ranges run from roughly $50,000 to $300,000 per vendor depending on scope, and some vendors credit the fee against the contract. Confirm terms with each vendor. Measure two things: how long your staff take to configure each scenario, and how many times they needed vendor help.
Make reference calls and commercial diligence do real work
Step 12: Choose your own references
Ask for the full client list in your region and pick from it yourself. Include at least one client that is mid-implementation and one that went live in the last two years. Questions to ask:
- How far did the go-live slip, and why?
- What did the vendor's contracted team get wrong?
- What does an upgrade cost you now?
- Would you choose this vendor again?
Informal conversations through industry associations often get more candid answers than a call the vendor arranges.
Step 13: Model ten-year total cost of ownership
License or subscription fees are often a minority of the total. The main cost drivers:
- Pricing basis. Per account, per customer, or subscription tiers. Model your growth plan, because per-account pricing can penalize a successful product launch.
- Maintenance or subscription uplift. On-premises maintenance is commonly a meaningful annual percentage of license value. Get indexation caps in writing.
- Implementation services. These can match or exceed the software cost over the program.
- Hosting, data migration, testing, parallel running and internal staff.
Typical 2026 market ranges vary widely by contract and region. Full replacement programs at mid-sized institutions often run into eight figures in USD, while SaaS cores for smaller digital banks may start at several hundred thousand dollars a year. Use these only as a sanity check and rely on vendor quotes.
Negotiate exit rights first, then choose
Step 14: Negotiate with two finalists, then decide
Keep two vendors in play until the main terms are agreed. Your leverage drops sharply once you announce a preferred bidder. Put these points first:
- Exit and data portability. Full data extraction in a documented format, transition support, and source code escrow or continuity provisions for on-premises deals. Many regulators expect a tested exit plan for critical outsourcing.
- Fixed or capped implementation fees tied to milestones that cover working functionality, not documents delivered.
- Contractual status for the RFP answers. Every "standard" and "configuration" answer becomes a warranty.
- Service levels with real credits for availability, incident response and end-of-day completion, plus audit and access rights for you and your regulator.
- Named key personnel from the vendor and its implementation partner.
Have external counsel with financial services outsourcing experience in each jurisdiction review the contract. Requirements differ enough between regimes that one template won't protect you everywhere.
The decision rule: pick the highest weighted score only if that vendor cleared every knock-out and your own staff completed the PoC scenarios with minimal vendor help. If the top scorer leaned on vendor engineers to get through the PoC, treat its margin as unproven and extend the PoC by two weeks before you sign.
Questions buyers ask
How long does a core banking RFP usually take?
A mid-sized bank or lender typically needs seven to twelve months from mandate to signed contract. Smaller digital banks with a narrow product set can finish in four to six months, and large multi-country institutions often take more than a year. Skipping the proof of concept is the most common way teams shorten the timeline, and usually the most costly.
How many vendors should be on a core banking shortlist?
Longlist 8 to 12, then use a short RFI to cut to four to six for the full RFP. Take two or three to scripted demos and a sandbox proof of concept. With more than three in the PoC, your team is spread too thin to test any platform properly.
Should we pay for a core banking proof of concept?
Usually yes. Paying gets you a properly provisioned sandbox, real vendor engineers and a scope you control. Typical 2026 market ranges are roughly $50,000 to $300,000 per vendor depending on scope. Ask whether the fee is credited against the contract, and confirm terms with each vendor.
Do we need a consultant to run vendor selection?
Not necessarily, but an independent advisor helps if your team hasn't replaced a core before. Advisors bring pricing benchmarks and scenario libraries. Check whether the firm has implementation partnerships with any shortlisted vendor, and get conflicts disclosed in writing.
What are the biggest red flags during core banking evaluation?
Vendors who switch to slides when a scripted scenario gets difficult, refuse to let your staff configure products in the sandbox, or offer only hand-picked references. Vague answers on upgrade costs and data extraction at exit are also warning signs, as is a requirement marked "standard" that turns out to need custom development once you test it.
How should regulatory requirements be handled across different countries?
List each jurisdiction's data residency, outsourcing, cloud and reporting rules as separate requirements, and treat the critical ones as knock-outs. EU, UK, Indian, Singaporean and Gulf regulators each have their own expectations for third-party risk and exit planning. Have your compliance team and licensed local counsel confirm the requirements for every country in scope.
Written by the Apexnode Tech editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.