How do banks run an RFP (request for proposal)?

How does a bank pick a new system? Usually it comes from a build or buy decision, and if it is buy, it goes through an RFP (request for proposal). From the outside it looks like bureaucracy, yes kinda, but I have been the person in the room for these, and the shape is almost the same every time. There are 5 stages you need to know

Who makes the decision

A bank RFP is never one team's decision. The business owner wants the feature. Procurement runs the process and the commercials. Risk, compliance and security bring their own checklists, because a vendor that stores or processes customer data is a risk the bank has to explain to auditors and the regulator later.

My seat is around the technical decision. Does it fit our architecture? How does it integrate with the core banking? How does it integrate with our API gateway, data lake and warehouse? Can it handle our volume? Can we run it ourselves? Can it be customized easily if our business changes? ...

Stage 1: Requirements

Write the requirements before you talk to any vendor. Split them into must-have and nice-to-have, and agree on the scoring weights upfront. Think of it like a PRD with some breakdown of feature detail. From a technical point of view, this is where the non-functional requirements go in: volume, availability, integration, where the data is stored.

Why so strict about the order? Because a vendor demo is a sales performance. The presenter does it all the time. A bank team sees one maybe once a year. Nobody walks out of a good demo neutral. If you write the scoring sheet after the demo, it will somehow fit the favorite, and everyone will honestly believe they came up with it on their own.

Stage 2: Longlist and then shortlist

Scan the market, send an RFI (request for information) to all potential vendors, then judge them on 3 questions: can it do the job, will the vendor still be around, and what do the references say.

Then shortlist to 3-5 real candidates. More than that just burns evaluation time. Every serious candidate costs your people weeks of workshops, so no more than 5.

Stage 3: The proposal

All vendors answer the same structured document, via an Excel sheet or a procurement system, so you can compare them answer by answer. This is where the famous scoring matrix comes from, and everyone pretends the matrix picks the winner.

The vendors wrote the answers. They answer a questionnaire the way a student answers an exam: to get the mark. "Yes, configurable" is the most expensive sentence in enterprise software. When I review a proposal, that is the line I push on. Configurable by whom? With code or with settings? Does it survive the next upgrade?

A matrix full of green cells tells you they may be a fit. The PoC tells you if they are.

Stage 4: The evidence

This is where it actually gets decided. Shortlisted systems, your data, your scenarios, and pass/fail thresholds fixed before the PoC (proof of concept) starts. A demo shows what the product can do. Your PoC shows what it will do for your bank.

A good PoC needs 2 things:

  • Ugliest cases. The old file format from the core, the end-of-day batch, the month-end volume, the weird local product nobody else has. A PoC on the vendor's happy path is just a longer demo.
  • Thresholds written down before anything runs.

Stage 5: Negotiation and reference checks

Price, support SLAs, exit terms, and real conversations with other banks that bought the thing 2-3 years ago. Those folks have opinions by now. Ask them. Talk about the exit early, because signing day is the last day you have any leverage. Data export format, help with the transition, escrow: negotiate the divorce while everyone is still in love.

That's the whole RFP process for me. The key thing is still knowing what you want before you reach out to vendors, then letting the PoC on your own data decide. The rest is just logistics.