The Quiet Craft of Picking Payment Rails

5 MINS

In every payments roadmap I have worked on, there is a moment where someone — usually a stakeholder who has just been to a fintech conference — asks, "why don't we use rail X?" And the team has to explain, again, that rails are not interchangeable. They are choices with consequences that last for years.

This is a short field note on how I think about that choice.

Rails are not just APIs

A payment rail is a bundle of an API, a settlement model, a regulator, a counterparty network, an FX policy, and a dispute process. The API is the smallest part. When you sign up to a rail, you are signing up to all of it.

That is why "let's just integrate rail X" is rarely the right way to frame a discussion. The right framing is closer to: what does this rail let us promise our customers, and what does it force us to do operationally for the next decade?

The four questions I ask before adding a rail

Whenever a new rail goes on the table, I run it through the same checklist:

What is the worst settlement story? Not the average — the day-7-of-the-quarter, currency-controls-changed, counterparty-bank-down kind of day.

Who picks up the phone? If a payment is stuck, is there a real human at a real desk who will work with us, or is it a contact form in a different time zone?

What does the cost curve look like at 10x volume? Pricing that works at pilot scale often doesn't survive growth.

Can we exit? A rail you can't decommission cleanly is a strategic liability. You will notice none of these are "is the API nice?" The API matters, but only after the other four are acceptable.

Why FX makes this even harder

In domestic payments, you can sometimes brute-force your way through a bad rail with operational effort. In FX, you cannot. The exposure is too dynamic. A bad FX rail leaks money in places that don't show up cleanly on a P&L until quarter end — and then they show up loudly.

I learned this the hard way scoping cross-border platforms. The most expensive lesson: a rail with a bad pricing feed will quietly destroy your margin while every dashboard tells you everything is fine. The fix isn't a better dashboard. The fix is to not pick that rail.

A short note to product peers

If you are a PO or PM owning a payments roadmap, my unsolicited advice:

Spend a day with the operations team that handles exceptions. It will change your prioritisation.

Ask treasury what their nightmare is. Build for that.

Don't let "modern" be a tiebreaker. "Modern" is a marketing word; "predictable" is a product word. Payment rails are one of those product decisions where the small, careful, slightly boring choice is almost always the right one. The exciting choice usually shows up later as a postmortem.

Scrum, the Way a Banker Still Loves It

Open Banking Is Mostly Plumbing — and That's the Point

Paul skipped presentations and built real AI products.

Paul K Paul was part of the March 2026 cohort at Curious PM, alongside 17 other talented participants.