Field notes · Bank transformation

Why Banks Lose, and It Isn't the Tech

The mainframe is not the moat. The governance is. Everything else is downstream of who is allowed to say yes.

Topic · incumbents vs fintechs synthesized from years of notes Read · ~6 min Stance · culture is the bottleneck, code is the alibi

What I've come to believe

Banks do not lose to fintechs because of legacy code. They lose because they are organized to prevent failure in a market that rewards reducing the cost of it. The technology is downstream of the culture, and the culture is the one thing no vendor can sell you.

The incumbent struggle is almost always framed as a tech-debt problem: aging mainframes, decades of tangled code, a stack too brittle to change. That framing is comforting because it implies the fix is a budget line. After years spent around risk and transformation, I read it differently. The legacy stack is real, but it is mostly a costume worn by an organizational-design problem. The thing actually holding banks in place is how they decide, who gets to decide, and what happens to the person who decides wrong.

01The through-line

Run across the sources and one number keeps surfacing: for roughly every dollar a bank spends on new technology, a large multiple of it goes not into the new thing but into tying that new thing back to the old systems it has to coexist with. The result is a transformation budget that looks enormous and moves almost nothing.

Set that against the speed gap. Challenger shops deploy software constantly, many times a day, in small reversible pieces. A traditional bank ships in large, rare, high-ceremony events, sometimes once a quarter. Same industry, two completely different metabolisms.

A nine-month release cycle does not just move slowly. It ships into a market that has already changed, which means the bank is forced to be certain about ideas that are stale on arrival.

Most of the money never reaches the future, and the slow cadence guarantees it arrives late. Figures are paraphrased from the notes and are illustrative.

02Three reads

The bottleneck is governance, not code. Change boards, sign-off chains, and a culture where "no" is the safe career move cost a bank more speed than any mainframe does. You can re-platform the technology and still ship once a quarter if the approval path is untouched. The org chart is the architecture that matters.
Speed compounds while certainty decays. Small, frequent, reversible releases learn faster and fail cheaper, so the gap between a challenger and an incumbent widens every quarter even if they started level. The incumbent is not just slower, it is paying more to be wrong later.
Power sits with the wrong people. Where lawyers and program managers outrank engineers and product, the institution optimizes for not getting blamed, which is the precise opposite of building. Fintechs put builders in charge and accept that a mistake is a bad afternoon. Banks put gatekeepers in charge and turn a mistake into a bad six months, so everyone learns to avoid the attempt.

03How I read it

My take

Here is the part the battlefield framing misses, and it is the part I care about most. The transformation energy almost always goes to the shiny front door: onboarding, instant payments, the slick embedded experience, the things that demo well in a board meeting. The unglamorous back-end journeys get left in the legacy estate, and those are exactly the journeys that decide whether the system serves a person on a bad month: hardship, collections, disputes, re-aging, the call where someone says they cannot pay.

Those journeys never get the thin-slice modernization treatment, because they do not photograph well and nobody is promoted for them. So they stay where the spaghetti and the "no" culture both still live. That is my contrarian test for whether a bank has actually transformed: do not open its app and admire the account opening. Call it and say you have lost your job and cannot make this month's payment. The distance between those two experiences is the real measure of the transformation, and at most institutions it is enormous.

The mainframe is not the moat. Neither is the app. The moat is whether the institution can extend its best engineering to its least glamorous customer, and that is a governance choice, not a technology one.

04Where this is going

Banks have the balance sheets, the licenses, the customers, and the trust. On paper they should win. They lose only by declining to dismantle the governance built for an analog world of single mainframes and quarterly releases. The institutions that break through will not be the ones that spent the most on cloud. They will be the ones that moved power to the people who build, and that chose, deliberately, to point that capability at the hardship path nobody else wanted to touch.

Threads worth pulling: the "Decoding: Banks" series (11:FS), especially the episodes on software and legacy, the fintech insurgency, the "Red Queen" speed problem, and the transformation playbook. The spend ratio and cadence figures are paraphrased from those notes and are illustrative.