It is a good story. But the more interesting story sits about nine thousand kilometres south of Dublin, and it has been quietly running for thirteen years.
If we want to understand what good looks like in retail banking, and more importantly what good looks like when software is built around people rather than the other way around, the answer is not in London or Berlin or San Francisco. It is in São Paulo.
Nubank now serves 131 million customers across Brazil, Mexico and Colombia. To put that in proportion, it is the largest private financial institution in Brazil by customer count, serving roughly 61 percent of the country's adult population.
Some numbers. Quarterly revenue runs at over four billion dollars. Net income for Q3 2025 alone was 783 million dollars - with annualised return on equity of 31 percent. Eighty to ninety percent of customer acquisition is organic word of mouth. The cost to serve a customer has actually fallen, from around 90 cents to 80 cents, even as the customer base more than doubled. Yes, some numbers.
These numbers are exceptional on their own. They become more extraordinary when you see what they replaced. Brazilian retail banking in 2013 was an oligopoly of five institutions, charging some of the highest interest rates and fees in the world, requiring physical branch visits and paper forms to open even a credit card account. Cristina Junqueira, who co-founded Nubank after five years inside one of those incumbents, has been blunt about the founding emotion: outrage at paying the highest fees in the world for the worst customer experience in the world.
What is striking is not just that Nubank won. It is how completely it has redefined what banking customers will tolerate. The company has held a Net Promoter Score (NPS) around 90 for five consecutive years, and won Brazil's Reclame Aqui customer service award nine years running. Most enterprise software cannot do that for a quarter.
There is a habit in Irish & European tech to treat innovation as something that happens in a corridor between San Francisco, London (and, on a generous day, Stockholm or Berlin). When we benchmark a fintech, we benchmark against Monzo, Starling, Revolut. When we look for a design language, we look at Apple, Stripe, Linear. When we talk about platform engineering, we cite the FAANGs.
That habit is increasingly expensive. The most aggressive product innovation in financial services over the last decade has not happened in a Eurocentric market. It has happened in places where the incumbents were so bad, and the regulatory and demographic pressure so acute, that there was no choice but to rebuild from first principles. Brazil. Indonesia. Kenya, with M-Pesa. India, with UPI and the broader India Stack. The Philippines. The constraint was the catalyst.
Monzo's pitch in Ireland this month is essentially a polite version of this same insight applied locally. Michael Carney, Monzo's EU CEO, made the point cleanly at launch: "I don't think many people today say that they would love their banking experience and I think that's our difference." But the truth Monzo will not say out loud is that the gap between what Irish customers tolerate and what they could have is not a gap of single percentage points. It is a chasm. Nubank has spent thirteen years showing what is on the other side of that chasm.
It is tempting to read the Nubank story as a design story. Beautiful purple card. Clean app. Lovely typography. NuDS, the internal design system, runs to over three hundred thousand lines of code and a hundred-plus reusable components, which means every product launches with the same visual grammar a customer already knows.
But the design language is downstream of something more important. Nubank's product philosophy is that the customer should never be asked to absorb the bank's complexity. Every regulatory requirement, every legacy oddity, every internal team boundary, is treated as the bank's problem to absorb on the customer's behalf - not to surface as an additional field on a form.
This is harder than it sounds, especially in a regulated sector. The default position of regulated software is defensive. The customer is asked to acknowledge, accept, confirm and re-confirm, because every screen is doing double duty as evidence in a future compliance review. The result is software that is technically correct but humanly… hostile.
Nubank's contribution is the demonstration that it does not have to be this way. You can be Central Bank-regulated, you can serve over a hundred million customers, you can run a 30 billion dollar credit portfolio, and you can still answer 94 percent of support calls in under 45 seconds, in conversational language, without making the customer feel small.
Monzo can do well in Ireland. The waitlist is real, the product is good, the timing is right, and 89 percent of Irish consumers are with AIB, BOI or PTSB (full disclosure: a Marino Software customer). Which is to say the addressable market of mildly-frustrated current account holders is large. The 1.6 percent variable savings rate is a useful wedge. The free business accounts are smart. Monzo is one of the best executed digital banks in Europe.
But Monzo at its current scale, around fifteen million customers in the UK, is roughly an eighth the size of Nubank. And while Monzo's NPS is good for a UK bank, it does not approach the sustained 90 that Nubank has held. The reason is not that Monzo's designers are less talented. It is that the cultural and regulatory water Monzo swims in is the same water as the incumbents it is competing against. The incremental improvement available is real but bounded. Nubank had to build an entire alternative grammar of banking because the existing one was not improvable.
The opportunity for any organisation watching Monzo's Irish launch is to look past Monzo and ask the harder question. What would our product look like if we treated the customer's time and attention as the most valuable resource in the system? Not as a UX nicety, but as the actual organising principle.
At Marino, software made human is how we think about our own product work, particularly as we move deeper into AI-native delivery and engineering.
The phrase is easy to misread as a design slogan. It is not. It is a claim about where the burden of complexity should sit. Software made human means the system absorbs the complexity so the person does not have to. It means regulatory requirements get translated into experiences a customer can navigate without a phone call. It means AI capability is deployed to remove friction, not to add a chatbot layer over the same broken process. It means the test of a product is not whether it is technically correct, but whether the human on the other end of it ends the interaction more capable, more confident and less anxious than they started.
This is what Nubank has been doing, in a regulated sector, at scale, for thirteen years. It is what the best version of Monzo is reaching for in Ireland. And it is the standard that any of us building software for regulated industries, financial services, healthcare, the public sector, should be measuring ourselves against.
The lesson from São Paulo is not that we should copy Nubank. It is that the ceiling is much higher than we have been pretending. The customers know it. They have been telling us for years, in churn data, in NPS scores, in the queue of a hundred thousand people who signed up for a Monzo account before the bank had even opened its Irish doors.
The question is whether the rest of us are going to keep designing for the ceiling we inherited, or for the one Nubank has already proven is possible.
Monzo's Irish launch
Nubank: scale, performance and customer experience
NuDS (Nubank Design System)
Have a project in mind, or just starting to think one through? We’re good at both.
Get in touch
