让用户狂热到主动安利：Nubank 如何把口碑做成增长引擎

Host Lenny Rachitsky opens with striking facts: Nubank is bigger than Coinbase, Robinhood, Affirm, SoFi and Lemonade combined, 80% to 90% of its growth comes through word of mouth, and it has more customers than Bank of America despite operating in only three Latin American countries. Duggal, who joined about five years ago, says much of what he describes predates him, but it is part of what drew him in.

The first ingredient is tapping a very deep pain point. Brazil's large incumbent banks were among the most profitable and among the most hated in the world. Duggal recalls a client in São Paulo in 1997 pointing at his bank branch and saying he hated them, a line that stayed with him for twenty years.

The second is a culture that is alive in employees' minds. He guesses 80% to 90% of staff could recite the five values, the first being "we want our customers to love us fanatically." During his interview he tried to test whether that was real or just a sign on the wall, and he says that when hard decisions arise Nubank biases in favor of the customer's love.

Only then does the tradecraft matter: customer discovery before building, innovating on customers' behalf by observing pain points they no longer notice, and measuring honestly whether they love the product. The other values are being hungry and challenging the status quo, building strong and diverse teams, pursuing smart efficiency, and thinking and acting like owners, not renters.

Duggal admits Nubank does not always live up to its principles, but it borrows techniques widely. One is the Amazon-style mock press release: before any engineer is assigned, the team must explain to the intended customer in two paragraphs why they should care. Product and design reviews several times a week ask why a feature is fundamentally different on quality, complexity and price, ideally all three at once.

After launch, the team relies on the Sean Ellis score, which asks customers how disappointed they would be if the product went away. Ellis's threshold for product-market fit is 40% answering "very disappointed," but Nubank raises it to 50% because Brazilians are inherently polite and optimistic. Nubank rarely scales a product before it hits the threshold, and the question is baked into post-launch reviews rather than written down as a hard rule.

At company level Nubank is "addicted" to NPS, with scores that skew into the 70s, 80s and occasionally 90s, and in Mexico it recorded 94-95 in the first couple of years. A dip of one or two points is cause for alarm. It also tracks churn, since a leaky bucket undermines the viral acquisition loop and forces heavy marketing spend.

Duggal recounts a phrase from their lead designer that he and the co-founders adopted: "good enough isn't good enough; is it great enough?" He contrasts this with companies that pour venture money into Google and Facebook marketing, arguing it is better to invest in making the product so good that marketing becomes a small portion of spend.

Duggal's example is the payments assistant, a Brazil-only product that helps customers manage bills across four payment rails, including the PIX instant payments system. Unreliable rails and the risk of a negative credit-bureau mark made it a real pain point. At launch the Sean Ellis score was borderline 40%, but customers with four or more bills on at least two rails scored 70%.

The team made it easy to onboard multiple bills across multiple rails, letting automation deliver the benefit, and iterated dozens of times to grow that cohort. Duggal says what looks like unexciting bill autopay becomes a fundamentally different experience when built this way. The product went from struggling in the hundreds of thousands about 15 months earlier to over 10 million monthly actives. A similar cohort analysis shaped the high-income Ultravioleta rewards card.

To get feedback fast, Nubank tells product managers and designers not to wait weeks for a thousand-person survey with cross tabs. Instead they should literally pick up the phone and call ten bullseye customers, without routing the work through researchers. By the fifth call you can usually predict what customers six through ten will say, and tone of voice reveals nuance statistics cannot.

Rachitsky adds Jeff Bezos's line that when data and anecdote disagree, the anecdote is usually right, and Duggal agrees. Nubank uses various survey vendors but does not standardize on any, since the mindset matters more than the tool.

Duggal describes the intense pressure on a line product manager to scale something the whole executive team has been counting on, and says the PM's job is often to say no. He frequently tells Nubank staff that they will not take a small problem and scale it, because that produces a big mess. If it is not working at small scale, iterate until it does, and scaling often takes care of itself once customers are excitedly telling friends.

Big companies can make almost any product look good for months or years because their huge customer base will try it, he says, but that is "bouncing a dead cat." The Sean Ellis score brings some science to what is otherwise an art-like judgment call. Nubank's Ultravioleta card launched on July 4, 2021 and began scaling two to two and a half years later, after the team found where it resonated.

On how PMs can push back, Duggal credits culture. As owners, not renters, team members are expected to deliver bad news early, even when it is inconvenient or a promotion is at stake. Senior leaders who have scaled startups usually respect clarity, data and a high bar, and strong-willed executives with opinions are part of the job description.

He adds a pragmatic argument: if you believe a product is not ready, you will still own the mess later and everyone will forget they told you to scale, so you might as well pivot earlier when it is easier.

Duggal calls user research the area where he still wrestles most, and the one with the greatest inconsistency in product development. His first lesson is that the anecdote usually trumps the data, and sophisticated analysis can lose sight of the question. Second, teams often skip articulating a very crisp hypothesis, which leaves research results as merely an interesting conversation.

His third caution is that once you have a hypothesis you can fall in love with it and act as its lawyer rather than its judge. The fourth is to observe more than you ask and to ask indirect questions from multiple directions instead of "would you love this product?" Rachitsky clarifies that the point is to have a view of what customers would say that signals real excitement, so that you recognize it when you hear it.

For an observation example, Duggal cites Procter & Gamble's Swiffer, born from noticing the burden of mopping people had normalized. For a live hypothesis, he points to the joint bank account, which he says dates from an era when women could not open accounts without their husbands' approval. Nubank's hypothesis is that financial lives are inherently social, yet it is easier to share a Spotify playlist than a savings goal.

When testing it, researchers should not sell. Even if a customer says something in the hoped-for direction, the researcher should play devil's advocate and make the customer sell the idea to them.

Duggal spent almost a decade in strategy consulting before Google, where he says it took a couple of years to realize his job was to ship code, not make PowerPoints. He pushes back on Silicon Valley's dogma that strategy is easy and execution is everything, noting that we hear about companies that got both right, not those whose great execution was "multiplied by zero."

He quotes Kevin Systrom from early Instagram: "We may not be right, but at least we are clear." Clarity lets you detect when you are going off course. Drawing on Richard Rumelt's *Good Strategy Bad Strategy*, Duggal says strategy is not an ambitious goal or vision but a coherent plan to apply strengths in a leveraged way against a core problem.

Nubank's 2014 strategy, he says, was specific: young, urban, middle-class Brazilians who hated credit card fees and could not get credit without a history, served by a no-fee card that was possible because the company was branchless and digital. That detail focuses the minimal set of activities needed to win, whereas lacking coherence leads to serving everyone or entering Mexico prematurely.

He argues that concentration builds wealth while diversification preserves it, and startups are building. Hedging is appealing and sounds smart, but concentrated bets require courage. He worries about successful companies with one or two monster products that grow conservative and forget focus, calling that a slow path to death.

Duggal says he has had a mantra since he joined: "We're in the business of being fundamentally different, not incrementally better." It is now printed on product team schwag. He notes that not every one of a hundred launches a year can be fundamentally different, but Nubank searches for anchor features that are, and builds sustaining innovation around them.

On the recurring podcast debate about category creation, Duggal acknowledges Todd Jackson's counterargument to go where customers already spend money, but says successful companies at company level are typically in the new-category business. Nubank set out to build a branchless bank, putting a bank branch in your pocket, with a digital cost structure that enabled disruptive pricing across the board.

He adds that unlike much of the world, Nubank chose to do credit first rather than a bank account, which is ten times harder and a hundred times riskier. Those concentrated bets and category definition were, he believes, key to success. Even in established markets, such as Brazil's *consignado* payroll-secured lending for government employees, Nubank tried to be different by going direct to consumers rather than through middlemen and undercutting pricing.

Duggal's conclusion is that when entering a category where customers already spend money, you should aim to reinvent it even if you are not inventing it. Rachitsky notes the do-not-create-a-category side seems to be winning in volume among guests, and Duggal happily accepts the contrarian role.

Duggal says Nubank has roughly a dozen to twenty core product lines depending on how you count. The company began as a Brazilian credit card monoliner, then in its second phase set out to become a full-solution Latin American bank. It launched a bank account, lending, investments, insurance and small business products, and about half of its 90-plus million customers use it as their primary banking relationship.

It also proved that the model was exportable by launching in Mexico and Colombia, and is bringing the full product suite there. A third phase moves beyond financial services, including an e-commerce marketplace inside the app, and a vision of a personal banker for everyone, something he says only around 10 million people around the world can afford today.

Asked about sequencing new lines, Duggal says there was no scientific master plan. It came from smart people with a culture of robust, candid debate and constant re-examination, with obvious inputs such as market size but a lot of art. Rachitsky notes that the early lesson of gating on the 50% score still seems central, and Duggal agrees that Nubank avoids spending effort fixing massive problems it has over-scaled before they were ready.

Duggal lays out what he calls unproven hypotheses. Banking should be holistic across life's financial seasons: spend, save, invest, borrow and protect. Nubank also aims to build what it believes is the first global bank on a single code base, in contrast to global banks like Citi or Santander assembled through acquisitions on distinct technology infrastructures, even though regulations are local and stakes are high.

The third hypothesis is social mechanics around financial life, not another WhatsApp or Twitter, combined with self-driving automation. He asks why people must remember to pay bills or save for a child's education, noting that he and his wife open different bank accounts just to manage goals. Customers everywhere, he says, live "harshly unoptimized financial lives," and a personal banker beside you could tell you the smart move after a marriage, child or house purchase.

In the AI Corner, Duggal says Nubank uses tools like a ring-fenced ChatGPT to improve customer experience and efficiency, and is now thinking about AI-native products. He recalls Fidji Simo's analogy of becoming mobile native at Facebook, arguing companies must ask what they would design if AI existed from the start rather than appending it at the corners. He admits Nubank has not cracked it, but he asks why the company that reinvents banking should not come from São Paulo, Mexico City or Bogotá.

In the failure corner, Duggal tells of leaving grad school at Harvard's Kennedy School in 2006 after a friend's startup, which aimed to bring AdWords-style advertising to terrestrial radio, was acquired by Google. The hypothesis proved harder than expected: broadcast advertising served brand goals rather than bottom-of-funnel ones, needed reach and frequency methods the team did not know, and had scarce inventory controlled by powerful players like NBC, CBS and Clear Channel.

Google pivoted, concluding that rather than chasing TV, TV should come to Google via YouTube, and Duggal's group was collateral damage, with a layoff looming. He had dropped out of two programs and moved 3,000 miles west, and he kept a song with the line "just because I'm losing doesn't mean I've lost" on repeat. After the DoubleClick acquisition he got the mandate for brand display advertising, which he says became Google's second-largest business.

His lesson is that bloody-mindedness is underrated: persistence combined with a clear-eyed view of the odds and openness to market feedback about what to hold and what to let go. His closing synthesis is that as an insurgent you cannot fight on ground where incumbents stand firm, and noise has grown so much that incrementally better no longer breaks through.

In the lightning round he recommends *Play to Win*, *Where to Play and How to Win*, and Lee Kuan Yew's *From Third World to First*. He applies his fundamentally-different motto to life choices too, and he invites listeners to join Nubank's "purple revolution," a nod to the future-is-purple slogan born in 2016 when customers rallied to urge the central bank not to let Nubank shut down.