How Wise Earns 70% of Its Growth by Word of Mouth

Peiris, who has been chief product officer at Wise for more than 11 years, describes the company as a fix for cross-border money movement, which is slow, expensive and hard. It has three products: money transfer, an account for people and businesses, and an enterprise product that puts Wise's infrastructure inside banks and other apps. It now moves about $12 billion a month, is growing 30 to 40% a year, charges about 0.65% on average, and has been profitable for more than four years with 20% EBITDA margins.

The statistic he is proudest of, and the hardest to achieve, is that 70% of the people who found out about Wise last month heard about it through word of mouth. Wise has 16 million customers, about 10 million of them active, and adds roughly a million a quarter. Of a million new joiners, he says, about 700,000 learned of Wise from a friend.

Measuring this is simple: Wise asks customers, using a question built into the sign-up flow. It overlays those answers on referrer and cookie data to work out how much of the "direct" traffic is really word of mouth. Host Lenny Rachitsky notes that Airbnb did exactly the same thing, with a small prompt asking "how did you hear about us?"

Peiris explains that Wise leaned on word of mouth from the start because money is the ultimate commodity. A business that charges to move it must feel special, and it has little margin to spend on paid marketing. When he joined about a decade earlier there was no playbook for word of mouth, so Wise began by asking customers and then used net promoter score as the best proxy for what drives recommendations.

When Wise overlaid referral data on its NPS survey, the pattern was striking. Invite rates were very low among low scorers, and moving up through the 7-8 group, then 8-9, then 9-10, roughly doubled the number of people customers told at each step. That is why Peiris prefers moving NPS from 30 to 50 over a one-off 10% gain in conversion rate: it raises the viral coefficient of the whole base.

The comments matter as much as the score. In the first years Wise emailed every NPS comment to the whole company each week, and customers kept saying the same thing: make it faster, cheaper and easier to use. Those became the three product pillars, though Peiris admits the early debates were chaotic, with the team arguing over whether trust or something else belonged on the list.

Peiris says he hit the trust problem hardest in marketing, where no landing-page badge could compete with a friend's word. People who arrive through ads often did so because a friend had told them earlier and they remembered it when they needed to send money. Solving trust head-on means unpicking worries like "is my money safe?", whereas delivering an experience so good that people recommend it is a shortcut around many of those problems.

Wise's NPS is about 70, which he says is higher than the iPhone or Google search, while banks and financial services sit around -30. Most teams stop when the product works, he says, and some then polish conversion. To get to recommendation you have to give users "an experience they didn't know was previously possible," such as instant transfers or prices falling from around six percent to 0.35%.

Peiris sees a common misstep in plugging in someone else's infrastructure to open a new market quickly, because it never grows like a hockey stick. Instead he asks for the theoretical minimum cost and maximum speed, and how to reach them in two years rather than inching forward. Rachitsky relates this to working backwards from the ideal, as at Airbnb, where Pixar-style storyboards of the ideal host and guest journey led to instant booking.

Wise launched at about 0.5% and is now around 0.35%, in an industry where retail customers often pay six to seven percent. Peiris says the key was a project, led by him some four or five years earlier, to allocate every bill in the P&L back to the customer or transaction that generated it, and then add a margin. This usually shows that around 20% of customers generate 80% of costs, so those routes pay what they cost and everyone else gets a lower price.

At the transaction level there are only three costs: people, realized risk and partner fees. Wise invests as much cash flow as it can in engineering to remove each. FX risk exposure has been halved over a few years, and people costs, which he thinks of as the cost of poor quality, improve about 20% a year through automation. Step changes, though, need bolder ideas.

One such idea came from Singapore. Regulators granted Wise a license on condition that it meet every customer face to face, so Wise opened an office there, and customers hated it. The company got them to complain to the government rather than to Wise, and after about a year of lobbying and a year of building it won what Peiris calls the world's first eKYC license, letting customers verify with a selfie and a photo of their ID.

Peiris says people recommend for rational reasons, such as price and speed, but also for emotional ones that he prefers to call mission rather than brand. Wise's mission is to make the world's money move instantly, at the touch of a button, for almost nothing. When it rebranded about eight or nine years earlier, it emailed the mission to customers, and the email was forwarded so widely that it brought in more new customers than any other marketing, despite having no call to action or sign-up button.

He credits the authenticity of the mission: Wise's founders were genuinely upset by how much it cost to move money, and the company is full of immigrants and people on visas who have lived with the problem. Customers helping Wise lobby for lower costs, which lets Wise cut prices further, which attracts more customers, creates a flywheel. Rachitsky adds a framework from his Airbnb work: separate what people truly need from their perceived fears, and address both.

Peiris also notes that most businesses keep two lists, one of things that help customers and one of things that make money. At Wise there is a single list of what makes customers happy, prioritized by impact on the hardest problems. If those are solved, he argues, growth and shareholder value follow.

On price, Wise did not run experiments to see whether each cut paid back. Peiris says the strategic bet at the heart of the company is that the lowest-cost, fastest, highest-quality platform will win the world's volume, and measuring each increment would be slower than just investing. The goal then becomes to cut costs by a set amount each period and pass most of the savings on to customers.

He extends this to UX: "You can't split test your way to love." He tells of a product manager who joined the referral team and planned to test everything before deciding what to build. Peiris gave him three weeks to pick one thing, talk to customers, and build his own gut feel before launching and testing it, arguing that data keeps hubris in check while conviction allows much larger changes.

Rachitsky says Airbnb has shifted the same way, toward building products founders and executives are excited about, though he would personally still lean toward experimenting. Peiris concedes the risk that conviction is just opinion, and says Wise still uses betas and small holdout groups to detect harm. Teams are asked for impact each quarter, through pre-post analysis or split tests, but analysis should not hold up a release that will not be rolled back.

Teams are also encouraged to pursue ideas leaders doubt. His head of SEO built a currency converter even though cofounder Kristo called it a crazy idea, and it now drives large amounts of traffic through which people discover Wise.

Peiris says Wise's pillars resemble what Jeff Bezos saw at Amazon: things customers will always want. Wise began with transfers, but realized that people sending only $10 would not download an app when they could use their bank, which led to the account. For businesses, international banking is painful, since getting an Australian bank account normally means flying there and incorporating, while Wise offers local account details in a few clicks.

Launching the account in a market brought about 20 to 30% more cross-border volume from that market. With about $12.5 billion in deposits, the next customer demand was a return. Wise deliberately has no banking license, so it puts customer money in government bonds and sells them dynamically when a card is used, which supports an interest rate of roughly 5% at the time of recording.

Asked how others should approach word of mouth, Peiris says to talk to customers, ask what it would take to be 10x better, and pick the thing everyone thinks is impossible. Instead of finding a cheaper partner bank, Wise went to the cheapest one of all, the central bank, and now has accounts at the Bank of England and other central banks. Each took years of lobbying, helped by a culture, set by the founders, that doing something hard and new is the only worthwhile work.

Peiris says international banks are really collections of local tech stacks with deep integrations, while PayPal has a single global stack but few local connections. Wise sits in between, with one global tech stack and deep local infrastructure. The onboarding flow, for example, uses the same code in Brazil, New York and Australia, even though regulation differs: Japan, he notes, requires photos of the front, back and side of the ID.

To make this work, a global product team owns overall KPIs such as conversion rate, while regional teams own conversion and KYC cost in their markets and contribute directly to the global code base. This is weak product ownership: anyone can submit changes, but the center owns the vision, and the shared model evolves as new market needs, such as Japan's, emerge. This avoids the usual pattern where international is an afterthought competing for priority.

On team structure, Wise began with autonomous teams on a KPI tree, with each team presenting its plan every quarter. That broke down at around 30 teams, so it moved toward a Spotify-style model of squads and tribes. Squads are organized around products, and each squad director owns the vision and results, while tribes provide light-touch strategy.

After 12 years of running a referral program and testing chocolate, cash, and many reward levels (about "three for $100" landed as a sweet spot), a change that produced a 300% increase stood out. A product manager noticed that the post-transfer email said customers had saved $10, but they believed they had saved money without believing the number. He and a designer sketched alternatives and showed them to people in a coffee shop until they arrived at a graph.

The graph compares what the customer would have paid hidden in the bank's exchange rate and fees with what Wise charges. Peiris says people reacted by saying they would never use their bank again. Placing it on the transfer success page beside share and invite buttons tripled the sharing rate, and the comparison later moved into marketing and the homepage. Wise logs prices from about 70 bank accounts around the world to keep it accurate, and Peiris still holds about 17 of them.

Peiris sees the same logic with speed: customers do not realize a transfer was instant, so an animation at the end shows the money has arrived and referrals jump. He calls closing the gap between what has been done and what is perceived "product marketing within the product," and Rachitsky calls it the reality-and-perception framework in reverse: helping people appreciate how remarkable the product already is.