#75 - The Iceberg Theory of Fintech

The greatest Fintechs in the world have built deep underlying capabilities that are unseen. This is the difference between building a Fintech business and a Fintech Solution

Illustrated by Mary Mogoi - Website

Hi all - This is the 75th edition of Frontier Fintech. A big thanks to my regular readers and subscribers. To those who are yet to subscribe, hit the subscribe button below and share with your colleagues and friends. Support Frontier Fintech by becoming a paid subscriber🚀

For group discounts, write to me at samora@frontierfintech.io. If you can’t afford a subscription, consider referring a friend. The more you refer the longer your complimentary subscription is.

Reach out at samora@frontierfintech.io for sponsorships, partner pieces and advisory work. Spaces are becoming limited on my advisory hours. I help clients with market entry, market mapping, strategic insights, sounding board to founders and general advisory across Pan-African Fintech.

---

📢📢I’m carrying out a Reader Survey over the next month to get insights from you on what works, what I should improve, what you value as well as some information that would help me make Frontier Fintech a thriving Media Business. Please fill out the survey below.📢📢 It shouldn’t take more than 3 minutes of your time. To make it worth your time, I’ll give you a months free subscription for filling the survey

---

Sponsored by Oradian

Built for Local Realities

Scaling a fintech or financial institution in Africa means navigating unique regulatory, operational, and market-specific challenges. Many global core banking solutions—built with mature markets in mind—lack the flexibility to accommodate local realities. A prime example? Loan fee structures. In many African markets, lenders need the ability to amortize fees over the length of a loan—a feature that most global core banking platforms don’t natively support. Fintechs and banks are forced to rely on cumbersome workarounds or build expensive custom solutions just to meet everyday business needs.

Oradian solves this by offering a core banking system designed specifically for dynamic markets. Our configurable platform adapts to local requirements, allowing financial institutions to operate seamlessly without compromise. From flexible fee structures to regulatory-ready workflows, Oradian provides the infrastructure you need to scale efficiently. Stop forcing global solutions to fit local challenges. Discover how Fairmoney uses Oradian to power its advanced lending business;

---

Introduction

All

Here’s a Quick Update on the top priorities for X.com (Paypal not X as in Twitter) over the next month.

1. Fraud Prevention: Max Levchin will be leading this effort, and Sarah Imbach will be coordinating the requisite engineering, financial and operational pieces of the company needed to make this happen. The good news is that the fraud crisis is readily containable and that we have a number of great front-end solutions (preventing fraudsters before they enter the system) and back-end solutions (detecting fraudsters once they enter the system).

2. Product cycle/V1 Platform. The product cycle is going to be accelerated as fast as possible, and therefore all engineering resources will be concetrated on the V1 Platform…

3. Branding. There will be no change in our dual branding. The product will be called PayPal (because that’s what consumers are familiar with) and the company will be called X.com (because that’s what investors are familiar with).

4. X-Finance - We will close down X-Finance Operations and Start to consolidate everything into Paypal. All of those people working on X-Finance will be shifted over to the PayPal product, because that one requires all our focus at this point in time.

Thanks,

Peter.

For this article, I’ve greatly referenced The Founders: The Story of Paypal and the Entrepreneurs Who Shaped Silicon Valley by Jimmy Soni

The Looming Fraud Crisis at Paypal

This was written on September 28th, 2000 by Peter Thiel. The context was that Elon Musk had just been kicked out by the board as CEO and Thiel was now in charge ideally as interim CEO but begrudgingly for himself as the actual CEO. Paypal was in a teenage crisis of sorts, the company had raised a massive fund raise in 1999 but was quickly running out of cash as it scaled. Up to this point, the entire focus of the company had been on growing and improving the product with Musk and David Sacks being critical in this strategic approach. One of the ways they super-charged growth was through offering bonuses of up to US$ 20 to any new customer who registered whilst making it extremely easy to onboard as a user. Unfortunately, the outcome was fraud.

Enter Roelof Botha. Botha was an actuarial science student who was studying at Stanford. He had met Musk serendipitously in the fall of 1999 and Musk immediately tried to recruit him as he did with every intelligent person he met. Botha emitted a precocious intelligence and grasp of numbers. Moreover, he combined this with an infectious fascination with the world of business. Botha declined the attempt at recruitment as he was on a student visa, besides, he didn’t want to drop out of Stanford. However, Botha said “There are people you meet, and then two weeks later, you can’t remember anything about them… Elon lingers”. Botha continued thinking about X.com as it was then called, his conclusion was that “it din’t have a natural advantage” particularly, it didn’t have any network effects and its unit economics didn’t make sense. However as fate would have it, a financial crisis in South Africa decimated his personal savings and he needed a job to make rent.

He reached out to Musk who asked Peter Thiel to interview him. Thiel threw a question at Botha;

There’s a perfectly round table of indeterminate length, and you don’t know the length in advance. Two players have a bag of quarters of infinite depth. Each player can place a coin in the table, and they can touch but not overlap. The last person who puts a coin down the table and fills it up wins the game. Is there a way to guarantee victory in advance, and does that involve going first or second.

These types of brain teasers were a regular thing at Paypal, arguably a grouping of some of the smartest people on earth. Botha fit right in as he aced the answer and got started building a financial model for the company. After working long hours on the model, he quickly realised that things could go awry pretty fast for the company if they didn’t get a handle on fraud. Specifically, there was a time lag between when a chargeback was reported and when it was paid. In his words “the chargebacks we were seeing in May were related to transactions from February or March”. Given that the company was growing exponentially, future chargebacks and fraudulent transactions would also grow exponentially. “If we don’t fix this, we are going to die”. In addition to this, there was a growing and ominous trend where international criminals took advantage of PayPal by using stolen credit cards, foreign drop-ship sites and even shell companies. He explained that “this unauthorised fraud had no brake on it”.

The massive challenge was that PayPal needed to continue growing by being user-friendly whilst devising ever more intelligent ways of battling fraud. With cash soon running out and in the midst of the dot-com crash, this was a matter of life and death as there was no one to save them.

A picture of the PayPal Mafia with David Sacks, Peter Thiel, Max Levchin, Reid Hoffman, Roelof Botha and Keith Rabois.

Max and his Fraudbusters

Enter Max Levchin the CTO of PayPal and his young and talented team of engineers. Max understood this problem better than most. In fact Botha in early 2000s discussed this issue with Max. Botha, as an actuarial science student, had used chain-ladder analysis to forecast the growing cost of fraud and shared his insights with Max. Max then called Luke Nosek way past midnight in the summer of 2000 and said “Luke, I think we’re dead”. Such was the scale of fraud. This was prior to the board coup and Musk was still in charge, Max had some time on his hands and he immersed himself in understanding this fraud menace. Together with his team he got working on some innovations that are still critical to payment companies to this day.

1. Gausebeck-Levchin test - The first challenge the tacked was that fraudsters were using bots to automatically open PayPal accounts and earn the US$ 10-20 bonuses that the company paid out to new customers. The solution relied on a simple intuition borne out of the Turing Test; “I propose to consider the question, can machines think?”. Simply, what can a human do very easily that a machine can’t. The reason this question was framed like this was that defeating bot fraud required PayPal to devise a method to differentiate between a human and a robot. Supported by David Gausebeck, the two came up with a very simple solution. “Why don’t we put images of characters and require a user to type them in?” Gausebeck asked Max. They then added lines and distorted the words a bit more. The Gausebeck-Levchin test became the first commercial application of a Completely Automated Public Turing Test to Tell Computers and Humans Apart or (CAPTCHA) as we now know it. This innovation proved surprisingly robust and has become a key component of the internet. Interestingly, when Max proposed this to David Sacks who was obsessed with simplifying the user experience, he replied “Are you fucking kidding me No one’s going to understand this… You want to do what to my sign-up page?” The tension between Sacks and Max later became a critical pillar of PayPal’s success;
   

   

   The Picture above shows an early version of the Captcha test - This one was patented by Alta Vista
   

2. Visualising Fraud - Fighting fraud is like a game of whack-a-mole. Once you deal with one issue, another pops up. There was a persistent fraudster called Igor and one of his tricks was to create two accounts, both legitimate looking to pass the initial screening process. Then after enough time had passed, he would use one account to purchase goods from the other using a stolen credit card number. The fake seller would then withdraw the money to a non-Paypal Bank account. Detecting such fraud typically had been done manually by a fraud team. Levchin together with a genius of a computer scientist called Bob Frezza thought that rather than going through transactions manually, why can’t we visualise them so that we can “see the patterns of fraud”. Together with a team of engineers, they built this visualisation product that showed transactions between PayPal accounts with the thickness of each line representing the amounts being transacted. This enhanced human intuition and enabled the fraud team to see fraud as it happened. Through additional work, Levchin and his teams could now not only just compare numbers against numbers i.e. amounts and frequencies, they could now compare patterns to patterns. This enabled them to get alerts when one pattern of a transaction resembled an earlier pattern of fraud. This new fraud detection system was aptly named IGOR after the famous fraudster who inspired its creation. It has become a modern pillar of fraud detection in banks and global payment companies.
   

3. Random Forests - Following on from their culture of hiring smart people and tasking them to figure out seemingly intractable problems, a young fraud analyst called Mike Greenfield fit this mould perfectly. One of the things Mike Greenfield did was that he built software that generated multiple decision trees—an early application of random forests—to spot fraudulent transactions with pinpoint accuracy. Mike basically threw tons of transaction data into the software and eventually it turned out to be great at identifying merchant fraud. As he explained “After going down eighteen steps, we’d say, Okay, this transaction has a twenty percent chance of being bad. This other one has a 0.01 percent chance”. Whereas banks and payment companies at the time had basic regression models, PayPal had pioneered big data and advanced fraud models.

The True Differentiator

Max Levchin reflected at the time that;

PayPal is actually a, more or less, commodity business… it sounds very cool and innovative.. moving money on the internet. But the credit card interfaces have existed for twenty years… All we really did was put a very pretty web front on it and let people use their email addresses instead of their account number… but beneath the surface, PayPal’s core innovation sparked. The submerged part of PayPal is this massive, and very, very numerically driven risk management system which allows us to instantaneously tell when you’re moving money to someone else, with a very high degree of certainty whether the money you’re moving is yours, or whether you got it illegally and we might be on the hook later on to help the authorities investigate or retrieve the money

Luke Nosek added a very interesting spin on this when he said that “Fraud saved us on accident… and it was cheaper than buying Super Bowl ads”

Fraud detection and the work of Levchin and his genius mathematicians and engineers laid the foundation for modern day PayPal. Thiel later reflected that the work of Levchin’s team drove fraudsters to other platforms ensuring that PayPal had the lowest fraud rates in the market. Botha acknowledged that the reduced fraud costs by US$ 2m a month. Enough to ensure survival. What’s more critical is that engineering and product efforts could then be diverted to growth. The outcome was initially a US$ 1.5 billion acquisition of PayPal by eBay. However the larger outcome is that PayPal is a US$ 70 billion company with annual revenues of US$ 31 billion and a TPV of US$ 1.7 trillion dollars.

Interestingly, prior to eBay acquiring PayPal, they had spent significant resources trying to ensure that their own product Billpoint would be the default payment system on eBay ahead of PayPal. This was a bruising battle between the year 2000 and 2002. However, PayPal was such a superior product that they ended up buying. Meg Whitman, then CEO of eBay reflected on the aggression and smarts of the PayPal team, acknowledging them as an exceptional team.

The Iceberg Theory of Fintech

“The key to investing is not assessing how much an industry is going to affect society, or how much it will grow, but rather determining the competitive advantage of any given company and, above all, the durability of that advantage.” - Warren Buffett

This was quite an introductory section but it matters. The larger point like Max Levchin stated is that behind every successful Fintech company is a submerged iceberg that contains the company’s core value proposition. It’s an interesting insight and one that was the subject of an entire book by the Co-Founder of Square, Jim McKelvey. Whereas I call it an Iceberg, Jim referred to it as the innovation Stack. The Innovation Stack according to Jim McKelvey is a series of interlocking innovations that together build an unbeatable business. Rather than a single breakthrough, an innovation stack is a layered system of solutions: solving one problem leads to the next, resulting in a cascade of novel capabilities. This is like an iceberg, where the visible product or service is just the tip, and the bulk of value lies beneath the surface in the form of proprietary technology, processes, and partnerships that competitors can’t easily see or copy.

I’ve experienced this at a smaller scale in work in a bank. We had built the leading agency banking system where each agent was also a merchant. It was a mix of traditional agency banking coupled with merchant acceptance and vendor capabilities. For us, the underlying stack was built around our “Sales Force Efficiency” in terms of daily agent management, innovations around float management and our relationships with Telcos as well as having a mix of ex FMCG leaders in the broader agency team. This was my first experience from a managerial perspective with the idea of an iceberg in Fintech. Other banks launched their agency programs but they struggled to catch on largely cause they copied the branding and app and not the submerged capabilities.

Interestingly, this idea is at the centre of all successful Fintech companies in the world. It’s not necessarily novel. It’s the core idea behind Zero to One by Peter Thiel and Thinking in Systems by Donella Meadows and Diana Wright. Both books are a must read. In Zero to One, Thiel emphasises the need to create a 10x improvement on the existing solutions, going from Zero to One. This requires proprietary technology or know-how. In Thiel’s words, a true engineering advantage.

Icebergs in Action - Africa

In Africa, there are a couple of interesting examples of the Iceberg theory in action.

M-Pesa by Safaricom

Source: Quartz Africa - The entire country is not green by accident.

The Tip of the Icerberg - At the tip is a simple STK and mobile app that enables payments across P2P, merchant and remittance use cases supported by a network of 688,000 agents as well as over one million merchants.

Beneath the Surface - Beneath the surface is a combination of a number of factors. Of course fraud capabilities, an entire telco infrastructure and an advanced mobile money system are all known. One thing that goes unsaid though is their world class ground game - Behind M-Pesa’s success is an exceptional ground game of agent management, merchant acquisition and general sales capabilities. Built from the start and relying on Safaricom’s FMCG heritage, M-Pesa has always focused on extensive agent and merchant management. I remember when I was working at a company called Zimele, M-Pesa was helping us enable our customers to save through M-Pesa. These were merchant capabilities. M-Pesa’s team would conduct trainings every Saturday for close to two months around reconciliations, transaction management and branding. I then realised that the success of M-Pesa wasn’t by accident. To beat M-Pesa, you have to beat an over 20 year investment in on-ground execution.

Moniepoint

Source: Rest of the World

The Tip of the Iceberg - At the tip of Moniepoint’s iceberg is a business banking product, an agent product as well as an extensive merchant solution that combines off-line acquiring and transaction processing.