February 2023 - Anatomy of the Swipe by Ahmed Siddiqui

This month we dive back into the world of payments and take a refreshed look at how payments companies work.

Tech Themes

Authorization. How do credit card’s even work? What is on the magnetic stripe? It turns out that each stripe has an alternating set of tiny magnets on it that produce a magnetic field around the card. The card reader on a POS system is a solinoid; when the magnets swipe through the solinoid it creates a change in the magnetic field also known as magnetic flux. The POS processes the changes in current as its swiped through the Solinoid, and is able to understand the credit card via common card standards, created by ISO. The challenge that existed with the internet, is how to ensure safe transactions when you are paying someone you clearly don’t know. The card companies created something called the card verification value (CVV), an extra number directly intended NOT to be written anywhere except for the back of your credit card. CVV codes were originally created by an Equifax employee in the UK, and initially rolled out by NatWest bank, eventually expanding to Mastercard (1997), Amex (1999), and Visa (2001). However, the early internet still had a massive fraud problem, as Roelof Botha discussed about Paypal in his Tim Ferris podcast appearance. In addition, card authorization was still quite difficult in Europe, where you would frequently have to call to provide authentication for cross-border purchases. In 1993, Europay, Mastercard, and Visa formed EMVco, to add additional fraud protection to cards. They were subsequently joined by other key members including American Express, Discover, JCB International, China UnionPay. Europay eventually merged with Mastercard in 2002 prior to Mastercard’s IPO. EMVco creates a standard for EMV chips, which embed small integrated circuits that produce a single-use cryptographic key for a merchant to decrypt and authorize a transaction. It is also incredibly difficult to clone a chip card, whereas magnetic stripes are fairly easy to copy. The increase in fraud protection was so great that it caused a liability shift, whereby the merchant (rather than the issuer) could become liable for fraud if a EMV chip card was not used, and a swipe was used instead. The latest innovation in card security is 3D Secure 2, which “allows businesses and their payment provider to send more data elements on each transaction to the cardholder’s bank. This includes payment-specific data like the shipping address, as well as contextual data, such as the customer’s device ID or previous transaction history.” The 3D Secure 2 also improves the UX of its “challenge flow,” which is an instance when a frictionless authentication wasn’t possible, for whatever reason. The challenge flow forces the user to authenticate the transaction through their banking application of choice, which is much more secure than just approving every transaction the network sees. Every day payments become more at risk and more secure!
Zelle and banks Funded Payments. Similar to Visa and Mastercard, Zelle is a cash transfer system originally created by a consortium of banks. In 2011, Bank of America, JP Morgan, Wells Fargo, and several other banks built a Paypal competitor called clearXchange. The new company would charge financial institutions to use the service, with most banks assuming the charges on behalf of their consumers. At the time, Venmo was beginning to take off with consumers utilizing its simple Peer-to-peer payments service. In 2012, Braintree acquired Venmo for just $26m. The low purchase price was the result of a lack of coherent business model, given Venmo’s founding by college roommates who were looking to send money easily. Later, in 2013, Braintree was acquired by Paypal for $800m. Braintree is an acquirer processor and introduced a business model to Venmo, namely investing the float of customer funds held in the Venmo ecosystem. In 2016, clearXchange was acquired by another bank run service called Early Warning Service, which provides risk management services to many financial institutions. Early Warning was itself created in 1990 by Bank of America, BB&T, Capital One, JP Morgan, and Wells Fargo. Early Warning launched Zelle in 2017, utilizing clearXchange’s underlying technology. Zelle is now massive, processing over $1m in volume a minute, which is more than competitors Venmo and Cash App. Zelle’s most recent stats are mind-blowing: 2.3B payments with $629B in volume. Look at creation of zelle and other examples of banks creating new companies (like visa/mastercard)
Money for Nothing, SaaS for Free. A new brand of payments companies are popping up that seek to turn a traditional SaaS model on its head. Divvy, the spend management platform, gives its SaaS software away for free, instead monetizing just the payments processed through its product. Its quite wild to see a company build complex software just to give it away, right? This strategy has been copied many times over, as we talked about when discussing Netscape and Slack. Whether its open source or just free commercial software, the free-ness of it makes it attractive. However, if everyone is free, then you still have to compete on merit. Ramp, a competitor to Divvy, raised $750m at an $8.1B valuation in 2022, and processed over $5B in payment volume in 2021. Divvy was acquired by Bill.com in 2021 for $2.5B in a mix of cash and stock. At the time of acquisition Divvy was doing just over $100m of annualized revenue, and about $4B of TPV, suggesting a take rate of ~2.5%. Ramp allegedly did about $100m in annualized revenue in 2022. While Divvy was able to find a successful exit through a willing buyer (a buyer who’s stock has declined 65% from all time highs), I’m not sure Ramp will find an easy buyer at $8.1B, but it may find the public markets if it can market itself as a cost-saving, AI, finance play. I’m just not sure that you can build a really big business by only processing payments and giving away complex software for free. We will see in time!

Business Themes

Issuer, Issuer Processor, Acquirer, Acquirer Processor. The arc of an individual payment can be broken into its constituent parts. The Issuer is normally a bank that issues credit cards to its banking customers. A bank may use an issuer processor to manage a connection to the card networks (like Visa and Mastercard) and accept/decline transactions. Examples of issuer processors include Marqeta, TSYS, and Galileo. Global Payments and TSYS merged in 2019, in an absolutely massive $21.5B deal, after Fiserv acquired FirstData for $22B and Fidelity National Services acquired WorldPay for $34B. 2019 was definitely a banner year for payments mergers, but signs of strain are already happening, with FIS announcing they’d be spinning out WorldPay in 2023. Back to our transaction - the merchant will have a payment terminal of some sort, and will use an acquirer processor like Chase Paymentech, Tabapay, or Fiserv, that will also have a fast connection to the card networks to request approval for a transaction. Lastly, we have the acquiring bank, the merchant’s bank account. When we look back at these big deals, its clear that each player was trying to round out its processing capabilities - TSYS (issuer processor) with Global Payments (a merchant acquirer), Fiserv (merchant acquirer) with First Data (issuer processor), and FIS (issuer processor) with WorldPay (merchant acquirer). FIS, Fiserv, and Global Payments have struggled to win over investors over the last five years, following these big deals.
Chargebacks. When a consumer doesn’t want to pay for an item, it can request a chargeback. The reasons for a chargeback can be numerous and valid, including fraud, item not as described, duplicate transactions, and more. In the event of fraud, a cardholder would dispute the charge with their bank. The bank would freeze their card and file a chargeback with the card network (Visa as an example). Visa would send a provisional credit to the customers card and take the money back from the merchant, then it would send the chargeback request detail to the merchant. If the merchant doesn’t dispute the chargeback, it will be assessed a $25-35 chargeback fee by Visa. However, if it does dispute the chargeback, and correctly can identify that the card actually did purchase the goods, then the issuer, the bank that is “underwriting” credit to the consumer, can be put on the hook for the funds used for the purchase. I never knew that both the merchant and issuer can be on the hook for chargebacks, but not the network! Another way in which Visa and Mastercard make money through the ecosystem!
Marqeta’s confusion. For our first payments book, we took a look at several of the new players in the credit card ecosystem, including Marqeta, Adyen, and Stripe. Its been quite a two years! Marqeta went public in June 2021, valuing the company at $15B. Stripe raised additional funds at a $95B valuation, and Adyen’s valuation hit $98B! But oh how the times change! Just two years later, and Stripe’s valuation is back at $50B, including a massively dilutive $6.5B raise to pay for employee taxes in option conversion. Adyen’s valuation sits at $53B today, a close to 50% decline, despite growing EBITDA 16% to 728M in 2022.. Marqeta may have had the worst time of all, which is said because Ahmed Siddiqui, worked at Marqeta for a number of years. Marqeta’s stock fell 85%, its CEO/Founder left the company, its gross margins have compressed from high 40’s back to down to low 40s, and its main customer Block has become an even larger customer, now driving 77% of its revenue. Marqeta went from next generation issuer processor and Stripe wannabe to an outsourced custom development shop for Block. My guess is its actual reputation sits somewhere in between the two. Expectations for Marqeta have fallen off a cliff, and its market cap sits at a tiny $2.6B. I’m not sure Block would be an immediate acquirer, because the market for issuer processing is incredibly competitive and Block has had its own stock price troubles. A spun out WorldPay could make sense as an acquirer. Visa would have made sense as acquirer, because it owns about 2.5% of Marqeta and has for many years, but their recent acquisition of Pismo, believed to be a LATAM focused and better version of Marqeta. It’s unlikely Marqeta will exist long as a small solo issuer processor!
Dig Deeper

June 2021 - Letters to the Nomad Partnership 2001-2013 (Nick Sleep's and Qais Zakaria's Investor Letters)

This month we review a unique source of information - mysterious fund manager Nick Sleep’s investment letters. Sleep had an extremely successful run and identified several very interesting companies and characteristics of those companies which made for great investments. He was early to uncover Amazon, Costco, and others - riding their stocks into the stratosphere over the last 20 years. These letters cover the internet bubble, the 08/09 crisis, and all types of interesting businesses across the world.

Tech Themes

Scale Benefits Shared. Nick Sleep’s favored business model is what he calls Scale Benefits Shared. The idea is straight forward and appears across industries. Geico, Amazon, and Costco all have this business model. Its simple - companies start with low prices and spend only on the most important things. Over time as the company scales (more insured drivers, more online orders, more stores) they pass on the benefits of scale to the customer with even further lower prices. The consumer then buys more with the low-cost provider. This has a devastating effect on competition - it forces companies to exit the industry because the one sharing the scale benefits has to become hyper-efficient to continue to make the business model work. “In the case of Costco scale efficiency gains are passed back to the consumer in order to drive further revenue growth. That way customers at one of the first Costco stores (outside Seattle) benefit from the firm’s expansion (into say Ohio) as they also gain from the decline in supplier prices. This keeps the old stores growing too. The point is that having shared the cost savings, the customer reciprocates, with the result that revenues per foot of retailing space at Costco exceed that at the next highest rival (WalMart’s Sam’s Club) by about fifty percent.” Jeff Bezos was also very focused on this, his 2006 annual letter highlighted as much: “Our judgment is that relentlessly returning efficiency improvements and scale economies to customers in the form of lower prices creates a virtuous cycle that leads over the long-term to a much larger dollar amount of free cash flow, and thereby to a much more valuable Amazon.com. We have made similar judgments around Free Super Saver Shipping and Amazon Prime, both of which are expensive in the short term and – we believe – important and valuable in the long term.” So what companies today are returning scale efficiencies with customers? One recent example is Snowflake - which is a super expensive solution but is at least posturing correctly in favor of this model - the recent earnings call highlighted that they had figured out a better way to store data, resulting in a storage price decrease for customers. Fivetran’s recent cloud data warehouse comparison showed Snowflake was both cheaper and faster than competitors Redshift and Bigquery - a good spot to be in! Another example of this might be Cloudflare - they are lower cost than any other CDN in the market and have millions of free customers. Improvements made to the core security+CDN engine, threat graph, and POP locations result in better performance for all of their free users, which leads to more free users, more threats, vulnerabilities, and location/network demands - a very virtuous cycle!
The Miracle of Compound Growth & Its Obviousness. While appreciated in some circles, compounding is revered by Warren Buffett and Nick Sleep - it’s a miracle worth celebrating every day. Sleep takes this idea one step further, after discussing how the average hold period of stocks has fallen significantly over the past few decades: “The fund management industry has it that owning shares for a long time is futile as the future is unknowable and what is known is discounted. We respectfully disagree. Indeed, the evidence may suggest that investors rarely appropriately value truly great companies.” This is quite a natural phenomenon as well - when Google IPO’d in 2004 for a whopping $23bn, were investors really valuing the company appropriately? Were Visa ($18Bn valuation, largest US IPO in history) and Mastercard ($5.3Bn valuation) being valued appropriately? Even big companies like Apple in 2016 valued at $600Bn were arguably not valued appropriately. Hindsight is obvious, but the durability of compounding in great businesses is truly a myth to behold. That’s why Sleep and Zakaria wound down the partnership in 2014, opting to return LP money and only own Berkshire, Costco, and Amazon for the next decade (so far that’s been a great decision!). While frequently cited as a key investing principle, compounding in technology, experiences, art, and life are rarely discussed, maybe because they are too obvious. Examples of compounding (re-investing interest/dividends and waiting) abound: Moore’s Law, Picasso’s art training, Satya Nadella’s experience running Bing and Azure before becoming CEO, and Beatles playing clubs for years before breaking on the scene. Compounding is a universal law that applies to so much!
Information Overload. Sleep makes a very important but subtle point toward the end of his letters about the importance of reflective thinking:
BBC Interviewer: “David Attenborough, you visited the North and South Poles, you witnessed all of life in-between from the canopies of the tropical rainforest to giant earthworms in Australia, it must be true, must it not, and it is a quite staggering thought, that you have seen more of the world than anybody else who has ever lived?”
David Attenborough: “Well…I suppose so…but then on the other hand it is fairly salutary to remember that perhaps the greatest naturalist that ever lived and had more effect on our thinking than anybody, Charles Darwin, only spent four years travelling and the rest of the time thinking.”
Sleep: “Oh! David Attenborough’s modesty is delightful but notice also, if you will, the model of behaviour he observed in Charles Darwin: study intensely, go away, and really think.”
There is no doubt that the information age has ushered in a new normal for daily data flow and news. New information is constant and people have the ability to be up to date on everything, all the time. While there are benefits to an always-on world, the pace of information flow can be overwhelming and cause companies and individuals to lose sight of important strategic decisions. Bill Gates famously took a “think week” each year where he would lock himself in a cabin with no internet connection and scan over hundreds of investment proposals from Microsoft employees. A Harvard study showed that reflection can even improve job performance. Sometimes the constant data flow can be a distraction from what might be a very obvious decision given a set of circumstances. Remember to take some time to think!

Business Themes

Psychological Mistakes. Sleep touches on several different psychological problems and challenges within investing and business, including the role of Social Proof in decision making. Social proof occurs when individuals look to others to determine how to behave in a given situation. A classic example of Social Proof comes from an experiment done by Psychologist, Stanley Milgram, in which he had groups of people stare up at the sky on a crowded street corner in New York City. When five people were standing and looking up (as opposed to a single person), many more people also stopped to look up, driven by the group behavior. This principle shows up all the time in business and is a major proponent in financial bubbles. People see others making successful investments at high valuations and that drives them to do the same. It can also drive product and strategic decisions - companies launching dot-com names in the 90’s to drive their stock price up, companies launching corporate venture arms in rising markets, companies today deciding they need a down-market “product-led growth” engine. As famed investor Stan Druckenmiller notes, its hard to sit idly by while others (who may be less informed) crush certain types of investments: “I bought $6 billion worth of tech stocks, and in six weeks I had lost $3 billion in that one play. You asked me what I learned. I didn’t learn anything. I already knew that I wasn’t supposed to do that. I was just an emotional basketcase and I couldn’t help myself. So maybe I learned not to do it again, but I already knew that.”
Incentives, Psychology, and Ownership Mindset. Incentives are incredibly powerful in business and its surprisingly difficult to get people to do the right thing. Sleep spends a lot of time on incentives and the so-called Principal-Agent Conflict. Often times the Principal (Owner, Boss, Purchaser, etc.) may employ an Agent (Employee, Contractor, Service) to accomplish something. However the goals and priorities of the principal may not align with that agent. As an example, when your car breaks down and you need to go to a local mechanic to fix it, you (the principal) want to find someone to fix the car as well and as cheaply as possible. However, the agent (the mechanic) may be incentivized to create the biggest bill possible to drive business for their garage. Here we see the potential for misaligned incentives. After 5 years of really strong investment results, Sleep and Zakaria noticed a misaligned incentive of their own: “Which brings me to the subject of the existing performance fee. Eagle-eyed investors will not have failed but notice the near 200 basis point difference between gross and net performance this year, reflecting the performance fee earned. We are in this position because performance for all investors is in excess of 6% per annum compounded. But given historic performance, that may be the case for a very long time. Indeed, we are so far ahead of the hurdle that if the Partnership now earned pass-book rates of return, say 5% per annum, we would continue to “earn” 20% performance fees (1% of assets) for thirty years, that is, until the hurdle caught up with actual results. During those thirty years, which would see me through to retirement, we would have added no value over the money market rates you can earn yourself, but we would still have been paid a “performance fee”. We are only in this position because we have done so well, and one could argue that contractually we have earned the right by dint of performance, but just look at the conflicts!” They could have invested in treasury bonds and collected a performance fee for years to come but they knew that was unfair to limited partners. So the duo created a resetting fee structure, that allowed LPs to claw back performance fees if Nomad did not exceed the 6% hurdle rate for a given year. This kept the pair focused on driving continued strong results through the life of the partnership.
Discovery & Pace. Nick Sleep and Qais Zakaria looked for interesting companies in interesting situations. Their pace is simply astounding: “When Zak and I trawled through the detritus of the stock market these last eighteen months (around a thousand annual reports read and three hundred companies interviewed)…” Sleep and Zakaria put up numbers: 55 annual reports per month (~2 per day), 17 companies interviewed per month (meeting every other day)! That is so much reading. Its partially unsurprising that after a while they started to be able to find things in the annual reports that piqued their interest. Not only did they find retrospectively obvious gems like Amazon and Costco, they also looked all around the world for mispricings and interesting opportunities. One of their successful international investments took place in Zimbabwe, where they noticed significant mispricing involving the Harare Stock Exchange, which opened in 1896 but only started allowing foreign investment in 1993. While Nomad certainly made its name on the Scaled efficiencies shared investment model, Zimbabwe offered Sleep and Zakaria to prioritize their second model: “We have little more than a handful of distinct investment models, which overlap to some extent, and Zimcem is a good example of a second model namely, ‘deep discount to replacement cost with latent pricing power.’” Zimcem was the country’s second-largest cement producer, which traded at a massive discount to replacement cost due to terrible business conditions (inflation growing faster than the price of cement). Not only did Sleep find a weird, mispriced asset, he also employed a unique way of acquiring shares to further increase his margin of safety. “The official exchange rate at the time of writing is Z$9,100 to the U$1. The unofficial, street rate is around Z$17,000 to the U$1. In other words, the Central Bank values its own currency at over twice the price set by the public with the effect that money entering the country via the Central Bank buys approximately half as much as at the street rate. Fortunately, there is an alternative to the Central Bank for foreign investors, which is to purchase Old Mutual shares in Johannesburg, re-register the same shares in Harare and then sell the shares in Harare. This we have done.“ By doing this, Nomad was able to purchase shares at a discounted exchange rate (they would also face the exchange rate on sale, so not entirely increasing the margin of safety). The weird and off the beaten path investments and companies can offer rich rewards to those who are patient. This was the approach Warren Buffett employed early on in his career, until he started focusing on “wonderful businesses” at Charlie Munger’s recommendation.

Dig Deeper

March 2021 - Payments Systems in the U.S. by Carol Coye Benson, Scott Loftesness, and Russ Jones

This month we dive into the fintech space for the first time! Glenbrook Partners is a famous payments consulting company. This classic book describes the history and current state of the many financial systems we use every day. While the book is a bit dated and reads like a textbook, it throws in some great real-world observations and provides a great foundation for any payments novice!

Tech Themes

Mapping Open-Loop and Closed-Loop Networks. The major credit and debit card providers (Visa, Mastercard, American Express, China UnionPay, and Discover) all compete for the same spots in customer wallets but have unique and differing backgrounds and mechanics. The first credit card on the scene was the BankAmericard in the late 1950’s. As it took off, Bank of America started licensing the technology all across the US and created National BankAmericard Inc. (NBI) to facilitate its card program. NBI merged with its international counterpart (IBANCO) to form Visa in the mid-1970’s. Another group of California banks had created the Interbank Card Association (ICA) to compete with Visa and in 1979 renamed itself Mastercard. Both organizations remained owned by the banks until their IPO’s in 2006 (Mastercard) and 2008 (Visa). Both of these companies are known as open-loop networks, that is they work with any bank and require banks to sign up customers and merchants. As the bank points out, “This structure allows the two end parties to transact with each other without having direct relationships with each other’s banks.” This convenient feature of open-loop payments systems means that they can scale incredibly quickly. Any time a bank signs up a new customer or merchant, they immediately have access to the network of all other banks on the Mastercard / Visa network. In contrast to open-loop systems, American Express and Discover operate largely closed-loop systems, where they enroll each merchant and customer individually. Because of this onerous task of finding and signing up every single consumer/merchant, Amex and Discover cannot scale to nearly the size of Visa/Mastercard. However, there is no bank intermediation and the networks get total access to all transaction data, making them a go-to solution for things like loyalty programs, where a merchant may want to leverage data to target specific brand benefits at a customer. Open-loop systems like Apple Pay (its tied to your bank account) and closed-loop systems like Starbuck’s purchasing app (funds are pre-loaded and can only be redeemed at Starbucks) can be found everywhere. Even Snowflake, the data warehouse provider and subject of last month’s TBOTM is a closed-loop payments network. Customers buy Snowflake credits up-front, which can only be used to redeem Snowflake compute services. In contrast, AWS and other cloud’s are beginning to offer more open-loop style networks, where AWS credits can be redeemed against non-AWS software. Side note - these credit systems and odd-pricing structures deliberately mislead customers and obfuscate actual costs, allowing the cloud companies to better control gross margins and revenue growth. It’s fascinating to view the world through this open-loop / closed-loop dynamic.
New Kids on the Block - What are Stripe, Adyen, and Marqeta? Stripe recently raised at a minuscule valuation of $95B, making it the highest valued private startup (ever?!). Marqeta, its API/card-issuing counterpart, is prepping a 2021 IPO that may value it at $10B. Adyen, a Dutch public company is worth close to $60B (Visa is worth $440B for comparison). Stripe and Marqeta are API-based payment service providers, which allow businesses to easily accept online payments and issue debit and credit cards for a variety of use cases. Adyen is a merchant account provider, which means it actually maintains the merchant account used to run a company’s business - this often comes with enormous scale benefits and reduced costs, which is why large customers like Nike have opted for Adyen. This merchant account clearing process can take quite a while which is why Stripe is focused on SMB’s - a business can sign up as a Stripe customer and almost immediately begin accepting online payments on the internet. Stripe and Marqeta’s API’s allow a seamless integration into payment checkout flows. On top of this basic but highly now simplified use case, Stripe and Marqeta (and Adyen) allow companies to issue debit and credit cards for all sorts of use cases. This is creating an absolute BOOM in fintech, as companies seek to try new and innovative ways of issuing credit/debit cards - such as expense management, banking-as-a-service, and buy-now-pay-later. Why is this now such a big thing when Stripe, Adyen, and Marqeta were all created before 2011? In 2016, Visa launched its first developer API’s which allowed companies like Stripe, Adyen, and Marqeta to become licensed Visa card issuers - now any merchant could issue their own branded Visa card. That is why Andreessen Horowitz’s fintech partner Angela Strange proclaimed: “Every company will be a fintech company.” (this is also clearly some VC marketing)! Mastercard followed suit in 2019, launching its open API called the Mastercard Innovation Engine. The big networks decided to support innovation - Visa is an investor in Stripe and Marqeta, AmEx is an investor in Stripe, and Mastercard is an investor in Marqeta. Surprisingly, no network providers are investors in Adyen. Fintech innovation has always seen that the upstarts re-write the incumbents (Visa and Mastercard are bigger than the banks with much better business models) - will the same happen here?
Building a High Availability System. Do Mastercard and Visa have the highest availability needs of any system? Obviously, people are angry when Slack or Google Cloud goes down, but think about how many people are affected when Visa or Mastercard goes down? In 2018, a UK hardware failure prompted a five-hour outage at Visa: “Disgruntled customers at supermarkets, petrol stations and abroad vented their frustrations on social media when there was little information from the financial services firm. Bank transactions were also hit.” High availability is a measure of system uptime: “Availability is often expressed as a percentage indicating how much uptime is expected from a particular system or component in a given period of time, where a value of 100% would indicate that the system never fails. For instance, a system that guarantees 99% of availability in a period of one year can have up to 3.65 days of downtime (1%).” According to Statista, Visa handles ~185B transactions per year (a cool 6,000 per second), while UnionPay comes in second with 131B and Mastercard in third with 108B. For the last twelve months end June 30, 2020, Visa processed $8.7T in payments volume which means that the average transaction was ~$47. At 6,000 transactions per second, Visa loses $282,000 in payment volume every second it’s down. Mastercard and Visa have always been historically very cagey about disclosing data center operations (the only article I could find is from 2013) though they control their own operations much like other technology giants. “One of the keys to the [Visa] network's performance, Quinlan says, is capacity. And Visa has lots of it. Its two data centers--which are mirror images of each other and can operate interchangeably--are configured to process as many as 30,000 simultaneous transactions, or nearly three times as much as they've ever been asked to handle. Inside the pods, 376 servers, 277 switches, 85 routers, and 42 firewalls--all connected by 3,000 miles of cable--hum around the clock, enabling transactions around the globe in near real-time and keeping Visa's business running.” The data infrastructure challenges that payments systems are subjected to are massive and yet they all seem to perform very well. I’d love to learn more about how they do it!

Business Themes

What is interchange and why does it exist? BigCommerce has a great simple definition for interchange: “Interchange fees are transaction fees that the merchant's bank account must pay whenever a customer uses a credit/debit card to make a purchase from their store. The fees are paid to the card-issuing bank to cover handling costs, fraud and bad debt costs and the risk involved in approving the payment.” What is crazy about interchange is that it is not the banks, but the networks (Mastercard, Visa, China UnionPay) that set interchange rates. On top of that, the networks set the rates but receive no revenue from interchange itself. As the book points out: “Since the card netork’s issuing customers are the recipients of interchange fees, the level of interchange that a network sets is an important element in the network’s competitive position. A higher level of interchange on one network’s card products naturally makes that network’s card products more attractive to card issuers.” The incentives here are wild - the card issuers (banks) want higher interchange because they receive the interchange from the merchant’s bank in a transaction, the card networks want more card issuing customers and offering higher interchange rates better positions them in competitive battles. The merchant is left worse off by higher interchange rates, as the merchant bank almost always passes this fee on to the merchant itself ($100 received via credit card turns out to only be $97 when it gets to their bank account because of fees). Visa and Mastercard have different interchange rates for every type of transaction and acceptance method - making it a complicated nightmare to actually understand their fees. The networks and their issuers may claim that increased interchange fees allow banks to invest more in fraud protection, risk management, and handling costs, but there is no way to verify this claim. This has caused a crazy war between merchants, the card networks, and the card issuers.
Why is Jamie Dimon so pissed about fintechs? In a recent interview, Jamie Dimon, CEO of JP Morgan Chase, recently called fintechs “examples of unfair competition.” Dimon is angry about the famous (or infamous) Durbin Amendment, which was a last-minute addition included in the landmark Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010. The Durbin amendment attempted to cap the interchange amount that could be charged by banks and tier the interchange rates based on the assets of the bank. In theory, capping the rates would mean that merchants paid less in fees, and the merchant would pass these lower fees onto the consumer by giving them lower prices thus spurring demand. The tiering would mean banks with >$10B in assets under management would make less in interchange fees, leveling the playing field for smaller banks and credit unions. “The regulated [bank with >$10B in assets] debit fee is 0.05% + $0.21, while the unregulated is 1.60% + $0.05. Before the Durbin Amendment the fee was 1.190% + $0.10.” While this did lower debit card interchange, a few unintended consequences resulted: 1. Regulators expected that banks would make substantially less revenue, however, they failed to recognize that banks might increase other fees to offset this lost revenue stream: “Banks have cut back on offering rewards for their debit cards. Banks have also started charging more for their checking accounts or they require a larger monthly balance.” In addition, many smaller banks couldn’t recoup the lost revenue amount, leading to many bankruptcies and consolidation. 2. Because a flat rate fee was introduced regardless of transaction size, smaller merchants were charged more in interchange than the prior system (which was pro-rated based on $ amount). “One problem with the Durbin Amendment is that it didn’t take small transactions into account,” said Ellen Cunningham, processing expert at CardFellow.com. “On a small transaction, 22 cents is a bigger bite than on a larger transaction. Convenience stores, coffee shops and others with smaller sales benefited from the original system, with a lower per-transaction fee even if it came with a higher percentage.” These small retailers ended up raising prices in some instances to combat these additional fees - causing the law to have the opposite effect of lowering costs to consumers. Dimon is angry that this law has allowed fintech companies to start charging higher prices for debit card transactions. As shown above, smaller banks earn a substantial amount more in interchange fees. These smaller banks are moving quickly to partner with fintechs, which now power hundreds of millions of dollars in account balances and Dimon believes they are not spending enough attention on anti-money laundering and fraud practices. In addition, fintech’s are making money in suspect ways - Chime makes 21% of its revenue through high out-of-network ATM fees, and cash advance companies like Dave, Branch, and Earnin’ are offering what amount to pay-day loans to customers.
Mastercard and Visa: A history of regulation. Visa and Mastercard have been the subject of many regulatory battles over the years. The US Justice Department announced in March that it would be investigating Visa over online debit-card practices. In 1996, Visa and Mastercard were sued by merchants and settled for $3B. In 1998, the Department of Justice won a case against Visa and Mastercard for not allowing issuing banks to work with other card networks like AmEx and Discover. In 2009, Mastercard and Visa were sued by the European Union and forced to reduce debit card swipe fees by 0.2%. In 2012, Mastercard and Visa were sued for price-fixing fees and were forced to pay $6.25B in a settlement. The networks have been sued by the US, Europe, Australia, New Zealand, ATM Operators, Intuit, Starbucks, Amazon, Walmart, and many more. Each time they have been forced to modify fees and practices to ensure competition. However, this has also re-inforced their dominance as the biggest payment networks which is why no competitors have been established since the creation of the networks in the 1970’s. Also, leave it to the banks to establish a revenue source that is so good that it is almost entirely undefeatable by legislation. When, if ever, will Visa and Mastercard not be dominant payments companies?

Dig Deeper

July 2020 - Innovator's Dilemma by Clayton Christensen

This month we review the technology classic, the Innovator’s Dilemma, by Clayton Christensen. The book attempts to answer the age-old question: why do dominant companies eventually fail?

Tech Themes

The Actual Definition of Disruptive Technology. Disruption is a term that is frequently thrown around in Silicon Valley circles. Every startup thinks its technology is disruptive, meaning it changes how the customer currently performs a task or service. The actual definition, discussed in detail throughout the book, is relatively specific. Christensen re-emphasizes this distinction in a 2015 Harvard Business Review article: "Specifically, as incumbents focus on improving their products and services for their most demanding (and usually most profitable) customers, they exceed the needs of some segments and ignore the needs of others. Entrants that prove disruptive begin by successfully targeting those overlooked segments, gaining a foothold by delivering more-suitable functionality—frequently at a lower price. Incumbents, chasing higher profitability in more-demanding segments, tend not to respond vigorously. Entrants then move upmarket, delivering the performance that incumbents' mainstream customers require, while preserving the advantages that drove their early success. When mainstream customers start adopting the entrants' offerings in volume, disruption has occurred." The book posits that there are generally two types of innovation: sustaining and disruptive. While disruptive innovation focuses on low-end or new, small market entry, sustaining innovation merely continues markets along their already determined axes. For example, in the book, Christensen discusses the disk drive industry, mapping out the jumps which pack more memory and power into each subsequent product release. There is a slew of sustaining jumps for each disruptive jump that improves product performance for existing customers but doesn't necessarily get non-customers to become customers. It is only when new use cases emerge, like rugged disk usage and PCs arrive, that disruption occurs. Understanding the specific definition can help companies and individuals better navigate muddled tech messaging; Uber, for example, is shown to be a sustaining technology because its market already existed, and the company didn't offer lower prices or a new business model. Understanding the intricacies of the definition can help incumbents spot disruptive competitors.
Value Networks. Value networks are an underappreciated and somewhat confusing topic covered in The Innovator's Dilemma's early chapters. A value network is defined as "The context within which a firm identifies and responds to customers' needs, solves problems, procures input, reacts to competitors, and strives for profit." A value network seems all-encompassing on the surface. In reality, a value network serves to simplify the lens through which an organization must make complex decisions every day. Shown as a nested product architecture, a value network attempts to show where a company interacts with other products. By distilling the product down to its most atomic components (literally computer hardware), we can see all of the considerations that impact a business. Once we have this holistic view, we can consider the decisions and tradeoffs that face an organization every day. The takeaway here is that organizations care about different levels of performance for different products. For example, when looking at cloud computing services at AWS, Azure, or GCP, we see Amazon EC2 instances, Azure VMs, and Google Cloud VMs with different operating systems, different purposes (general, compute, memory), and different sizes. General-purpose might be fine for basic enterprise applications, while gaming applications might need compute-optimized, and real-time big data analytics may need a memory-optimized VM. While it gets somewhat forgotten throughout the book, this point means that organizations focused on producing only compute-intensive machines may not be the best for memory-intensive, because the customers of the organization may not have a use for them. In the book's example, some customers (of bigger memory providers) looked at smaller memory applications and said there was no need. In reality, there was massive demand in the rugged, portable market for smaller memory disks. When approaching disruptive innovation, it's essential to recognize your organization's current value network so that you don't target new technologies at those who don't need it.
Product Commoditization. Christensen spends a lot of time describing the dynamics of the disk drive industry, where companies continually supplied increasingly smaller drives with better performance. Christensen's description of commoditization is very interesting: "A product becomes a commodity within a specific market segment when the repeated changes in the basis of competition, completely play themselves out, that is, when market needs on each attribute or dimension of performance have been fully satisfied by more than one available product." At this point, products begin competing primarily on price. In the disk drive industry, companies first competed on capacity, then on size, then on reliability, and finally on price. This price war is reminiscent of the current state of the Continuous Integration / Continuous Deployment (CI/CD) market, a subsegment of DevOps software. Companies in the space, including Github, CircleCI, Gitlab, and others are now competing primarily on price to win new business. Each of the cloud providers has similar technologies native to their public cloud offerings (AWS CodePipeline and CloudFormation, GitHub Actions, Google Cloud Build). They are giving it away for free because of their scale. The building block of CI/CD software is git, an open-source version control system founded by Linux founder Linus Torvalds. With all the providers leveraging a massive open-source project, there is little room for true differentiation. Christensen even says: "It may, in fact, be the case that the product offerings of competitors in a market continue to be differentiated from each other. But differentiation loses its meaning when the features and functionality have exceeded what the market demands." Only time will tell whether these companies can pivot into burgeoning highly differentiated technologies.

Business Themes

Resources-Processes-Value (RPV) Framework. The RPV framework is a powerful lens for understanding the challenges that large businesses face. Companies have resources (people, assets, technology, product designs, brands, information, cash, relationships with customers, etc.) that can be transformed into greater value products and services. The way organizations go about converting these resources is the organization's processes. These processes can be formal (documented sales strategies, for example) or informal (culture and habitual routines). Processes are the big reasons organizations struggle to deal with emerging technologies. Because culture and habit are ingrained in the organization, the same process used to launch a mature, slow-growing market may be applied to a fast-growing, dynamic sector. Christensen puts it best: "This means the very mechanisms through which organizations create value are intrinsically inimical to change." Lastly, companies have values, or "the standards by which employees make prioritization decisions." When there is a mismatch between the resources, processes, and values of an organization and the product or market that an organization is chasing, its rare the business can be successful in competing in the disruptive market. To see this misalignment in action, Christensen describes a meeting with a CEO who had identified the disruptive change happening in the disk-drive market and had gotten a product to market to meet the growing market. In response to a publication showing the fast growth of the market, the CEO lamented to Christensen: "I know that's what they think, but they're wrong. There isn't a market. We've had that drive in our catalog for 18 months. Everyone knows we've got it, but nobody wants it." The issue was not the product or market demand, but the organization's values. As Christensen continues, "But among the employees, there was nothing about an $80 million, low-end market that solved the growth and profit problems of a multi-billion dollar company – especially when capable competitors were doing all they could to steal away the customers providing those billions. And way at the other end of the company there was nothing about supplying prototype companies of 1.8-inch drives to an automaker that solved the problem of meeting the 1994 quotas of salespeople whose contacts and expertise were based so solidly in the computer industry." The CEO cared about the product, but his team did not. The RPV framework helps evaluate large companies and the challenges they face in launching new products.
How to manage through technological change. Christensen points out three primary ways of managing through disruptive technology change: 1. "Acquire a different organization whose processes and values are a close match with the new task." 2. "Try to change the processes and values of the current organization." 3. "Separate out an independent organization and develop within it the new processes and values that are required to solve the new problem." Acquisitions are a way to get out ahead of disruptive change. There are so many examples but two recent ones come to mind: Microsoft's acquisition of Github and Facebook's acquisition of Instagram. Microsoft paid a whopping $7.5B for Github in 2018 when the Github was rumored to be at roughly $200M in revenue (37.5x Revenue multiple!). Github was undoubtedly a mature business with a great product, but it didn't have a ton of enterprise adoption. Diane Greene at Google Cloud, tried to get Sundar Pichai to pay more, but he said no. Github has changed Azure's position within the market and continued its anti-Amazon strategy of pushing open-source technology. In contrast to the Github acquisition, Instagram was only 13 employees when it was acquired for $1B. Zuckerberg saw the threat the social network represented to Facebook, and today the acquisition is regularly touted as one of the best ever. Instagram was developing a social network solely based on photographs, right at the time every person suddenly had an excellent smartphone camera in their pocket. The acquisition occurred right when the market was ballooning, and Facebook capitalized on that growth. The second way of managing technological change is through changing cultural norms. This is rarely successful, because you are fighting against all of the processes and values deeply embedded in the organization. Indra Nooyi cited a desire to move faster on culture as one of her biggest regrets as a young executive: "I’d say I was a little too respectful of the heritage and culture [of PepsiCo]. You’ve got to make a break with the past. I was more patient than I should’ve been. When you know you have to make a change, at some point you have to say enough is enough. The people who have been in the company for 20-30 years pull you down. If I had to do it all over again, I might have hastened the pace of change even more." Lastly, Christensen prescribes creating an independent organization matched to the resources, processes, and values that the new market requires. Three great spin-out, spin-in examples with different flavors of this come to mind. First, Cisco developed a spin-ins practice whereby they would take members of their organization and start a new company that they would fund to develop a new process. The spin-ins worked for a time but caused major cultural issues. Second, as we've discussed, one of the key reasons AWS was born was that Chris Pinkham was in South Africa, thousands of miles away from Amazon Corporate in Seattle; this distance and that team's focus allowed it to come up with a major advance in computing. Lastly, Mastercard started Mastercard Labs a few years ago. CEO Ajay Banga told his team: "I need two commercial products in three years." He doesn't tell his CFO their budget, and he is the only person from his executive team that interacts with the business. This separation of resources, processes, and values allows those smaller organizations to be more nimble in finding emerging technology products and markets.
Discovering Emerging Markets.
The resources-processes-values framework can also show us why established firms fail to address emerging markets. Established companies rely on formal budgeting and forecasting processes whereby resources are allocated based on market estimates and revenue forecasts. Christensen highlights several important factors for tackling emerging markets, including focusing on ideas, failure, and learning. Underpinning all of these ideas is the impossibility of predicting the scale and growth rate of disruptive technologies: "Experts' forecasts will always be wrong. It is simply impossible to predict with any useful degree of precision how disruptive products will be used or how large their markets will be." Because of this challenge, relying too heavily on these estimates to underpin financial projections can cause businesses to view initial market development as a failure or not worthy of the companies time. When HP launched a new 1.3-inch disk drive, which could be embedded in PDAs, the company mandated that its revenues had to scale up to $150M within three years, in line with market estimates. That market never materialized, and the initiative was abandoned as a failed investment. Christensen argues that because disruptive technologies are threats, planning has to come after action, and thus strategic and financial planning must be discovery-based rather than execution-based. Companies should focus on learning their customer's needs and the right business model to attack the problem, rather than plan to execute their initial vision. As he puts it: "Research has shown, in fact, that the vast majority of successful new business ventures, abandoned their original business strategies when they began implementing their initial plans and learned what would and would not work." One big fan of Christensen's work is Jeff Bezos, and its easy to see why with Amazon's focus on releasing new products in this discovery manner. The pace of product releases is simply staggering (~almost one per day). Bezos even talked about this exact issue in his 2016 shareholder letter: "The senior team at Amazon is determined to keep our decision-making velocity high. Speed matters in business – plus a high-velocity decision making environment is more fun too. We don't know all the answers, but here are some thoughts. First, never use a one-size-fits-all decision-making process. Many decisions are reversible, two-way doors. Those decisions can use a light-weight process. For those, so what if you're wrong? I wrote about this in more detail in last year's letter. Second, most decisions should probably be made with somewhere around 70% of the information you wish you had. If you wait for 90%, in most cases, you're probably being slow." Amazon is one of the first large organizations to truly embrace this decision-making style, and clearly, the results speak for themselves.