Wave hunting

June 2020

If you’re optimizing for impact, choosing a material problem matters far more than your personal effort and your ability to execute.

I don’t know where I heard the analogue, and it’s not perfect, but it’s stuck with me as a way to think about choosing problems to work on. It goes something like;

“if you had to bet on two swimmers in a race, and one was an exceptional swimmer in calm waters, and the other was an average swimmer riding a tsunami, which would you bet on?”

In case it’s unclear, a tsunami carrying an average swimmer will win 10 times out of 10 in a race with even Michael Phelps. This resonated as I thought of what didn’t work at my last startup and where we could have done better. We spent 4 years building an incredible amount of things and doing a (in my opinion) better than average job of micro-optimizations to make the product and business work. We read all the things, worked all the hours, spent all our savings and it still didn’t matter.

I think our biggest headwind was working on something that didn’t matter, in a domain that wasn’t growing in any macro sense.

I spent 2014 through 2019 working on Cash App, and I think we out-executed for the most part and did (and still do) a good job. I also think Square/Cash App rode (and are still riding) a few massive waves that cut across rapid technology adoption, new technologies becoming widespread that weren’t before, and discontinuous changes in the economic structure of financial services. A large domain with lots of growth and macro change is a force multiplier for strong execution. Example waves (from my time on Cash) are below, but they’re meant to be illustrative.

Time to money

The money movement system in the US has long imposed a tax on the most vulnerable. Most money in the US moves via wire, followed by ACH - the ACH network processed $55.8 trillion in volume in 2019. By comparison, Visa processed $11 trillion in volume in 2018 (actually this volume also moved over the ACH network - Visa was merely the messaging system for these payments). From an end user perspective, money moved via ACH moves relatively slowly (1 - 2 business days, not at all on weekends) relative to money moved by the card networks (instantly for all intents and purposes). There’s a lot to unpack here which I won’t get into for the moment. In terms of utility, consumers in the US know that they cannot depend on money moved via ACH in an emergency. This means that there’s always been a high value placed on being able to move money instantly. In the past this was wires; most banks would charge you $25 to wire money to another bank, and it would only work if both banks were open.

This meant that the slowness in the ACH system was a tax on the most vulnerable Americans - if you’re living paycheck to paycheck and something catastrophic happens, and you need to access to funds really quickly, you’re either paying a really high fee or its just not happening for you. And heaven help you if it’s the weekend.

Over the last few years, the card networks and NACHA have invested heavily in making it possible for consumers to access their money faster. There have been several iterations of innovation here.

Next Day ACH

Square & Stripe productized this in merchant payments first; before they came along, most small businesses would get access to funds 5 - 7 days after processing a credit card payment. Square and Stripe gave you your money within a day or 2. There is some element of credit risk to doing this, but the risk is embedded in the system (Square/Stripe would be out of the money if the system failed) rather than an attribute of the merchant (ie, whether the merchant was a bad actor or bad credit, had no impact on whether they should receive funds the next day).

Instant Push to Debit

The next iteration was making the instant push-to-debit transaction type widely available. Cash App was the first to do this broadly for consumer, but everyone from Venmo, Apple Cash, Paypal, Dave and several others have since followed suit. This iteration took a transaction type that had existed for a while but wasn’t really applied in consumer use cases (if I remember correctly, the transaction type was typically used for to push gas station refunds or something of that nature. Don’t quote me on this). The transaction type essentially enabled a merchant acquirer to push funds onto a card with a realtime message, and the SLA on making funds available was functionally instant (sometimes could take up to 30 minutes, but the card issuer would make the funds available for the customer to use generally right away). Prior to the 2010s, this transaction type existed on some payment networks some times and was only sometimes honored by some banks. Over the latter half of the decade, a lot was done to standardize and distribute it for broad, straightforward use, primarily starting with Cash App. Now, you can access it via Stripe instant payouts: https://stripe.com/docs/payouts , and lots of companies use it for lots of things.

Mobile & impact on distribution

There’s tons of writing on the internet about the impact of mobile on various industries. The rise of mobile phones (and the app store in particular) had (and are continuing to have) a pretty transformational effect on the banking industry. Over the last hundred odd years, in order to succeed as a bank you needed the best distribution, and the way to win in distribution was to have branches near the most potential customers. The winners over the last hundred years (Bank of America, Chase, and Wells primarily) built and acquired their way to victory, by building a massive branch network.

As mobile has saturated the market, there’s become a new way to win distribution. Every consumer has the functional equivalent of a “branch” in their pocket; the App Store/Google Play. Branch in this context means an access point for a large variety of services - deposits, credit cards, mortgages, investments, retirement products and more are already available in the App Store today.

The analogue for foot traffic is App Store visits. By being a top 20 app, consumers are just more likely to come across you when going about their business, it’s an impression you don’t have to earn, and as a result when they are having a problem that you solve, you benefit from being there, and the cost of trying you is relatively low.

This new way of winning distribution is quite different from the old way. Yes it’s also capital intensive, but you’re not buying real estate. You have low fixed costs, and variable costs that scale with usage (vs. real estate servicing that you pay regardless of use). And where branch-banking might have had winner-take-all dynamic regionally, there is now at least the possibility that it has a winner take all dynamic nationally; it’s not as though banks in Kansas and banks in Alaska did fundamentally different things for the last 100 years. They did mostly the same things, and just differentiated on literal geography

What’s been missing is the depth. Even today, for instance, there really isn’t a mobile first banking product that has the depth of financial features that Chase provides. That’s changing though, and tons of challenger/neobanks have increasingly broad and rich feature sets. For what it’s worth, I also think this same dynamic is playing out around the world and it’s one of the waves being ridden by startups like Monzo, N26 and others.

Square is riding this wave, as is Cash App. I think Chase, Wells and others realize mobile is a big opportunity, but I don’t think they realize it might be existential.

Dodd Frank and the impact of the Durbin amendment

Roughly speaking, the Durbin amendment bounded how much large banks (above $10b in assets) could earn off debit card interchange/swipe fees. For the largest banks, interchange was the primary way they monetized consumers who lived paycheck to paycheck. If you live paycheck to paycheck, you have high volatility in your income and expenses and you generally don’t keep a balance in your bank account, so the bank relies on interchange fees to profitably service you.

Because Durbin capped interchange for larger banks, they all of a sudden could no longer profitably service a large proportion of Americans (you’ve heard this group commonly referred to as the underbanked or unbanked). This change hit the market circa 2011, and had a few material effects.

Reducing the cost of processing

Merchant processors like Stripe and Square charge merchants a % of receipts and optimize costs under the hood. The biggest banks in America, which make up ~80% of the debit cards in the country, were most impacted by Durbin. The amount they earned on interchange dropped materially, and because more consumers spend on debit than credit, the cost of processing debit card payments for the biggest banks, dropped for companies like Square and Stripe, improving their economic profiles in a stroke.

Forcing large banks to rely on overdraft fees

Because of this hit, big banks began charging and raising overdraft fees. For a consumer who doesn’t keep a balance, overdrafts became so expensive that it simply didn’t make sense to keep an account with a bank that would charge those. This created two opportunities: 1. Products like Dave, Status, Bridget and Earnin have emerged to provide overdraft protection, and 2. Products like Cash, Chime, Venmo and others have emerged to provide free checking (or checking like) accounts.

Enabling challenger banks

A secondary effect of Durbin; big banks like Chase and Wells could no longer profitably service consumers living paycheck to paycheck, and as a result they no longer spent marketing dollars to acquire those customers. This left (and continues to leave) an opening for new entrants - most startups couldn’t compete with Chase if trying to acquire customers head to head. It’s just too expensive. But if Chase is no longer competing for a segment, that left an opening and tons of products have driven through it. Here’s a helpful list that someone created. Many of the products on that list would not exist were it not for the macro changes happening over the last decade.

Problem selection = Hunting for waves

Looking at problems through this lens is pretty helpful. Operating in a large and growing market is a force multiplier on your efforts; you can be average on execution and still land in a pretty good place. Operating in a stagnant or shrinking market is a fractionalizer; you might do ok but your execution needs to be that much tighter, and competition will be fierce, and in the end it still might not matter.

In many ways, an early stage startup’s job is to hunt for waves - they’re running a search function for finding valuable problems, and building solutions for them. I often see people poke fun at founders who pivot; there’s probably something wrong with flailing, but it’s really healthy to be able to discard problems you’ve realized are not sufficiently impactful (or for which the effort spent solving the problem outstrips the return). If you find yourself swimming against a wave, it is really important to stop, because the outcome will be the same; the wave will defeat you, all you choose is when. You can stop now voluntarily or stop when you’ve exhausted your resources. Most big companies are bad at this (discarding bad ideas fast) because in many ways it’s equivalent to losing face.

Today when deciding what to work on, or learning about what friends are working on, or talking to other founders who are optimizing for maximum impact, I push pretty hard on problem selection. The hierarchy I like to use is

  • best case: find a large market with lots of growth and change.
  • Failing that, find a small niche with lots of growth
  • Failing that, find a large market with lots of change
  • Failing that, find a large market
  • Don’t fail that.

Fantastic waves and where to find them

A friend whose been in Silicon Valley for a long time recently commented that a lot of the biggest opportunities left for technology are these truly large domains in the world, that software has only started to touch. He mentioned financial services and healthcare as two where software has only scratched the surface. Extending this, the others I think (all for different reasons) are logistics, freight, construction, natural resources, energy, industrials, and mining. There are probably others I’m not considering, but these are the obvious ones. When hunting waves, you’re more likely to find large, valuable ones, outside the domain of pure, traditional software.

Thanks to Temi Omojola, Nafis Jamal, Charles Birnbaum and Robert Jawl for reading this in draft form.