The Business of Wallets

What Exactly is a Wallet?

Nobody carries a leather wallet anymore. OK, that's an exaggeration - some people do. But a huge and growing chunk of consumers have ditched physical wallets entirely. Their IDs, their cards, their cash equivalents - all of it lives on a phone now, or shoved into a phone case as a backup. The industry saw this coming ages ago, and for years companies have been jockeying to get the most valuable bits of people's financial lives into electronic systems that stick. They called the customer-facing piece of those systems "electronic wallets." Which, predictably, got shortened to just "wallets."

The term has bounced around a lot. Product teams at various companies have tried to one-up each other, or to argue that what they're building is really wallet-adjacent, or a superset of wallet functionality, or something grander. Fine. But it is worth stepping back and just looking at what actually got pitched - and sold - as a wallet around the world.

One clarification that became necessary around 2024: crypto folks hijacked the word. In their world, a "wallet" means "a collection of private keys controlled by a single entity," and then more specifically "the software that, given an instruction to move money from A to B, assembles that money from separately kept accounts corresponding to a list of private keys and sends the relevant crypto network a candidate transaction for inclusion in the blockchain." Those are not the wallets I'm talking about here. But for anyone deep in crypto, contrasting how crypto wallets make money versus how these wallets make money is genuinely instructive.

The Regulatory Question

Here is something I've watched trip up smart people over and over in fintech: relatively small product questions swing enormous doors when it comes to regulatory and partner complexity. The small question with massive implications for wallets is this: "Can it hold a balance of actual money?" Because if it holds actual money, regulators start squinting at it and thinking it looks an awful lot like a deposit product.

And regulators really, really prefer that deposits stay within the regulated banking sector. The biggest reason is straightforward - they want households' immediately accessible stored funds to be safe and, you know, actually accessible. There's a follow-up reason that's less appreciated outside specialist circles: regulated banks are bound to long lists of consumer protection items at the transaction level, not just the institution level. A lot of abuse in the economy happens in $50 and $5,000 increments rather than multi-billion dollar increments. Regulators sleep better at night knowing that this abuse happens at companies with teams of operators who are actually standing ready to fix things, rather than leaving users to eat losses on their own.

Early Internet: PayPal's Innovation

The granddaddy of all wallets was PayPal. Nearly 30 years ago, in the early days of Internet commerce, a huge number of users were genuinely afraid of typing credit card details into a website. (In some countries, the relevant instrument would have been banking details rather than cards.) But in places like the U.S. and Japan, the physical capability to pay with banking details lagged cards so dramatically that online commerce became synonymous with card use early on. That persisted for decades, though it's no longer strictly true.

The Security Scare

Younger users will not appreciate this, but there were front-page newspaper stories - real, prominent, ink-on-paper stories - that scared large parts of the population into believing that typing a credit card number into any keyboard was basically inviting hackers to clean out your accounts. There was supposedly this vast underground of fake e-commerce sites where criminal enterprises would spend millions building convincing replicas of real stores, all just to harvest credit card numbers.

Was this ever actually a major source of stolen cards? Almost certainly not. Not when those articles were written and not afterwards. But this wouldn't be the first or last time the media convinced itself of something untrue and couldn't find industry insiders willing to leak the SQL queries that would have dispelled their fantasies.

The largest source of stolen credit card information is, and basically always has been, scaled breaches of card issuers or companies that were legitimately presented hundreds of thousands or millions of cards through normal commerce. Organized crime does not outscale capitalism. The threat is when it piggybacks illegitimately on capitalism.

PayPal's Value Proposition

So PayPal's initial pitch was simple: move money from payment instruments to other people on the Internet (by volume, mostly eBay auction sellers) without ever showing those supposedly devious hackers your actual credit card numbers. That way exposure was capped at the single transaction in progress, and hopefully PayPal or the banks could step in if something went sideways.

But then eBay sellers ended up with money sitting in PayPal... and what exactly were they supposed to do with it?

Stocks vs. Flows

There are classically two ways to make money in financial product innovation. You can charge customers for stocks - ongoing, often percentage-based fees to hold assets for however long they want. Or you can charge them for flows - per-transaction fees that sometimes scale with size. Very frequently, from a single user's perspective, one is priced and the other is free. In basic bank accounts, stocks are priced (via interest rate spreads on deposits) but transactions are free or close to it. In credit card processing for businesses, stocks are essentially free but flows - incoming customer transactions - get priced.

So do wallets with embedded cash make money from stocks or flows? Both. The answer is both.

PayPal's Enduring Genius

Here's where it gets clever. Moving money into PayPal's ecosystem cost substantial amounts via card interchange. PayPal would set its pricing - historically about 2.9% plus about 30 cents - such that it sat above most interchange fees they'd absorb on the way in. Then they would strongly encourage users to keep balances within PayPal. And this was natural for many casual eBay sellers who were also buyers. If you're not a professional antique dealer, just someone selling old stuff and buying other old stuff, leaving your Internet money somewhere on the Internet until you next need money on the Internet works out fine.

When those customers did their next transactions - paying from PayPal balances to other PayPal users - money moved internally at database transaction speed and cost. Approximately instantaneous. Too cheap to meter. Then PayPal would try, with varying degrees of success, to charge the full 2.9% plus 30 cents. But this time they were making 290 basis points of margin rather than 80-120 basis points or so.

Supercharging margin possibilities by disintermediating card ecosystems is the primary economic advantage of wallets capable of holding balances. I keep coming back to this point because non-specialists routinely assume that things looking like bank deposits - and might actually be bank deposits under the hood - must earn revenue the way bank deposits earn revenue: via net interest margin. And they do, sure. In high-interest environments like the current U.S. market, this is lucrative. PayPal earned nearly $500 million in interest on customer balances last year. Most of that is very high margin revenue. But it is not the ballgame. The real prize is having vastly better economics on transactions.

The Math Per Account

Zoom into single-account economics and this becomes obvious. PayPal claims accounts on average have about 40 transactions per year. Assuming a typical consumer might carry $200 in balance and do 10 transactions each about $40 per year, that is something like $10 of margin from net interest and either about $1 of margin from transactions if transactions are funded with the most expensive credit cards - or about $15 of transaction margin if they're repeatedly funded by PayPal balance. A fifteen-fold difference.

A subtlety worth sitting with: how do you do $400 in transactions out of a $200 balance without extending credit? You convince consumers to let you top up their balances via ACH pulls or similar methods, which (unlike card transactions) are very close to free.

What Does an ACH Pull Cost?

Depends on a lot of things. Negotiating savvy, volume, who your bank partner is. But if someone guessed "maybe 5 cents when you're buying hundreds of millions of them," they'd be in the right ballpark.

If you go check what payment processors charge businesses for their own ACH pulls, quoted numbers land closer to 30 cents. This should surprise nobody - payment providers are in the business of earning margin on underlying payment rails, the same way Coca-Cola is in the business of earning margin on combinations of water, carbon dioxide, and corn derivatives. Coke also gets structurally better prices on corn because Coke buys the agricultural output of Iowa while you and I buy a few ears from the supermarket.

Many readers live in countries where interbank transfers don't go over ACH rails. In those nations as well, interbank transfers are broadly speaking much less expensive than most card transactions.

Extensions: Debit Cards

After running wallet operations for a few years, something predictable happens. Product teams start clamoring that they've got attractive ideas to make wallets stickier, get the best customers to move more transactions through the ecosystem, and decrease payment costs. They'll sometimes make incidental revenue too.

The Obvious No-Brainer

First up: debit cards backed by wallet balances, with cards issued by banking partners that (in the U.S.) are certain to be Durbin-exempt institutions. Product teams will argue this is a total no-brainer. And frankly, they're right. Currently, when users want to pull money out of ecosystems back to bank accounts, wallet providers are paying for that privilege. Why not flip the script and earn money instead? Get co-branded debit card offerings going, and banking partners will provide revenue shares on every transaction out. (Exact amounts are something of a trade secret, but thinking somewhere in the 100 basis points ballpark gets you close enough to pass a PM interview at a fintech company.)

Cash App is, if you squint, a wallet. Block's product managers would probably dispute this characterization heavily. They'd say, truthfully, that their core user feels like they're using Cash App rather than their linked Bank of America debit card when they pay with Cash App. But economically? It looks an awful lot like a wallet. And Cash App has what amounts to a debit card - users can take it anywhere in the economy that doesn't accept Cash App directly. Every time they do, Sutton Bank earns swipe fees, and portions of those become Cash App transaction revenue. These payments are reasonably assumed to be very large chunks of their $498 million in revenue in that segment.

Monetizing Payouts

Sometimes money gets trapped inside ecosystems and users need it out in a hurry. A common Cash App use case: you text someone requesting a small amount, or you ping roommates for their share of rent. Now you've got a few hundred bucks sitting in Cash App and you need to actually use it somewhere. Cash App will happily facilitate this, for 0.5-1.75% with a 25 cent minimum, depending on exactly which rails the money travels over.

Say your landlord requires rent payment via Zelle or check. You need to send money to an account at a bank that can handle both. Two options. Push it over via ACH, which typically arrives in a few days. Or use what is technically a quirky form of "reversing transactions that never happened" on debit cards, which arrives almost instantly.

Cash App likely charges 0.5% (25 cent minimum) if they push over cheaper ACH rails and 1.75% (25 cent minimum) over the more expensive debit card rails.

Credit Card Partnerships

Why only let people spend money they already have when they could spend money they might not have yet? Co-branded credit card offerings are the logical next step. Wallet providers will not themselves issue credit cards - no, regulators consider consumer credit issuance an exclusive domain of the regulated financial sector. So they partner with members of that august community, and those partners issue cards which simply happen to have wallet provider names printed on them.

Why Do This?

Because it lets customers spend money that doesn't exist in the ecosystem, inside the ecosystem, and instead of paying to move money in, wallet providers get paid coming and going. Users charge purchases to cards (for which co-branding partners pay fees) and then money moves to some business through the wallet (earning another fee). The wallet provider is also highly likely to pocket a success fee for every card account successfully opened, which can be used to incentivize users, kept entirely, or split to do both.

The Venmo card is a perfect example. Card issuers have been getting more sophisticated about differentiating offers, which has probably pushed PayPal (Venmo's parent) to get more sophisticated about how they make offers too. Venmo has been willing to pay $200 if applicants apply within 2 weeks and spend $1,000 in the next six months. You can construct ranges of per-account and usage-based payments from Synchrony Bank to PayPal that make this immediately incentive compatible for PayPal.

Negotiated Terms

A fun thing about negotiating issuing relationships: every single company and their dog will claim they have Alternative Data with Big Data levels of transaction history that lets their issuing bank partner underwrite heavy wallet users better than typical prospects. Spoiler: this almost never actually ends up mattering for credit decisions. FICO scores are unreasonably effective. Many teams have thought they could supplement FICO with another data source and get better loss rates. But just about the only additional data sources for which that is actually true are illegal to use.

The big win on loss rates isn't the additive data that wallets bring to the table. It is the negotiated terms between wallet providers and card issuers. Card issuers are not particularly interested in any individual consumer's decision to pay or skip on debts. They're trying to buy portfolios from wallet providers, and the negotiation happens at the portfolio level.

A fairly common term goes something like this: "Across all accounts we issue on the co-branded card, we model a certain loss percentage. Now we're going to construct a graph. If you slightly underperform that loss percentage, we're going to cut your payouts a bit. If you greatly underperform, we cut them by more. And if you somehow have an absurdly fraudulent user base which nonetheless makes it past our own underwriting, we will demand a payment from you."

Phone-Based Wallets

Many wallets exist as apps on phones. But phone makers have had a much bigger realization: phones have replaced leather wallets. Full stop. They now consider the walletness of phones to be as core a product feature as making actual phone calls.

The two most successful wallets in this category are Apple's (more or less coextensive with Apple Pay) and Google's, plus Samsung Pay.

Apple's Pitch to Banks

Apple's pitch to banks was essentially this: "OK, you might claim that many well-heeled users really love their American Express cards. Sure, that sounds reasonable. But do those users spend hours stroking their Amex cards lovingly? Do they pull them out to gaze at while sitting on toilets? No? Well, users care about our plastic-and-glass artifacts much more than they care about yours. Good news though. We can digitize your thing that users yawn about onto the thing that they love, and then it will outcompete every other payment method because it's not buried in their wallets - it is in the palms of their hands basically every waking hour."

This was contentious, to put it mildly. But banks agreed to pay Apple for the privilege of inclusion. Partly because Apple may have said something that rhymed with "If you don't do this, we will find two-sided payment networks willing to work with us. And if we can't find them, we will build one. People will use it, because they like us more than they like you."

Apple has been publicly reported to make 15 basis points on each transaction going over Apple Pay. Google reportedly doesn't make anything for intermediating transactions. If true, that might tell you something about the relative executional capabilities of Apple and Google on non-core products.

The Ads Business Parallel

Lots of scaled Internet companies have looked at these economic models and decided to jump in. And why wouldn't they? The economics are extremely compelling and additive to almost any business that already has to move money around on behalf of users.

This has led to a weird phenomenon: numerous one-button checkout options competing for placement on merchant checkout pages. Which brings us to an interesting parallel.

A Consumer Internet Product Manager's View

Think about what we're looking at here. Numerous different firms vying for user attention. Experiences between those firms that all lead to essentially the same outcome - users bought stuff for prices from businesses. Maybe a bit of user preference, but many users are highly persuadable about which button they click.

Any good consumer Internet product manager will say at this point: "Wait. This smells like an ads business. Do you have any idea how much money ads businesses make? It's absurd."

Now if product managers actually prioritize outcomes for the transacting businesses they represent, they might use prioritization algorithms other than "who paid me the most for placement." Businesses choosing their own ordering might say they really prefer to maximize net margin rather than simply maximize side payments. They're in the business of selling things to people, after all. Checkout pages are just incidental plumbing.

The Optimization Reality

Here's something that surprised almost everyone I've talked to about this: almost no business below the scale of the world's largest companies, and few enough of those, actually has teams running experiments on checkout flows to optimize for conversion rates. I found this striking when it came up through industry conversations over many consecutive years. Just... nobody doing it.

Some providers noticed this gap. They took advantage of rendering checkout flows for huge numbers of customers in parallel and introduced dynamic optimization of payment method presentation. The system uses previous user preference data across geographies, so people transacting in America get prompted with cards like they'd expect, while people transacting in Japan see both cards and popular Japanese payment methods. The conversion rate impact turned out to be much larger than most people guessed before actually running the numbers.

Fast Checkout Solutions

Most wallets exist to make wallet providers money through various revenue streams. But some offerings are designed differently - they're optimized to convert really well and generate network effects for the businesses using them.

The concept is straightforward. Businesses can let customers who begin checkout by sharing identifying information - an email address, a phone number - quickly reuse payment credentials they linked during a previous transaction, at that business or another business using the same system. Since a sizable fraction of businesses use these systems, users are highly likely to have saved payment methods the system already knows about. They just reuse it. No typing card numbers.

The Realization

There's a fun piece of history here. Someone once visited a fintech office and wanted to show off that they'd figured out how to query an internal data store. They asked: "What's your over/under on the percentage of cards used this year that we've seen before?" The number that came back was pretty mindboggling, even years ago. Obviously a killer marketing stat.

Similar products all grew out of the same insight: if a user has typed in card details once, and they're fine with it, they shouldn't have to type them again. That is an obvious user experience win. Businesses like it because it outconverts other checkout methods, and that can be dynamically measured all the time. Providers like it because improvements to conversion rate at the margin mean they collect fees on more transactions.

Another Business Rationale

Now imagine this. Providers' two largest costs are smart people and card interchange. One of those is fun to cut.

Many customers have strong feelings about how they pay for things. They love their credit card rewards. Sure, let them do what they want to do.

But other users don't have strong preferences at all. Maybe they are not in socioeconomic strata that directly benefit from rewards programs, or maybe they just don't care. For those users, businesses that are sensitive to interchange costs can subsidize incentives for typing low-cost payment credentials into their phones once. Those businesses might see repeat custom from those users, and recoup the incentive through lower payment costs over many transactions across months or years.

Think about ride sharing. Core customers ride twice daily, every weekday, for what could be years on end. And every single time the business charges them, it pays issuing banks again for the pleasure of transacting with a customer it already knows. If a ridesharing company could convince those customers to pay with bank accounts instead, the savings over the full length of those relationships would be big enough that the company could just directly pay people for that one simple action.

Then next time those riders check out at other businesses, if they've agreed to save those payment methods... those other businesses passively benefit from lower payment costs. The network effect kicks in without anyone doing anything.

A Scaled Renegotiation

What is really happening here, at the highest level, is a scaled renegotiation between businesses that ultimately pay for payments and the providers of payment services. The logic is pretty intuitive once you see it: more money gets paid for customers who genuinely love certain services. If they really love those services, great - hearts have been won, business will continue to be won. But if customers perceive services as basically interchangeable, then payment services will get bought from commodity providers at commodity prices.

The price of payment services is a contentious topic. Always has been. There's been substantial litigation about credit card interchange fees in the U.S., for example.

One small piece of that: Walmart has opposed a settlement, explaining to judges that it has the economic heft to negotiate serially with the largest banks in the U.S. on interchange and would do so if Visa and Mastercard got out of the way. Most companies do not have anywhere near the scale of Walmart or the operational capability to pick up the phone and engage in months-long bespoke negotiations with one bank, to say nothing of doing that with 20 of them. Walmart makes this point at length in its filings - they don't want the deals most companies would get.

But why shouldn't the Internet get the deals the Internet could negotiate, if the Internet were capable of negotiating on its own behalf? And that is precisely what some providers, as side effects of wallet-like products, are enabling: distributed, techno-social renegotiations of payment costs on behalf of the broader Internet. It is even aesthetically very Internet - not single big bang negotiations ratified by judges, but aggregates of millions of individual, purely voluntary decisions interacting with each other. This brings large numbers of small-to-huge businesses, and the customers using them, to different sides of a table. And given that those sides of the table represent something like percents of global GDP, they tend to get listened to.

Conclusion

The business of wallets is really a story about clever intermediation reshaping financial infrastructure. By holding customer balances and keeping transactions inside closed ecosystems, wallet providers flip payment economics on their head - from paying card networks 2-3% per transaction to paying nearly nothing for internal transfers.

That fundamental advantage - disintermediating card networks - is what drives the entire industry. Interest on balances provides nice supplemental income, sure. But the real prize has always been capturing transaction volume at vastly superior margins. PayPal figured this out first, and now the model is everywhere: Cash App, Venmo, Apple Pay, and dozens of others.

All the extensions - debit cards, credit cards, instant payouts, phone integration - serve the same purpose. They deepen moats and pull more transaction volume into ecosystems. Each additional service makes the wallet stickier, gives users another reason to keep money inside rather than pulling it back to a traditional bank. And for businesses, wallets are a mixed bag. Conversion rates go up when customers can check out fast with saved credentials. But payment costs swing wildly depending on which wallet customers pick and how those wallets route transactions under the hood.

What comes next probably involves continued tinkering with these business models - leveraging transaction data, optimizing checkout experiences, and negotiating better payment economics on behalf of merchant networks. As phones finish replacing physical wallets entirely, whoever controls that digital wallet interface controls some very valuable chokepoints in the payment ecosystem.

And that's ultimately why so many technology companies - messaging apps, ride sharing services, social networks - eventually build payment features. The economics are just too good to walk away from if you're any kind of scaled platform that touches commerce.