When you start learning about how companies work and what role they serve in the economy, you'll typically start with the simplest possible company: a handful of inputs, one output, which they hopefully sell for an amount that exceeds input costs by a sufficient margin to produce an economic profit. And there are many companies that fit this model today: copper mines make money from digging up and selling copper, retailers mostly buy goods and sell them for a markup, real estate companies mostly make their money from charging rent, etc.

And then there's a trickier category. Stockbrokers used to make their money from commissions, but depending on the model their big profit centers are either some net margin on interest spreads (i.e. paying less than they earn on cash balances, charging much more than their cost of funding for margin loans) or payment for order flow. The actual transactions matter, but their monetization is indirect. Oil companies do largely make their money from drilling for, extracting, moving, refining, and retailing oil and related products. But some of them also make a healthy slice of their profits from commodity trading: owning a big chunk of the supply chain gives them a reason to hedge, and it also gives them plenty of opportunities for quick profitable trades. This is sometimes a material profit contributor, but it isn't the main thing they do.

But then there are companies whose business model bears little resemblance to their core competencies. Google is good at doing search, but also good at selling ads—the vast majority of products they make are free to end users, and lucrative because Google sells other companies traffic from those users. Carmax loses money on the business of selling cars; last year, 118% of their net income came from their financing segment. Grocers, too, sometimes end up with a business model where selling food and household products is a slightly unprofitable activity, but selling shelf space to the producers of those products is almost pure profit and makes up for it. Non-low-budget airlines are, notoriously, a business whose loyalty programs are sometimes worth not just more than the airline business, but more than 100% of the total value of both.

There are also companies that try to do this, and sometimes fail  badly. PayPal's original plan was to make money on the float—i.e. the interest you get between when someone sends you $20 because you picked up the check and when you take the $20 out of PayPal. As it turns out, this period was not especially long, at least early on.¹ The early cruise ship industry had some exciting adventures with float; Carnival was founded by the then-CEO of Norwegian, because his investors told him to stop making speculative bets with the money passengers had put down as deposits. His response was to lend the deposits to a new company: Carnival. Starbucks gives us another fun example of float as financing. The float generated by purchased, yet unused Starbucks gift cards is in effect an interest free loan, which the company uses to fuel working capital needs and other investments. Better yet, some percentage of the gift card value (typically 6-8%) never gets redeemed—serving as a permanently “free” cash injection.

One thing the successful business-model-isn't-the-real-business companies have in common is that one element of their offering is closer to a traditional product/service, and another piece is either ads or financial services. The latter businesses are both sometimes opaque, and they have a similar relationship with marketing leads: in financial services, leads are incredibly expensive, so running a car dealership that's mostly in the business of getting good prospects for auto lenders works great. The lenders' business is highly commoditized—nobody's going to buy a worse car because they really want an Ally loan—and that means that it's expensive for them to use advertising to differentiate themselves. (Some of the world's most valuable fictional characters sell intangible financial products, specifically insurance: the GEICO Gecko, Progressive’s Flo, Allstate's Mayhem.) And ad businesses are, in a way, a kind of lead arbitrage: Google doesn't have to pay nearly as much to acquire a customer as they earn from that customer's searches, especially if the free product remains good.

Another thing these companies have in common is that, as tempting as it might be, there is no good way to separate the two businesses. If there were—if you could spin off Carmax the high-margin lead gen business from Carmax the negative margin auto retailer—that's what investors would want. But in practice, it is just one business, and splitting them up breaks the entire formula that makes them work.

The existence of these companies actually illustrates an economic principle that's usually applied to governments rather than the private sector. The government-focused version is "the economic incidence of taxation is independent of the legal incidence," in other words, whether you tax cigarette companies, cigarette retailers, or smokers, what you're really taxing is the purchase of cigarettes, and each party will adjust according to its own elasticity. No matter who pays in an accounting sense, the tax creates a wedge (equivalent to the tax’s value) between what the consumer pays and what the producer gets, and that wedge affects both sides' incentives. But that's really a special case of the general point that elasticities determine who it's optimal to charge and what it's optimal to charge for. Google as a business wouldn't work if you paid whatever their average revenue per search is for an ad-free experience, because you'd be thinking about money every time you thought about a search. It’s better to make search free and to monetize through advertisers, who were thinking about money all along. The same applies to the grocers: Procter & Gamble's willingness to pay for shelf space as a function of expected revenue is more rational than customers' willingness to rapidly switch between brands based on tiny price differences, so P&G pays to make the consumer's decision even easier. What all of these companies are doing is taking a multifaceted economic process and putting the complexity (and costs) in front of whoever's best equipped to handle it.

Read More in The Diff

We’ve looked at these business models from a variety of angles in The Diff, including:

• We’ve examined float-based business models ($).

• Airlines can be viewed through multiple unit economics lenses.

• Back in 2020, United did a very interesting maneuver where they borrowed against their loyalty business, not the overall business ($).

Here’s a quick Diff profile of Starbucks ($).

Join the discussion!

1  How much of this was consumers actively optimizing their liquidity in order to maximize interest income, and how much of it was because in the 90s it still seemed like a very bad idea to trust an Internet company, especially one with a catchy, consumer-friendly name, with their savings? Float-based businesses can work just fine; the insurance industry is built on the returns from invested float. But it works best in circumstances where people treat that company's obligations as money-good and are either getting some future benefit from their current cash outlay or are putting an immaterial sum to work.