There are, roughly speaking, two kinds of good businesses in existence:

1. The ones that sell something nobody else can offer, and that can thus be priced in a way that delivers comfortable margins, or

2. The ones that sell something that's widely-available, but at a lower cost than anyone else. For these companies cost structure defines everything they do, and pricing happens in a fairly narrow band.

In a way, these look like two pretty simple situations, where your optimal price is either a tiny increment below the competition (and a tiny bit above your marginal cost) or whatever price you want. But in practice, there's an interesting horseshoe theory effect here, where price-setters and cost-competitive price-takers both have more strategic dimensions than their competitors.

For a high-margin company with a useful product, the problem is that they have a steep S-curve that tops out quickly: if they're solving a million-dollar problem for $100k, whether that problem is "my company spends a million dollars a year on inefficient process X" or "I would happily pay $100k to stand out compared to all the other millionaires." The existence of pricing power implies a problem customers urgently want solved, which implies a market that gets saturated quickly. So high-margin companies that intend to keep growing are always looking for some way to invest in a complement to their core business, so they expand their market.¹ This also creates a new class of owners: the kind of person who builds a unique new product from zero to one is probably not the exact person who's excited to preside over an annual meeting about whether prices should go up 5.5% this year or a round 6%.² And they also worry that their high-margin product might make sense as a complement to someone else's business. No one wants to be in the position of web analytics providers in the early 2000s, who found that there was no business model quite as attractive as "build the best analytics package, give it away for free, and make money on search ads." Or perhaps today, being the system of record complement to someone’s relatively more valuable system of intelligence, until the record increasingly resides in the system of intelligence itself and can be given away for free.

On the low-margin side, cost advantages generally come from some kind of upfront investment, whether it's in building a bigger manufacturing facility or distribution networking, investing in R&D, or the implicit investment behind a corporate culture that's ridiculously fixated on cost-cutting.³ That last one doesn't sound like a real investment, but in practice it is: when a company achieves a superior cost structure through behaviors rather than some underlying advantage, that's built off of senior managers flying economy, brown-bagging lunch, haggling even when it's embarrassing, and otherwise going to inconvenient lengths that aren't worth it on an object level but are worth it if they lead to a company full of compulsive penny-pinchers.

That kind of cost advantage, the boring kind that mostly consists of constantly asking "does this really have to cost so much?" is part of a broader kind of process knowledge, which is some combination of discrete discoveries about ways to improve efficiency and some kind of general improvement in the skill of cutting costs. Continuous process improvement is one of the results of scaling, which means that when companies decide to run some kind of capex blitz in order to achieve unbeatable scale in some domain, part of what they're investing in is the attendant skills everyone will develop while using all that shiny new capital. So they're forced to continuously ask whether, and to what extent, they ought to be betting the entire company on some additional scaling. In many cases, this works well, especially when the product is general-purpose enough that it can be used in many different domains. Energy, compute, and transportation all tend to see higher usage when there's more supply available. (Urban planners like to note that adding lanes doesn't speed up traffic, but another way to say this is that in car-centric cities, people really like to drive, and will do as much of it as possible up to some speed-tolerance set-point.) Which means that, the more general a product is, the more likely it is that the winning company in the category repeatedly built enough capacity to capture more than 100% market share.

As a result of all of this, you can generalize about these companies and the quality of their management teams: high-margin companies picked the right business upfront, which could be skill or could be luck. Low-margin ones have generally picked a bad business, but if they've survived and grown they've made the best of it. On average, lower-margin companies will have better management because they have to, and because the business constantly provides feedback (mostly negative) and managerial skill. These businesses also have a small enough margin of error and big enough base of tacit knowledge/process advantage that they're safer from acquisitions. There are some higher-margin companies with exceptional management, often the ones that come from a time of lower repeat revenue and thus more existential crises just from standing still. But there's a reason private equity likes these companies: you can mess up a lot more and still run them, which means that if someone applies enough leverage to simulate a lower-margin business, they have an incentive to run the company like one.

Disclosure: Long AMZN.

We've covered the varieties of margin in many previous Diff issues, including:

• One of the advantages of low margins is that so many side businesses are margin-accretive ($).

• After 2022, tech reset to more of a margin story with growth as a secondary consideration ($), at least until the AI spending ramp really got going.

• The story of globalization is partly a story of moving annoyingly low-margin activities off the P&L ($).

• One of the questions growth forces you to ask is how many truly huge companies there should be ($).

• Sometimes, the margin picture is complicated because a company's margin structure evolves deal by deal ($).

Join the discussion!

1  The pressure here has changed as more high-margin products, particularly software, have moved to a subscription model. They won't necessarily grow (especially by landing new logos), but they also probably won't die (because they will find some way to be incrementally useful to their captive customer base or, more likely, ways to make switching off their software incrementally more painful, which are both two sides of the same coin).

2  In the case of Salesforce today, it looks like a round 6% is the answer they went with.

3  It can also partly come from business model innovations, two prominent examples are Amazon Prime or the famed Costco membership. A fun fact is that Amazon’s stock actually dropped by 15% the day Prime was announced in 2005. Partly because they missed earnings, but partly because few analysts understood how the company was going to make money offering folks 2 day shipping for free. Amazon understood that having a captive customer base and recurring revenue stream would allow them to make the necessary investments in process, technology, and infrastructure to do just that.