Who Controls Distribution: The Power of Leverage

From Owning Distribution:

Companies can invent, manufacture, ship, and sell products. But the end result of all of this is selling to the end user, and the party that touches the product last has the most influence. Some companies are content to provide internal components that customers have never heard of, because they compete on cost in a commoditized industry, or they sell something that those customers can’t get anywhere else. (Nvidia and Qualcomm are good examples of this in hardware: most end users don’t realize they’re interacting with these companies’ products, but if those products are essential to the interaction, that doesn’t hurt pricing power.)

Much of competition and strategic decision-making between companies comes down to owning distribution. Making a product people like and being the company people like because of that product are two completely different things. This is especially obvious in food delivery. Both platforms and restaurants are constantly working on quality control, in part because it’s hard to tell who is at fault when the fries show up late, cold, and soggy: is it Baemin’s fault for delivering late, or the restaurant’s fault for cooking too early or handing the food to the courier too late? They can’t solve this problem perfectly (getting consistent performance from a fragmented network of mostly high-turnover independent couriers, as Baemin does, would be a logistical miracle). But there are tools they can use to mitigate it: couriers tend to use insulated bags to keep food at a steady temperature, restaurants can package food better or remove items that don’t travel well from the delivery menu. The UI that appears after a customer places an order is, first, an attempt to reassure the customer and, second, an ongoing effort to shift blame when something goes wrong: you can see if the food was ready before the courier arrived, you can see if the courier took a detour, and if the order is left somewhere odd or inconvenient there will be a photo record.

This is one example of a more general problem: companies have multiple paths to reach the end customer, and even when they’re reaching the exact same customer, they have to decide how much economic upside and control they’re willing to give up to get there. In Baemin’s case, its commission rate last quarter was 9.8%, but that’s just the direct cost. Restaurants also have to think about whether joining a delivery service will, over the long term, shift customer preferences toward eating at home. In a one-off decision, it’s better to pay a 9.8% fee and collect 9,020 won out of 10,000 from a customer who otherwise wouldn’t have gone out to eat at all, rather than getting zero. But over the long run, that may amount to agreeing to an indefinite margin squeeze. And the more restaurants rely on delivery in general, and on a single platform like Baemin in particular, the harder it is to say no when commissions go up. (Or, equivalently but more subtly, when platforms increasingly weight search results toward ad revenue rather than customer satisfaction.)

Some companies navigate a world of multiple distribution channels quite well. Apple is both a participant in, and a perpetrator of, this world. Beyond its own stores, Apple sells iPhones through many retailers like Hi-Mart, and it probably gets favorable terms from them. That’s because Apple knows that 1) people will visit specific stores just to buy an iPhone, and can be persuaded to pick up higher-margin products like cases or headphones along with it, and 2) many loyal customers will find a way to buy an iPhone even if Hi-Mart doesn’t carry it.

At the same time, Apple imposes this exact multi-channel dilemma on companies that use the App Store. Netflix knows that the App Store is not the only reason people watch Netflix, and that if they really want to, they can watch shows in a mobile browser. Initially, Netflix concluded that Apple was entitled to (or could at least demand) a cut in exchange for letting people sign up for Netflix using the payment information they’d already registered with Apple. But they probably didn’t like this arrangement very much. The main “service” the App Store distribution channel provided was just pulling forward a transaction that was likely to happen anyway. A few years later, Netflix stopped letting new users sign up and pay through the App Store.

In the short term, companies want to own as many distribution channels as they can. The fixed cost (depreciation) of building a distribution channel can be much cheaper than the commissions paid to a seller, especially when the product is digital, with high fixed costs and low marginal costs. As they get closer to being market-dominant, and as the platforms they use start competing with them, it becomes rational to spend less on distribution.

But all of this implicitly assumes that we know what the main distribution channels are. In reality, we don’t. The U.S. Department of Justice fought hard to keep computer operating systems free from monopolistic influence, but that effort took so long that by the time it wrapped up, search engines and browsers had become the primary distribution tools people used. Even that can change as new platforms emerge and older ones become less important than the applications they enabled. Owning distribution matters, but understanding where distribution is headed is essential to knowing what’s actually worth owning.

The recent payment-default crises at Tmon and WeMakePrice, SK Square’s decision to walk away from its call option on 11st, and Emart’s strategic failure in acquiring eBay Korea all serve as fresh reminders that the essence of the e-commerce market is not discount rates, coupons, or user counts, but the physical distribution network.

Back when Coupang was still racking up cumulative losses in the trillions of won, I was one of the few who predicted it would turn profitable, and what I emphasized was that most e-commerce customers put the delivery experience first. The e-commerce ecosystem consists of producers, sellers, intermediaries, shippers, and end consumers, and is highly fragmented.

E-commerce platforms that spent the mid-to-late 2010s luring customers with discounts, coupons, and aggressive celebrity-driven marketing are now bankrupt or on the brink, while Coupang, which invested aggressively in logistics infrastructure, has turned profitable and is posting record results. Coupang, in other words, made the economically rational choice for the long term.

Marketing variable costs like discounts and coupons, which scale with gross merchandise value, may boost short-term sales and user numbers, but they have several effects that undermine the platform’s long-term economics. As the earlier essay argued, joining a delivery platform can temporarily lift a restaurant’s sales, but over time it can shift consumers’ dining preferences toward delivery and cause an indefinite erosion of margins. In a similar vein, a strategy of acquiring customers through discounts and coupons may lead to a temporary bump in revenue, but over the long run it turns price-sensitive shoppers into the platform’s core user base and permanently compresses margins. It is a fundamentally uneconomic strategy.

By contrast, an e-commerce strategy of investing in transport and delivery networks is highly economical over the long term. We already know that consumers care most about the delivery experience when they shop on e-commerce platforms (see revealed preference theory), so we can infer that, in generating demand, such investments are at least as effective as, if not superior to, discounts, coupons, or TV ads featuring celebrities. On top of that, they allow the company to enjoy the benefits of higher marginal revenue and lower marginal cost as the user base grows.

For example, a customer who likes Coupang’s delivery experience enough to open the app is highly likely to add a few more items to their cart while they’re at it (expanding marginal revenue). And as the user base grows, the probability that Coupang will deliver to multiple households in the same apartment complex rises (reducing marginal cost). If it costs Coupang 5,000 won to deliver to one customer in unit 301 of building A, the extra cost of delivering to a customer in unit 302 is close to zero—because it’s the same driver, in the same truck, burning the same fuel.

Using Coupang’s vast Rocket Delivery network and its easy-to-use app—these tangible and intangible assets—to generate additional value is what we call “operating leverage”.

If you want to succeed, you need to find ways to extend your strengths as far as possible at the lowest possible cost. And if you’re a thoughtful investor, you should look for companies that combine strong economic and technological moats with operating leverage. Naturally, the AWARE portfolio is always invested in such companies.