In The Diff, we'll sometimes talk about companies varying in their "economic surface area." This term isn't in wide use just yet, but it's a surprisingly powerful concept. To start off, it's helpful to think of most economic entities as a layer of abstraction between people, who are the end consumers of goods and services.¹ Everything else is some form of intermediation, which means that every company is in some sense intermediating between households-as-labor-providers, households-as-consumers, and households-as-savers (who invest some of their savings).²

Some companies have to view the broader economy through a narrow keyhole: a local contractor has visibility into macro forces that they don't have full control over, and they can give you anecdotes like "the labor market is looser: I'm actually able to hire non-felons!" or "consumer sentiment is great; the guy whose pool I installed last year is asking me to demolish it so he can replace it with a bigger pool." But they have a limited ability to forecast where these trends are going, other than by extrapolating based on previous patterns. And, given the information channels they have, they aren't well-positioned to make many causal inferences about what they're seeing: if that contractor had a hard time hiring in the summer of 2020, how much of it was because the economy was booming, and how much of it was because direct transfers to households meant that there wasn't as much urgency to work as usual? They probably had a view on this, but it was a view that had to be informed by consuming the same kinds of business and political news that everyone else had access to.

At the opposite end of the spectrum, big ad platforms have incredible visibility into the economy, especially if the ads they're displaying are on some search or social media property. They can see:

• What people are expressing interest in (are they booking vacations to Bali or somewhere closer to home?)

• How responsive they are to different price points in ads

• What the time-of-day pattern looks like—a recession is when adults aren't checking Instagram much 9-5, a depression is when they spend those hours toggling nonstop between Instagram and LinkedIn.

• They can look at advertiser confidence, and look at this by sector.

• Within their advertiser cohort, they can also look at whether companies are doing ads with a shorter payoff cycle (direct-response ads leading to a single purchase) or a longer and intangible one (brand ads that might eventually lead to recurring revenue)

• A social network can even take nuanced readings of the economic climate by looking at people's personal behaviors, like how often they're posting selfies after nights out, or how much their social media behavior implies continuing to be loyal to their current partner (divorce rates are cyclical, though they're probably a lagging indicator).

And these companies don't just have to blindly stare at the data and react: they can look at which variables are leading indicators, which ones lag, and how all of them together predict macroeconomic variables and actually constitute a good working model of the economy.

All of which is, unfortunately, of limited utility for these companies, because there's just not much they can do about all of this. They might adjust their hiring and capital expenditure plans a bit, lock down spending when they expect labor to get more expensive, and give investors more accurate guidance. But beyond that, it's hard to get that much direct use out of the data.

The exception to this is when they have some way to scalably measure economic performance in many domains, and a way to influence it. All of the big tech companies do this to some degree: when Meta decided that short-form video was a real category, they started paying people to upload it. Google saw how beneficial faster Internet connections were, and made their own (probably the biggest impact of Google Fiber is the indirect impact of other companies offering higher-speed Internet in non-Google Fiber markets specifically so Google Fiber wouldn't go there first). Their dominance in one market gave them visibility into another.

The company that's most defined by economic surface area is Amazon. They sell goods. They sell compute, an increasingly universal complement. They sell ads (of almost all kinds). They buy real estate, and use it to deliver their products faster. They've deployed Internet-connected speakers in their customers' homes, they make movies, and, as is fitting for a retailer, they have physical stores (that sell everything, including groceries!). Amazon's business model makes sense as a kind of incremental growth where the increments are a few steps past what a reasonable person would expect, but it's also a good model to look at the Amazon of ten years ago as an intelligence-gathering operation attempting to design the optimal Amazon of today, and for the Amazon of today to be doing exactly the same thing for the Amazon of 2035. Amazon has visibility into bottlenecks throughout its business, and can continuously expand into the most profitable ones.

It's always easy to speculate about future economic configurations, and in a sense every new company is a bet on what those future arrangements will be. But the more economic activity a company sees, and the more visibility it has into the underlying drivers of that activity, the more informed its bets will be.

Disclosure: Long AMZN, META.

For more in The Diff on economic surface area and its implications, see:

• This writeup of adtech company Criteo ($) notes that its effective take rate is basis points compared to the volume of economic activity it sees.

• Platforms sometimes conflict with other platforms, partly over who gets the most visibility.

• Sometimes, the right approach is making data-rich interaction a continuous default ($).

• Here's a look at how Amazon considers whether or not to let competitors buy ads ($).

Join the discussion!

1  To do this right, you need a very broad sense of "consumption." For example, where does a nonprofit that funds public art projects fit in? It's an employer, it buys capital assets, but it also exists as a form of consumption for 1) the people who enjoy whatever art it produces, and 2) whoever decided that a monetary donation produced more happiness for them than the money would if it were put to some other use. The second is more action-guiding, not because it matters more—plenty of nonprofits produce desirable goods whose exact utility is hard to measure—but because the donor is the person whose decision ultimately catalyzes the economic activity that the nonprofit is responsible for.

2  This concept is nicely illustrated in macro-economics classes through the circular flow diagram.