Why The Answer Is So Often Ads

Plus! Strategic Shutdowns; Verification, Ads, and Membership; The China Cap Table; Very Long-Term Investing; Branding a Bank; Diff Jobs

Byrne Hobart Byrne Hobart

Why The Answer Is So Often Ads

Of the ten largest companies in the world by market capitalization, two, Alphabet and Meta, are primarily in the business of selling ads. Two others, Microsoft and Amazon, have significant ad businesses that are meaningful contributors to long-term growth assumptions. And now Apple, the largest company in the world by market cap, is also experimenting with ads ($).

That should be surprising. Over the last century, ad spending has typically fluctuated in a surprisingly narrow range of 1.0% to 1.4% of GDP. Total war and financial crises are the only events that seem to push it to the lower end of that range, whereas the invention (or perfection) of radio, television, direct response mailings, classifieds, cable TV, online display ads, in-feed ads, interstitials, search, influencer marketing, and even newsletter ads have somehow not pushed that number up much. This isn’t the case for other industries; the consumer price index in the 1930s was 35% food and 11% apparel, compared to 13.5% food and 2.6% apparel today. (And we're consuming more calories and wrapping the resulting bodies in many more varieties of clothing than we did back then; the deflation in clothes over time is staggering.[1])

And yet, fortunes are being created; companies go from dorm room projects to testifying before Congress about their overwhelming dominance in less than a generation, powered by a business that, over the last hundred years, hasn't grown its share of spending.

This naturally leads to some discomfort. As Jeff Hammerbacher put it: "The best minds of my generation are thinking about how to make people click ads. That sucks." And it does! But you don’t have to see it that way; one useful division of the economy goes like this:

  1. There's making stuff.
  2. There's selling it—i.e. matching physical stuff to the buyer most likely to want it.
  3. There are services-other-than-ads.

As a general sketch, service businesses tend to see slow gains in productivity, while manufacturing businesses see high gains. But a corollary of this is that service businesses that are complements to manufacturing can ride on the coattails of higher manufacturing efficiency, at least indirectly. If you measure the value-add of a car dealership in terms of how many vehicles it sells compared to how long employees work, you will see gradual change. But if you compare the labor input from the dealership to the number of safe vehicle-miles driven, and the cost thereof, dealerships have gotten far better at adding value to the economy.

There's no field that better illustrates this than software, because it's a complement to transistors, which have seen insane improvements in productivity over long periods: most of the discrete objects humans have manufactured in our history are transistors, and the continuous drop in their cost has continually increased the value of labor that's complementary to cheap memory and processing power.[2]

Businesses don’t exist in a vacuum, and especially when we’re talking about the long-term drivers of productivity—i..e the drivers of economic growth beyond just deploying known technologies at bigger scale—it’s useful to put things in a macro context. In a global economic system with fluctuating exchange rates, a natural consequence of high growth in one export sector is a higher currency value, which tends to make other exports unattractive. This has a name, "Dutch disease," and while it's usually better than the alternative, it does need to be managed. In the US, there are actually two large and competitive export-driven sectors, one of which, the production of reserve currency, is unrelated to software (except in the sense that, like so many other industries, its processes have been heavily automated thereby). But the world's demand for dollars means there's global demand for Americans to consume.

And this environment happens to be very conducive to tech companies that monetize through ads: there's abundant consumer demand in the US for products that the US has a comparative advantage in consuming but not producing, i.e. anything that can create a dollar-denominated revenue source in a country that can't print dollars. The growth of global manufacturing in the last thirty years, mostly in China, has led to massive deflation in the price of physical goods, and has made the manufacturing sector more agile.

So ad-driven software companies in the US have, depending on how you count, two or three separate complements that are all subsidizing their growth: more diversity in products people could want, creating a more challenging problem matching supply to demand (and creating more rewards for matching them[3]; persistent global demand for dollars means both that there's demand for the US to consume more than it produces, and that other sectors that would otherwise want the software industry skill-set are priced out of exporting and thus can't afford the talent; and the hardware industry keeps making it cheaper to store and search vast amounts of data. Meanwhile, the usual economics of software businesses—high fixed costs, low marginal costs—mean that the same services that get proven out in the US can expand to other parts of the world where fewer of these tailwinds apply.

If ad spending as a share of GDP hasn't moved up much, but the market value of ad-driven companies is up a lot, it presents a few different possibilities: the market might be overestimating the potential of these models (but the P/E ratios of big tech companies really aren't much of a stretch given their historical growth); the market might be pricing in unprecedented growth in ad spending as a share of GDP (it doesn't take much extrapolation to get there, but nobody values Google based on a top-down model of GDP, ads' share of that, and Google's share of that); or, the ad industry could be a lot better than it used to be at cost-effectively matching supply and demand—it could be so much better, in fact, that it can support multiple high-margin intermediaries, all of whom actually report artificially low margins because they also invest heavily in unrelated R&D projects.

There are plenty of other reasons why the US was where many of these ad-based companies started—the US had a strong tech sector back before ads were a major driver, of course, but when that existing comparative advantage collided with new sources of growth, it accelerated the ad business more than all the rest. Which makes it interesting to read things like this piece on non-ad models for chatbots and look at just how many of them turn out to be ads after all: monetizing chatbots through APIs means asking the API buyer to figure out a model (I have a guess!); plugins tend to be bottom-of-the-funnel features for demand-aggregation companies; and licensing data once again pushes the problem back to the end user who buys the data and figures out how to monetize it. When there's a new product that simultaneously captures lots of attention and ingests lots of data, it's solved the two key problems ad platforms need to solve: how to get attention, and how to direct it to the right commercial message. There are, of course, plenty of other monetizable features of large language models; they're great for searching and summarizing, two things businesses will actually pay for as complements to their existing work. But given how lucrative the ads business has been for the companies that go all-in on it, the gravitational pull is strong here, too.

Just as there are enterprise search products that don't generalize the way Google does (but are much smaller), there will be plenty of specialized services that leverage AI, as there already are. But don't be surprised if the average person's experience of a chatbot in a few years is that it's a free product, it's consumer-friendly, and when appropriate it finds answers to users' questions that happen to entail making a purchase.

Disclosure: Long META, AMZN, and MSFT.

  1. Before the Industrial Revolution notes that clothing was so expensive during the renaissance that a major problem for hospitals was the theft of clothes from the deceased, and that this contributed to the spread of plague. ↩︎

  2. Which raises an interesting question: if chips have gotten so cheap, and there are so many more programmers, and programming-related fields have set such a high price on talent that the industry at least get a chance to hire anyone who is qualified and even slightly motivated by money—why is the average software experience so far from perfect? One possibility is that we're running through a society-wide version of Brook's Law: adding people to the project of digitizing the economy makes that project later. As the share of software built to interact with other software approaches 1, there will be a higher premium for extremely well-considered design decisions, but an industry geared towards growth is not designed to make those decisions naturally. ↩︎

  3. If you ever want to see this in action, look for a product on Amazon and then track it down on Alibaba and compare the prices. That price gap is partly Amazon's own vig for expediting the purchase, but the rest of the gap is basically the amount a manufacturer pays for a distributor/retailer/marketer to find ways to get the product to the specific end customers who want it. ↩︎

A Word From Our Sponsors

Elevate your investment research with Tegus

Designed to keep you ahead of the curve, our research platform empowers you to source new opportunities, map markets, and pressure test your investment theses at a fraction of the cost of other tools.

What sets us apart?

With custom-sourced expert calls at an average of $300 per call and the flexibility to customize call lengths, deep private company data that spans seed stage to mature companies and robust benchmarking, charing and comps, Tegus delivers the competitive edge you need to make informed decisions with confidence.
Trial Tegus today for free and experience the power of streamlined investment research.

Elsewhere

Strategic Shutdowns

Thursday's subscribers-only post was on why Netflix chose to shut down, rather than sell, their DVD business ($), and the core argument was that Netflix timed this so it coincided with cracking down on password-sharing so it could convert more DVD-subscription customers into streaming viewers. Further evidence for this has since emerged: Chicken Soup for the Soul Entertainment, the parent company of Redbox, has repeatedly offered to buy Netflix's DVD business, and been turned down. Presumably Chicken Soup can't offer an amount that equals the net present value of the new streaming subscribers Netflix will get if the original business just disappears.

Verification, Ads, and Membership

Twitter continues its hectic series of experiments with paid verification, and is now asking all advertisers to buy verified accounts, too. Making sense of every decision Twitter has made under Musk is an exercise in frustration, but this one does actually have some logic to it: replacing some incremental spending with recurring spending increases the amount of money those recurring spenders will optimally spend. And giving ad-buying users better features can keep them around on the site. For advertisers, it's turning the product from a pure ad product to a bundle of traffic plus extra features, with all the economic benefits that such bundles entail.

The China Cap Table

Foreign bondholders of Chinese property companies are getting less than they hoped for in restructurings ($, Economist). Finance is always an iterated game, where one party's openness to opportunistically one-sided deals is tempered by the fact that they might be on the receiving end some time. Unfortunately for this particular set of investors, the prospects for outside capital to be allowed into China, either by home governments or by China itself, are dimmer than they were when the bonds were initially issued. When that happens, it turns from an infinitely repeated game to a game with a finite number of repetitions that's approaching or has reached zero, and incentives change accordingly.

Very Long-Term Investing

A new investment club, subsidized by the CEO of value investing darling Markel, is requiring participants to lock up their investments for 25 years ($, WSJ). This kind of thing is usually proposed as a thought experiment (here's The Diff's, on why it would be hard to run an endowment for the Long Now). What it ends up selecting for is growth and stability, but there are two ways to go about this: one option is to identify a narrow set of companies that are hard to disrupt, reasonably well-run, and have good internal mechanisms for choosing a successor (at that time interval, it's not enough to know something like "the COO seems competent, and will be taking over in a few years—there's a nonzero chance that the CEO at the end of the period in question will join the company as an entry-level employee after the purchase is made!). The other option is to find a larger set of companies that have gigantic addressable markets and some reasonable shot of attaining them.

Valuation certainly matters a bit in both cases, but the converse of the fact that it fluctuates a lot more than fundamentals in the short term is that it can't contribute much in the long term. If a company's P/E collapses by half over the course of the investment, that's a 2.7% headwind to annual returns, which is nothing to sneeze at, but it is dwarfed by the long-run dispersion in how much companies can grow and how durable their margins are.

Branding a Bank

First Citizens, the bank that acquired SVB, is struggling to integrate it ($, FT) while deposits and personnel flee. Part of SVB's value was the brand name, but part of the value of that brand name was that it attracted a certain kind of banker, who developed relationships with particular companies and funds. (The funds are an important part of the model, since SVB has some nice products that helped funds report higher internal rates of return by delaying capital calls from LPs, and because founders will weight VCs' advice about financial matters highly when they're making decisions. First Citizens presumably wasn't aiming to buy a pool of complicated assets at a discount and then profitably liquidate them, but it's easier to buy a balance sheet than to acquire and retain talent.

The Diff previously covered the question of what makes a bank worth more than its book value last month, and also recently wrote about First Citizens ($).

Diff Jobs

Companies in the Diff network are actively looking for talent. A sampling of current open roles:

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

Find a Role

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.

Find Talent with Diff Jobs