This is the once-a-week free edition of The Diff, the newsletter about inflections in finance and technology. The free edition goes out to 17,207 subscribers, up 443 week-over-week.
In this issue:
Just One Thing or Every Single Thing?
Media Disintermediation
Amazon’s Tactical PR, Con’t.
Roll Your Own Green Zone
Just One Thing or Every Single Thing?
In investing, and in life, the more extraordinary results are, the more likely they are to be the result of getting literally one thing right. Google beat Yahoo because Yahoo had everything—sub-portals devoted to news, entertainment, sports, stocks, an email service, and a search engine—while Google spent all its early efforts on exactly one thing, making the best possible search engine. Early Facebook had a minimalist interface compared to other search engines, and a more restrictive approach to identity, but it was the one place online that was most likely to have a digital representation of your real-world friends. In venture and global macro investing, a single call can be career-defining; in 2006, John Paulson was just another member of the 0.1%’s struggling middle class: a very rich person, but not famous, and not someone who was building a durable institution. A few years later, he was the subject of media coverage that included a book called The Greatest Trade Ever.
Even when the outputs don’t follow this extreme distribution, you can step back one level and find a single determining factor. Renaissance Technologies probably can’t attribute the majority of its profits to any one strategy—it’s entirely possible to run a quantitative strategy where no single signal is profitable before transaction costs, but the cumulative results of many simultaneous signals is very profitable. But a company can specialize in recruiting the kinds of people who will invent such signals, and acquiring the many datasets that help train those signals. Steven King and Daniel Steel show, in an entirely different domain, that while the N of publications is large enough that no one book defines their careers, their book production function, with an N of 1, is career-defining.
The “just one thing” logic also works at the more granular level of decisions. In The Startup of You, Reid Hoffman talks about his rule for business trips: make the trip if it would be worth doing for one meeting alone, and then backfill everything else.[1] This is true for other decisions, too; if you have five equally compelling reasons to take a job, quit a job, choose a major, or marry someone, you probably don’t have any good reasons. (A good intuition pump here is to think ten years ahead: you’ll usually beat yourself up over very simple things like “I could have had X!” instead of complicated ones like “I could have had some X, a little Y, and a smidgen of Z, not to mention small doses of P and Q.”)
The weird thing is that, in practice, many of the organizations that focus on exactly one thing tend to demonstrate obsessiveness everywhere. Google beat Yahoo in part because Google was focused and Yahoo was scattershot, but today Google has a profusion of features and sub-products. Yahoo didn’t have an office suite, much less a hot air balloon-based ISP. Stripe wants to increase the GDP of the Internet (and ideally get 2.90% plus 30 cents of as much of it as possible)—books, a magazine, and carbon removal don’t seem like they’d appear this early on the to-do list. It’s not as if Visa publishes books or Paypal as a magazine.
Among investors, I’ve noticed a correlation between a) having encyclopedic knowledge of every detail of a given company, and b) having a very simple thesis for why it’s a good or bad investment. If the bet is “their biggest competitor just stopped discounting” or “the economics of their switch to a subscription model are better than most people realize,” is it really necessary to know how nice the CEO’s new house is, how much unvested stock their new CMO forfeited at the last job, or what product mix shifts led to slightly lower than expected gross margins nine quarters ago?[2]
One hypothesis is that as companies succeed, they get spoiled and lose focus. This certainly happens to some companies, and to some people (CEOs golf more when they get less equity compensation, and their companies perform worse, hedge fund managers who drive fast cars have worse risk-adjusted returns, and both marriage and divorce are associated with worse fund performance). But there’s a wide range of activities that can be distractions in one context and serious business in another. Some CEOs are probably net beneficiaries from all the golfing they do.[3] Some spend undue effort on vanity projects that are nominally related to the business but actually actually more about fun. On the other hand, it’s hard to think of something that sounds more like a vanity project than switching from a cost-effective IP arbitrage of buying DVDs and selling access to them to the much riskier business of streaming videos and making movies, but that’s been key to Netflix’s success. (In early 2012, Vanity Fair illustrated a profile of Netflix’s Reed Hastings with a picture of a burning DVD envelope and $20 bills wafting through the air over Hollywood. The actual VF story is less apocalyptic, but that illustration certainly captures the conventional wisdom at the time.)
The way to square these two worldviews—do exactly one thing perfectly, versus do basically everything—is to work backwards. Getting search, social, payments, or video right would be worth a lot of money. But if there were an obvious way to do it, it would have been done by now. The space of potential megacompanies is the space of big ideas with high execution risk.
And the signature for high execution risk is that the company doing the execution has to make at least a few counterintuitive decisions. It also has to make high-information decisions, in two senses:
Why did Google start offering Gmail? To get more users to perform searches while logged in, so Google could customize their search results better.
Why did Netflix fund House of Cards? To get original content, sure, but also to see what kind of return they’d get from it. If Netflix had built a model for 2019’s 371 original content releases back in 2012, they could plausibly have gotten production costs wrong by 2x in either direction, and there might have been an order of magnitude gap between the low end and high end of the expected financial impact. Every original Netflix produced narrowed the range of expected outcomes for adding the next piece of original content.
So the right way to look at strange diversification with One-Thing-focused companies is to ask: is this one instance of something that will be scaled up if it works or cancelled if it doesn’t? And if the scale is too big for that, the right question is: does this reduce one of the constraints the business faces? Stripe, for example, seems to be an unusually reading- and writing-focused company, and like any company in tech, one of their biggest constraints is recruiting. Publishing books—books that their target employees read, and books whose design means they’ll get prominent shelf placement and get picked up when they’re on a shelf—should be evaluated as a substitute for paying recruiters.
This also explains why companies put so much effort into things that look like incidental side projects. If they’re actually important to the mission, they’re essential to the core product. I’ve noticed a correlation between company success and the CEO’s theoretical ability and actual willingness to do individual contributor-level work well past the point at which it’s economically viable. Bill Gates said in 1997 that he’d last written code that shipped in 1989, when the company was doing $800m in annual revenue. Jeff Bezos says he still reads customer complaints sent to jeff@amazon.com, and forwards them to the appropriate parties. This is, in opportunity cost terms, a massive waste of time, but it seems to be the opportunity cost version of rolling coal, a way to very wastefully signal certain priorities.
For investors, detail obsession is a bit different. A variant market perception has a bit in common with a conspiracy theory: it’s a view that the rest of the world is missing how disparate data points fit together. In some cases, mostly on the short side, it is a conspiracy theory—the most viable conspiracy theories are the extremely boring ones; fewer alien crash-landings, more aggressive revenue recognition. But even optimistic investments have a whiff of conspiracy to them: if a company’s long-term economics are good enough, one of their big risks is that customers, competitors, or regulators figure it out early. These potentially-true conspiracy theories reward in-depth research, because additional data points either confirm or challenge a specific narrative. For any decision management makes, the conspiracy theorist investor can ask: does this decision make more sense in light of what I believe is going on, or less?
The way a business presents itself to the public is a carefully-chosen subset of reality. Some of that choosing is done by rulemaking bodies, or by investigative journalists, but a lot of it is up to the company. And while there’s a general incentive to tell the truth, there are always good reasons to mislead, usually by omission. This is pretty universal—your Instagram profile probably doesn’t offer a high-fidelity summary of what your life is like, although it’s mostly drawn from real life. There’s a U-shaped curve for the rewards to this: the companies that have the most interest in crafting a coherent story are a) the frauds, and b) the dangerously promising companies. There’s a messy middle of companies that don’t have a coherent strategy, or can’t execute on one; these are the companies where an interesting theory can be invented, but it quickly falls apart with more data.
Since reality has a surprising amount of detail, and since consistent and detailed deceptions are hard, it’s essential to ask what interesting story fits all the facts.
[1] Hoffman’s advice is particularly worth listening to because he’s been successful in several domains: working at one important startup, cofounding another, venture investing, and books, at least one of which changed the vocabulary for talking about tech companies. So his advice is probably more generalizable, relative to someone who had more extreme results in a narrower domain. There’s a quote attributed to J. Paul Getty that acknowledges the limits of advice from people who succeeded in just one field: “My formula for success is rise early, work late, and strike oil.”
[2] These are extreme, but they are both real-world examples. In the CEO case, for example, the situation was this: the CEO had held on to his stock after IPO, but years later, with the stock down a lot, he started selling. The company’s investor relations head was asked why, and claimed that the CEO was building a new house. The analyst replied with “If he’s selling at that clip, it must be a pretty nice house.” Further investigation revealed that it was, in fact, a really nice house in a very expensive neighborhood. This was not a fully mitigating factor—a CEO who really believed in the turnaround might have waited to buy his house—but once construction starts, it makes sense that it’s going to get paid for until it’s done.
[3] The right thing to look for is a CEO who a) golfs a lot, and b) is always a little bit worse at golfing than whoever else is playing. An early example of this is that the largest merger in history up to that point was probably arranged due to a golf game.
Elsewhere
Media Disintermediation
Coinbase published a story ahead of a negative New York Times story. The normal corporate communications approach is to either get ahead of the story by refuting it (if it’s clearly wrong) or to get right behind the story by responding to it after it comes out. Coinbase has apparently violated some sort of gentleman’s agreement by “front-running” the story. (It’s very funny to see journalists using securities law terminology to discuss this, since in securities law terms any scoop that comes from a well-informed source is a Reg FD violation. Selectively disclosing information to someone who will benefit—by getting a scoop—is generally frowned upon by the SEC.)
Setting aside the object-level question of what’s in the story, this is interesting as a shift in who controls the narrative. For most companies, it’s hard to go direct; company blogs typically don’t get much traffic, even if they’re widely-read, and it seems like many of them exist as a home for press releases that the company hopes major media outlets will rewrite. But Coinbase has had a direct-to-reader option ever since they released their “mission focus” blog post (which, incidentally, almost certainly catalyzed the NYT piece). Coinbase doesn’t have to hope its press releases get favorably rewritten; they can write press releases people will actually read.
A few other companies have made moves in this direction. Tesla produces slick product updates, but dissolved its PR department. For companies with less visibility, it’s hard to go direct. But, increasingly, it’s possible. And for controversial companies, who know that negative stories are easy to write, it’s optimal. They have more to say, and they benefit more from saying it themselves.
Amazon’s Tactical PR, Con’t.
In yesterday’s issue, I noted that Amazon is collaborating with the US government to stop IP-infringing products, which coincides with the S-1 filing of Wish, an e-commerce competitor with a reputation for looser IP controls. Since then, Amazon has also started helping European merchants register trademarks and protect IP. This could be coincidental, of course; Amazon has been getting IP infringement complaints for years. But it’s hard to explain the timing as anything other than a response to Wish.
Amazon does this in other areas, too. For example, some Amazon warehouse workers are planning a strike today, so Amazon is giving warehouse workers a bonus.
Roll Your Own Green Zone
Creating and enforcing quarantines is a classic externality problem: a noncompliant person gets the benefit of convenience, while the costs of their behavior are mostly borne by everyone else. Companies with monopolistic economics and thorough data are often able to internalize externalities and enforce some socially optimal behaviors on their own. Delta and Alitalia are offering quarantine-free flights between the US and Italy ($, FT) for travelers who are willing to get tested a few days before a flight, and then again on departure and arrival. Airlines have a direct financial stake in higher international travel, and the airlines whose route map skews to a particular pair of countries have a strong incentive to make those country pairs the best two to travel between.
In this case, it’s coincidental that the magnitude of Delta and Alitalia’s benefits happens to support a high level of pandemic control (airlines are a classic high fixed cost business, so incremental travel is very profitable). But the directional tendency is not coincidental at all: the more of a market one company or one set of closely collaborating companies controls, the more of a financial incentive they have to create positive externalities whose upside accrues to their bottom line.
