A few weeks ago, we considered what happens when companies discover that there's one metric they need to be laser-focused on maximizing. Sometimes it's total users, or time spent, but sometimes it's a quirkier one—there was a point in Microsoft's history where they realized that their top priority should be squashing bugs and addressing security vulnerabilities, and Amazon's priorities changed when there was a top-down mandate to make the different parts of the business connect to one another as if they were third-party services, because that's increasingly what they would be.

But there's a similar process when public markets cotton on to the right way to evaluate a company, and end up monomaniacally focused on a single metric or a cluster of related ones. In the early 2010s, two of the great tech growth stocks were Netflix and the company now known as Booking Holdings (formerly Priceline), and for neither of them was the most market-moving datapoint a GAAP metric. For Netflix, "the number" usually referred to some combination of their net new subscribers in the current quarter and their guidance for the next.¹ In Booking's case, the metric was their guidance for the next quarter's FX-adjusted growth in gross hotel bookings. The reason guidance mattered was that they reported fairly late in the quarter, so they knew a lot about the next quarter's number already. And the reason the focus was on a gross bookings number rather than a revenue take rate or a profitability metric was that they had good-enough metrics there already and the big question was how big they could get, and how quickly.²

For both companies, the metric investors cared about was the one that answered two surprisingly similar questions. For Netflix, the question was whether their model could really scale to every country, once it had been proven in the US. A bigger Netflix would have more negotiating power in buying media assets, could amortize these content costs over more viewers, and had the opportunity to experiment with local shows that could be popularized in other markets—like using viewership data for the original House of Cards to underwrite a remake, or seeing that Squid Game and Money Heist had global appeal. So the number was an abstract way of measuring the extent to which their expansion in Latin America implied similar future success in Europe and Asia. For Booking, it was a similar setup: the core Booking thesis was that the hotel market was fragmented but travel search was increasingly consolidated: it probably started on a search engine, then went to an online portal that would display a wide selection in a given area; since there was a correlation between how many viable options a travel site could show and how much revenue they'd get per click, the biggest tended to grow and the smallest had to sell or die. But the hotels weren't standing still; they, too, were a form of travel portal, and what they lacked in breadth (there are only so many Hilton-affiliated properties, at a finite number of price points), they made up in brand and trust, loyalty programs, credit card points, and the rest. So gross bookings growth was really a proxy for how much runway Booking had with unaffiliated hotels before they had to switch to knife-fighting the big chains over take rates.

When investors settle on one true metric, what it enables is a division of labor. Pod shops tend to look for short-term catalysts, and they're implicitly taking the general valuation model for a business as a given and then trying to precisely estimate the inputs. When there's only one main input, that creates demand for ways to measure it, and companies that trade off of one key variable tend to be the most profitable ones for alternative data providers to track. As it turns out, there are many indirect metrics you can use to track these businesses, whether they involve web scraping, clickstream data, tracking app downloads (the Netflix app is kind of useless unless you're paying for it), etc. So what they're doing is underwriting incremental execution risk, from a slow quarter, a weak rollout, a marketing campaign that didn't land, etc. Meanwhile, their implicit counterparty is a longer-term investor who's asking bigger questions—trying to understand how much pricing power Netflix will have once streaming is the default way to consume video, or figuring out whether hotels will get good at distribution faster than online booking platforms figure out loyalty programs. Another category of questions longer term investors ask target the premise of a company’s value proposition itself (sometimes called terminal value questions). For example, whether hotel discovery and booking will move from search and travel portals to chat interfaces and AI agents or whether high-fidelity, low cost AI-generated content would make an aggregator and distributor of premium content obsolete. 

This is a very happy relationship until the story changes. Sometimes, that's because of a one-off change in the market that completely changes the usual pace of compounding (Netflix and Booking.com both got one of those, in opposite directions, in 2020). Sometimes the narrative shifts towards margins, because investors have a line-of-sight on market dominance and start considering exactly what it's worth. And sometimes, it's because the model that made that indicator so important starts to break down—if a company is only rewarded for growth, the market is telling it to overspend on marketing, but eventually investors will start to ask whether that growth is worth it or not.

This disruption shows up in stock prices, but it's more extreme than whatever has changed fundamentally. For one thing, a stock is lower-risk, for all parties, if there's one main metric and there are large market participants who are closely tracking it. The fast-money participants know that longer-term investors are valuation-sensitive, and will tend to show up as buyers if the stock drops too much based on temporary weakness. Those real-money investors recognize that they're somewhat insulated from buying the right stock in the wrong quarter, because Citadel, Balyasny, Millennium, and all the rest are constantly competing to accurately price every incremental reading from the data. But when each side can't rely on the other to keep things efficient, everyone's optimal position size is lower, and the stock has to produce higher returns to compensate—which is a gentle way to say that the stock price has to go down.

Often, this is fundamental to what the company is doing, and can show up in the data even when the narrative works. Netflix was easier to track when the only way people got it was putting their credit card number into a website. Once they could subscribe through iOS, that had to be tracked, too, and it had to be incorporated into a model that just didn't have that many clean quarterly datapoints. Add in bundles from third parties, or access through smart TVs, and the story gets messier.³ For someone like Booking.com, this confusion might take the form of tradeoffs, where Booking accepted slower growth in the short term for a more loyal customer base, or shifted their money from direct response to brand advertisements in order to raise the returns of future direct response spending.

It's hard for a company to stay easy to analyze for too long. There can be magical years where it's riding the steepest part of the S-curve for one distribution channel that's trackable, and where the investors who've been tracking it the longest can trade not just the fact that their data predicts fundamentals but that data neophytes make predictable mistakes. But the reward for making lots of comparatively easy money is to be analyzing a company that's grown into something harder to understand.

The Number is a running sub-theme in many of The Diff’s discussions of hedge funds.

• We’ve covered the world of alternative data ($) and the challenges of monetizing it ($).

• We’ve examined Netflix pre-pandemic and post ($).

• Sometimes the same number can be read as bullish whether it’s going up or down ($).

• You’ll get most of the data you need to make an important business decision after you’ve made it.

Join the discussion!

1  They went through a long period where they kept insisting that they were producing their best guess, but kept blowing through that estimate. They did start missing eventually, so it's possible that they were repeatedly taken by surprise by how well this worked.

2  Booking definitely sandbagged; there was a long period where the way to get their FX-neutral number for the next quarter was to take the growth guide, add exactly seven points, and then adjust it by 1-2% in either direction based on whatever data you had. If you were trying to make money on a Booking.com earnings print, it was because you felt that you had an edge predicting the guidance, not the quarter.

3  Interestingly enough, the company is often getting better-informed over time about how to predict its business. Netflix has some idea of how viewing behavior and churn vary based on the device being used to access their service, and they can compare that against the cut that they'd give some third party for distributing their product. So one of the paradoxes of growth companies is that management's information advantage over investors improves even as there's more money riding on investors having good information.