The Gym Business: Usage, Slacking, and the Duel Between Margins and Lifetime Value

Plus! Diff Jobs; Twitter; Remote Work and Neurodiversity; Granularity; Unhealthy SPACs; More Evidence for the Travel Boom; Surcharges

Programming Note: the Diff weekend link roundup is off this week. Back to the regular schedule on Monday.

And: later today I’m doing a Callin show with Augstin Lebron, where we’ll talk about his excellent book and how to apply the mental frameworks of trading to other domains.

The Gym Business: Usage, Slacking, and the Duel Between Margins and Lifetime Value

One of my favorite seasonal charts of search volume is the query "lose weight"1:

There's a spike in January around New Years Resolution season, volume remains fairly high as summer ominously approaches, and, at the end of July, motivation collapses; fall attire does a decent job of hiding extra weight, and typical winter diets give people more to hide.

This trend drives the historical economics of the gym business, where you can break the audience into roughly three categories:

  1. People who sign up in early January and may literally never attend—sources vary, but up to two thirds of gym memberships go unused. For these customers, the closest analogue to gym membership is a lottery ticket: for an upfront cost you can buy a semi-credible fantasy about how different your life might be.
  2. One set of customers drives the main seasonal trend; they're on a roughly annual cycle of losing weight half the year and gaining it back the other half of the year.
  3. The smallest subset in terms of numbers but possibly a majority in terms of time spent, grunts emitted, etc. is the group of people who become absolutely maniacal about fitness.

Classic gym economics relies on the first set of users. The gym needs to balance the large volume of customers who will show up in January (maybe), work out diffidently for a while, and then never come back, against the smaller number of customers who make the whole enterprise worth keeping open in the last half of the year. Annoyingly for gym owners, the more intense exercisers will also tend to be pickier about things like hours of operation and equipment availability. (There's upside to these customers, though: for one thing, they're walking billboards, and the population of a gym during a new customer's tour is not an unweighted selection of customers but a selection of customers weighted by time they spend in the gym, i.e. the average fitness level in a gym at a given time will always be higher than the average fitness level of the gym's members.)

This model has some problems. The obvious one is that it's exploiting people who overestimate their willpower and sign up for a product that is, for them, a very bad deal. That can be unavoidable, and there are plenty of other industries that survive even though many people buy their products at a high point in their motivational cycle. In fact, for purchasing decisions that people consider repeatedly and periodically choose to make, you can think of them as having a winner's curse where the prices are roughly fixed but motivations vary.

The other problem with this model is that it under-exploits its best customers. Any time a business creates some consumer surplus for a subset of users, the natural evolution is towards a) capturing more of that surplus, and b) having a correspondingly larger budget for targeting those customers, both with better marketing and with a better product. Under-exploitation is probably a smaller social problem than over-exploitation, but one way to look at it is that a subset of gymgoers are deriving thousands or tens of thousands of dollars of value from their membership each year, and paying a tiny fraction of that. If they were charged what they were willing to pay, the optimal move for the gym in question would be to offer better services to everyone, in order to capture more of those high-value customers' dollars and to spot more of them early in their fitness-enthusiast careers. It's a bit like the way that high rollers subsidize Vegas buffets, or how the high margins on business- and first-class flights lead to more routes with economy seats, or how commuters ducking into Starbucks on the way to work subsidize a massive network of low-priced coworking venues. (As we've written before, this can also be done in a more literal way.)

That last point is particularly important, because the gym business is partly a real estate business. A gym is many things: a community center, an entertainment venue, or a big box full of exercise equipment and (hopefully) functioning AC and showers. Locations matter. Low-priced gyms can choose out-of-the-way areas and focus on a self-service model to keep costs down, while gyms aiming for high-income customers have to develop expertise in real estate. Not coincidentally, SoulCycle and Equinox, two fitness companies that seem to score well on revenue per square foot, are subsidiaries of real estate firm The Related Companies. This works in two directions: a company with lots of commercial real estate in pricey cities can squeeze in a gym when other clients don't quite fit, and it can use that gym to extract a premium on the rest of its holdings. It's a sort of Georgist business model, where the company benefits from real estate's monopoly characteristics, but also benefits from creating positive externalities.

These premium-priced fitness companies have a very different model from the classic January-focused business: for them, customers are paying enough that it's worth trying to retain them year to year, and one of the better metrics for that is whether or not those customers are actually working out. For heavy users, exercise can be pretty addictive (I mentioned a few months back ($) that Peloton's monthly churn rate of 0.8% is only a little lower than the 0.7% monthly churn rate for smoking, meaning that Peloton is one of the most addictive products of all time.) SoulCycle is vastly more expensive than a gym: a 30-pack of classes is $960, so an every-weekday habit, for example, runs about $8,352/year. And they don't offer much in the way of bulk discounts: one class is $38, and the 30-pack is $32/class.

If you think of higher usage as an implicit discount, then it's entirely plausible that a frequent gymgoer with a standard fixed-cost membership attending every weekday is paying 80% less per session than even a fairly committed person who makes it to the gym every week. The economics for class-based fitness are different from the economics for membership-based gyms because the classes are an attempt to monetize commitment, not over-ambition. It makes sense to give tiny bulk discounts because, on average, the tenth or hundredth SoulCycle class someone attends should be more fun than the average first class, simply because the people who aren't enjoying it will churn out early.

And that, naturally, brings up subscription models. The economics of an in-person class are limited by how many people can be physically crammed into one room and still see the instructor, so it makes sense to price per class. But if a single instructor can work with an audience of hundreds, or tens of thousands, then it's possible to focus the economics on acquiring users and minimizing churn, rather than getting value out of every single class. In-home fitness is a bet that this is doable—that the gym business can simultaneously skew more to entertainment and loyalty rather than guilt, and that it can offload the real estate and equipment capital expenditures onto the customers.

Covid naturally boosted this, and forced other models to adapt (Life Time Fitness, for example, describes itself as omnichannel—and in addition to offering online classes, it's experimented with co-living arrangements; once people aren't bound to a single city, they can still follow around a single fitness brand). Some of the fitness business got unbundled, but the pressure to monetize both the diffident gymgoers and the fanatics was already forcing gyms to operate on a hybrid membership-and-classes model. Now, that's more of a continuum; a gym can think of its business proposition to a single customer as, say, offering video classes for yoga and bodyweight exercises done at home, offering squat racks and weight machines to those customers when they're in the mood to lift, and using in-person classes at the gym as lead generation for more utilization of remote classes elsewhere. If more customers are sticking around specifically because they're using classes, then the physical gym is both a way to reduce churn for less frequent exercisers and a way to grab leads who will convert to higher-margin online customers.

It's odd that growth in the wellness industry can coincide with worse aggregate health statistics: as Robert Solow might have put it, you can see the fitness revolution everywhere but in the obesity statistics.

A lot of this is because of two factors: first, price discrimination as the driver of growth in the fitness industry, and second, fitness as a class marker. For the first, people who love working out have been free-riding on the spending of everyone else; the machines they use exist to convert vague New Year's intentions into annual contracts, and if everyone behaved the way they did the market-clearing price of a gym membership would be higher. The gap between, say, Blink Fitness at $10/month and Equinox at $200/month is some mix of better service and also paying a premium specifically to price some attendees out of doing curls in the squat rack or reviewing their Instagram feed between sets without getting off the machine.

Cardio is also a way to signal status. A theory of the athleisure class is beyond the scope of this piece, but it is interesting that cardio and endurance sports seem to be getting more common in high-status jobs, and have been for a while. (Triathlons started marketing to CEOs when one organizer realized that at a 2001 event, 8% of the attendees either ran their own business or operated a division of a major company.) Endurance sports are a kind of conspicuous consumption of time and willpower, demonstrating that someone with a demanding, time-consuming job somehow still has time to spare. In one extreme instance of this, the CEO of Take-Two Interactive has time for an intense fitness routine and time to write a book about it.

It's possible that causation runs the other way: training for a triathlon will certainly force you to develop discipline, which you could then apply to working harder. (And, of course, to planning your work—aspiring triathletes need to be brutally realistic about their training program.) On the other hand, spending that training time at work instead would also inculcate a habit of putting forth more effort.

There might be some kind of cross-training effect, perhaps because athletic training is more quantifiable than some parts of climbing the corporate ladder. In other words, people build habits that get a good return-on-discipline in areas where that can be measured, and then apply those habits to other domains.

But there's also a signaling aspect to all of this. A successful person who does endurance sports has chosen a hobby that takes up gobs of free time. There are athletic hobbies that are more time-efficient, like powerlifting. And if the goal is just to get lean, there's an even easier option: fasting requires a negative investment of time (but plenty of willpower).

Whatever the root cause, there's a subset of the US population that is increasingly willing to undergo punitive exercise routines. Even if they're spending fifteen hours a week training for an event, their time is valuable, and getting a little bit extra out of that time—or, if it's unpleasant, making it a bit more endurable—is something they'll pay for. So the growth of the fitness industry ends up being a trend that exacerbates visible inequality while reducing financial inequality: the customers who are committed to getting fit have the means and willingness to pay a lot for it.

Further reading: Sweat Equity is a good look at the rise of the fitness industry circa 2016. As with a previous Diff book recommendation, The Polyester Prince, there is, inevitably, at least one Kindle Unlimited romance with the same title, so please choose carefully.

Disclosure: I own shares of Peloton.

Diff Jobs

We've had a burst of new signups recently, so now's a great time for a reminder: Diff Jobs is a talent-matching service connecting Diff readers to interesting companies. It's free for job-seekers, and we're happy to work with an interesting set of companies tackling fun and lucrative problems. If you're looking for your next role—or not actively looking but curious about what's out there—please reach out.

Interesting new roles in the network:



Elon Musk has made an unsolicited offer to buy Twitter for $54.20/share, and also threatened to dump his stock if the company doesn't accept. As of this writing, Twitter is up 6.7% to $48.93, indicating some mix of uncertainty about how serious his offer is and certainty that he's entirely serious about blowing out of a 9.1% position if he doesn't get his way.

One theory on Musk's behavior is that he's mostly trolling, and that this is just a very expensive way to have fun. Maybe Abbie Hoffman was right but the joke only worked with a few more zeroes. Another possibility is that he's strategically trolling, and plans to profit from this situation by turning Twitter into more of a meme stock instead of a lower-margin meme-distribution venue. (This doesn't have to mean selling! Musk can borrow against his Twitter holdings if the stock appreciates.)

But another way to look at this is that Musk has spent the last decade or so in a unique position, where he's running capital-intensive companies that can only raise sufficient cheap capital because he's involved. Regardless of how SpaceX and Tesla are structured, Musk's control rests on the fact that both companies are hard to imagine without him, and get a valuation premium because of his outsized personality. Actually taking Twitter private would mean some combination of liquidating other assets, borrowing against them, or convincing someone to help him fund a Twitter deal. (Matt Levine has written through these options in more detail.)

Another thing to note is Musk's analysis style. The Ashlee Vance biography has this quote from an early SpaceX employee:

He's involved in just about everything... He wants answers that get down to the fundamental laws of physics. One thing he understands really well is the physics of the rockets.

If you're confident you can think faster than most people, it makes sense to get good at first-principles thinking so you can lazily evaluate reality instead of memorizing a bunch of disconnected facts. This style works best in fields where there are some broadly-agreed upon underlying rules, like physics, rather than hazy and contradictory generalizations, like debates over the limits and tradeoffs of free expression. It's entirely possible that Musk is genuinely convinced that a less moderated Twitter is an enormous social good, and that ensuring that it exists is worth putting other assets at risk—that he's obligated to be the person who saves Twitter from excessive bans the same way he feels obligated to make humans a multi-planetary species. It's a lot easier to derive unusual beliefs from first principles than it is to consistently act on them, but in Musk's case the domains he chose for that kind of thinking turned out to be lucrative enough that he can apply it somewhere else, too.

Remote Work and Neurodiversity

A more globalized and remote-friendly market for services has powerful effects, some of which The Diff has covered before. One counterintuitive side effect is that this model makes it easier for companies to hire neurodiverse workers ($, WSJ), who may simultaneously be better at performing their job duties and worse at dealing with some of the irritations of office work. Hiring has always meant offering a bundle of compensation options in exchange for getting a bundle from employees in return, and remote work makes it easier to shift away elements of this bundle. In-person interaction helps with some roles, but it's easy to get the dosage wrong, and it's very difficult for a company to create a culture where every employee gets the right balance between helpful interactions and annoying interruptions. Working over email, Slack, and Zoom ends up formalizing this in a useful way.

And this will probably extend to other areas, too: if a company is willing to have someone work in the office one or two days a week and remotely the rest of the week, it's not a stretch to think that they could work three or four days a week instead of full-time; coordinating with a part-time worker is not that much different from coordinating with a partly-remote one, especially for projects that require one person to spend an extended period on their tasks before one big deliverable.


Google is testing out something new in shopping search results: price comparisons to similar products.

This is very tricky to get right, since one of the major drivers of those price differences will be quality and brand differences. (This is certainly what you'll hear if you ask a seller why their version costs more.) It's a limitation of sorting and filtering that everything needs to be compared along rankable axes, and it leads to behavior changes from sellers. Large airlines have argued that their "basic economy" products are not something customers directly asked for, but something they implicitly wanted every time they went to a metasearch site and listed the results by price. Even if a budget carrier's tickets were $25 cheaper but charged $30 for a bag that was included in a different airline's basic fee, the budget carrier got the click. For retailers, what this means is that it's more important to acquire customers with surprisingly cheap products and then upsell them on something higher-margin.

Unhealthy SPACs

An index of SPACs that merged with target companies has declined by 31.6% since 2018, compared to a 50.4% gain for the S&P, with only a few days where returns were ahead of the overall market. SPACs have a selection problem: any SPAC with accurate and enticing future projections is just a few years away from being able to file for a conventional IPO, in which those projections will be audited historical data instead of optimistic PowerPoint slides. Given this, the companies that choose the SPAC route are either a) companies in a hurry to raise financing and go public, or b) companies that are not that likely to hit the numbers that justify their initial valuation. And since the supply of B can be expanded at will, while A is a lagging function of which companies got formed years ago, the SPAC population ends up selecting for disappointment. It's not intrinsic to the product itself, just a reflection of the current incentives.

For an earlier look at SPACs, their performance problems, and their incentive issues, see this Diff piece from 2020 ($).

More Evidence for the Travel Boom

A few weeks after Carnival cited an all-time weekly booking record, Delta says plane ticket sales for the last five weeks are at an all-time high ($, FT). Travel demand is partly a function of Covid and partly a function of Covid regulation, and in both cases can track not just the current situation but likely changes. So even pandemic-indifferent travelers might be reluctant to book a ticket if they worried their trip would be disrupted.


Part of my mental model of Amazon's retail business is that it tries to offer consumers as predictable an experience as possible—so they'll order from Amazon by default, and not think twice about renewing Prime—by offloading some of the inevitable volatility of business to sellers. A new instance of this is Amazon adding a 5% shipping surcharge to shippers' costs, which will not automatically be added to their prices ($, FT). This is not out of the ordinary for delivery companies; Fedex and UPS both pass on fuel surcharges. By imposing the cost on sellers rather than automatically raising prices, Amazon is basically daring them to be the first to fully reflect the charge in prices and lose their "buy-box" status as the default seller for a given product. (An especially troubling proposition if those shippers are also paying Amazon to store their products in the meantime.) This isn't a perfectly inelastic situation, and some sellers will raise their prices in response, but Amazon's approach makes it more likely that sellers will bear more of the cost.

Disclosure: I own shares of Amazon.