Seventy-five churches close in the Netherlands every year. Some of them reopen as padel courts.
That reads like a joke about decline. It is not. It is a transfer of social infrastructure from one balance sheet to another, and nobody wrote the terms.
Here is the transfer in numbers. The Netherlands had roughly 12,800 cafés in 2013. On 1 January 2026 it had 10,038. In the same country, padel went from 475 courts in 2020 to 3,523 at the end of 2025, across 780 locations. Some 876,000 people played at least once last year.
Nothing disappeared. The third place changed hands, changed buildings, and changed its revenue model.
01 / From pay per visit to pay per month
The café sold you a moment. You walked in, you paid at the point of consumption, you left. High margin per visit, low frequency, and the operator learned nothing about you.
The club sells you a habit. You pay on the first of the month whether you show up or not. Low margin per visit, high frequency, and the operator knows your name, your entry time, and the fact that you stopped coming three weeks ago.
That is not a small commercial difference. It is a different asset class wearing the same shoes.
Real estate has one dominant formula for the plinth, and that formula is turnover. Rent as a percentage of what goes through the till. Turnover rent is an elegant instrument and it does exactly what it was built to do. It selects for spend per visit.
Which means it selects against frequency.
The tenant who takes fifty euro off one person once a month wins. The tenant who takes twenty-five euro a month off three thousand people who each come twice a week loses, because the till reads lower and the rent per square metre says he cannot afford the corner unit.
So the corner unit goes to the phone shop. The padel hall goes to the industrial estate at the edge of town. And then we write reports about the loss of community.
02 / It is not the gym. It is the whole category.
Britain had 45 public saunas in 2023. It had 147 by the start of 2025. It has more than 600 today. That count comes from the British Sauna Society's own map, so read the slope rather than the point, but the slope is not in doubt.
In New York, Bathhouse raised 35 million dollars in May 2026 to go from three sites to ten, and says it runs around a thousand customers a day per location. A day pass is 29 to 39 dollars. A membership is 145 to 225 dollars a month. Its Los Angeles flagship takes 55,000 square feet of the old Amoeba Records building, on a twenty-five year lease.
Othership, four sites across Toronto and New York, lists "alcohol free" as a company value. One of its founders explains the appeal in a sentence that should worry anyone who has ever programmed a residential lobby: the bathhouse is a good place for conversation because nobody is on their phone.
Above that sits a tier that no longer pretends to be about exercise. Remedy Place operates what it calls the World's First Social Wellness Clubs, with the registered trademark symbol in the sentence. The third place is now intellectual property. Surrenne, in the Maybourne group's Knightsbridge hotel, charges 10,000 pound a year plus a 5,000 pound joining fee, caps membership at 100, and sold out. Six Senses Place opened in London this March at 3,000 to 37,000 pound a year, inside a one billion pound mixed-use redevelopment.
Underneath all of it sits the gym. Europe now holds 75.5 million memberships across 67,515 clubs, a market of 39.1 billion euro growing at 9 percent a year. That is 9.3 percent of the entire population paying monthly for a building they enter for no reason other than to be in it.
The spa sector as a whole ran 201,861 sites and 157 billion dollars of revenue in 2024, growing 14.6 percent in a single year.
And the substitution is measurable, if you accept the source. In a ClassPass survey of around 900 of its own members this spring, 53 percent of 18 to 34 year olds had booked a workout instead of a night out in the past year. A vendor asking its own customers a question it wanted a particular answer to, so hold it loosely. It is still the only number that measures the swap directly.
Now the part that cuts against the easy version of this story, because that version is wrong.
Dutch young adults are not lonelier than everyone else. In 2025, 8.5 percent of 15 to 25 year olds reported strong loneliness against 10.1 percent of the population, and the figure has been falling since the 2022 peak. Dutch students did not drink less in 2025 than in 2023. And American drinking, down from 62 percent of adults in 2023 to 54 percent in 2025, sat at 54 percent again in July 2026. The collapse stabilised.
This is not a rescue mission. It is a market.
03 / None of this is new. Only the invoice is.
Sober socialising, communal bathing, the gym as a social venue: all of it has been built before, at scale, and mostly by people who are dead.
A fourth-century catalogue of Rome lists 856 neighbourhood baths beside 11 great complexes, and the historian who works with that list calls the figure not above suspicion, so take the order of magnitude. The Baths of Caracalla held two libraries and two exercise yards, and entry cost a quadrans, the smallest coin in circulation. What a Roman did in there, in one scholar's summary, combined what we would now call the gymnasium, the bathroom and the massage parlour.
In 1879, Dickens's Dictionary of London counted nearly 3,000 coffee taverns in the United Kingdom, under the control of nearly 80 companies. Alcohol-free venues at national scale, commercially operated by dozens of limited companies, with a philanthropic vehicle above them: the Coffee Public House Association, chaired by the Duke of Westminster, described itself as a promoter and encourager of coffee taverns, not a trader in them, and existed to make the necessary advances. That is an impact fund with a soft-loan facility. It is a hundred and forty-seven years old.
And then the one that should stop the room. In 1921, the YMCA across North America took in 4.4 million dollars of member dues and 10.4 million dollars of gifts. It took in 21.9 million dollars from what its own Year Book called business and club features: restaurants, dormitories, camps, baths. Out of 47.3 million dollars total, commercial operation was 46 percent of income, nearly five times the membership dues.
The gym with a café, a spa, rooms upstairs and a coaching programme is not an invention of this decade. It is a century old, and the model was proven by a Christian youth charity.
So what actually changed? Three things, and all three are financial.
The surplus leaves the building. In 1921 the YMCA's commercial income went back into the same association that produced it. In 2026 it goes out at 26.6 times EBITDA to whoever buys the operator.
The door acquired a price gradient. Oldenburg published The Great Good Place in 1989 and made accessibility constitutive; the standard academic reading of his criterion is that there are no physical, policy or monetary barriers to entrance. A sauna session in Brooklyn is 39 dollars. Surrenne is 10,000 pound a year, capped at a hundred people, sold out. Both are third places. They are not the same third place. Oldenburg saw this coming, incidentally, and wrote in 1996 that chain venues make less hardy third places than independent ones because they pull money out of the community they sit in.
The public side is withdrawing while the private side scales. Amsterdam's public bathhouses recorded 1.9 million visits in 1956 and 80,000 in 1986. Tokyo had 2,687 sento in 1968 and 444 in 2023. England has lost 500 swimming pools since 2010, with 42 percent of those closures falling after 2020, and English local-authority youth provision is down 76 percent in real terms since 2010-11. Britain's Club and Institute Union had over 4,000 clubs and around four million members in the early 1970s. It has about 1,800 clubs and just under a million members now.
Nobody underwrote the swap. It is not in anybody's model, on either side.
I should be precise about how far this goes, because the strong version does not survive the data. Germany's sports associations recorded 28.8 million memberships across roughly 86,000 clubs on 1 January 2024, an all-time high, against 11.7 million commercial fitness members. There the association sector is two and a half times bigger and still growing. In the Netherlands it is convergence rather than a crossing: association membership moved from 32 percent in 2012 to 29 percent in 2022, commercial from 18.9 percent to a 2018 peak of 23.2 and back down to 20. The crossing appears in one cohort only, among 20 to 34 year olds, where commercial membership has run level with or above the club since 2016. And CBS reported in December 2025 that association membership overall fell from 70 percent in 2012-14 to 62 percent in 2023-24, lowest among 25 to 35 year olds, and running from 43.9 percent in the poorest income quartile to 74.7 percent in the richest.
So the claim is narrower than the headline, and more useful than it. The third place was not replaced. A second one was built beside it, priced by income, and only one of the two shows up in a rent roll.
04 / The word with the invoice attached
"Soft" is the most expensive word in real estate. "Amenity" is the runner-up. Amenity is what we call a function when we have already decided it can be cut: a square metre count, a fit-out budget, and no line in the model for what it produces.
CBRE tracked European retail destinations from January 2024 to July 2025. Assets with a fitness, healthcare or wellness tenant sat at plus 0.1 percent footfall against 2019. Assets without one sat at minus 4.7 percent. A gap of 4.8 points.
Say the honest thing first: that is an association, not a proven cause. Good assets attract good operators. But CBRE also found fitness floorspace up 106 percent since 2019 across the portfolios it manages. The market has already voted with its square metres. The underwriting has not caught up.
05 / The operators worked it out first
Life Time in the United States did 2.99 billion dollars of revenue in 2025 across 189 clubs and 822,380 memberships. Average revenue per membership: 3,531 dollars a year. Roughly 28 percent of that is not the membership at all. It is the food, the spa, the coaching. The hours around the workout. The same instinct the YMCA had in 1921, with better lighting.
Then look at what they do with the buildings. Life Time closed 200 million dollars of sale-leasebacks in April 2026 and plans another 200 million this year. It is selling the bricks and keeping the members, and building 327 branded residences next to a club in Phoenix.
Now put two valuations side by side. Third Space in London did 99 million pound of revenue and 26.3 million pound of adjusted EBITDA in 2024, and went to market in February 2026 at around 700 million pound. Call it 26.6 times operating earnings. UK gym property, the freehold itself, traded in 2024 at gross yields between 6.25 and 7.5 percent. Call it thirteen to sixteen times the rent.
Same building. The operating cashflow was worth more than the rent it pays.
That is my own arithmetic, not a published finding, and an operator's EBITDA carries risk a landlord's rent does not. Take the direction, not the decimal. The direction is the point. Marriott paid 355 million dollars for the citizenM brand and did not get a single building. I wrote earlier this month that the operation is the asset. This is the same sentence in gym clothes.
06 / The membership is portable. The building is not.
Look at what the operators actually sell, and it is not a room.
Third Space in London prices one club at 245 pound a month. Access to the group of clubs is 299. The top tier is 345. Same equipment, same trainers, same classes. That extra hundred pounds is the network, and it is sitting on their public price list.
ClassPass took the idea to its conclusion and sells the network with no buildings at all. Strava passed a million clubs on a platform whose venue is the street. Life Time runs one membership across 195 centres. In every case the unit of the business is the member, not the site.
That has a consequence the property side has never priced. Your address is one node in somebody else's roster. The habit you help create travels: it survives a holiday, a move, a change of job, because it was never attached to your postcode. And the party that can see it, hold it and charge for it is the operator, not the owner of the room it happened in.
It also changes where the third place starts. It no longer starts at the door. It starts three days earlier, on a phone, with a sign-up that asks what time you train, whether you do that alone or in a group, what you are working towards. Then the programme is built on the answers.
That matching is not decoration. It is the retention model. Members who train in groups stay 22.6 months against 16.2 for those who only use the gym floor, and gym-floor-only members are 56 percent more likely to cancel. Industry figures, so hold them loosely, but the direction is consistent and it is not about the equipment. Nobody added a machine. Somebody asked a question.
Now hold that against how a residential landlord takes on a resident.
A new tenant answers more questions to join a gym for 25 euro a month than to enter a ten-year relationship with your building. We ask for income, employer and a deposit. We do not ask when they are home, whether they cook, whether they would run a Tuesday session, what they would leave the flat for. Then we programme the plinth on a leasing assumption and wonder why the common room is empty.
And here is the part that should interest anyone holding more than one asset. A residential fund with eight thousand homes across thirty buildings already has a larger natural roster than most gym chains. It has the addresses, the plinths, the tenancy relationship and the renewal moment. It has never issued a membership.
That is the move I would make before any of this touches a lease. One membership across your own portfolio. Digital onboarding at signing, with three questions that actually shape service. Access that travels between your buildings, so a resident moving from one of your addresses to another keeps the habit instead of restarting it. Programming built on stated preference rather than on a floor plan.
Then the plinth stops being twelve unrelated units in twelve unrelated buildings and becomes one operating network with a roster attached. And for the first time you are on the right side of the value it produces.
07 / What I would put in the term sheet
First, the reason I built this.
This year, in one of the best-known monumental buildings in this country, the ground floor is being programmed with a gym. Not because anybody worked out what that street needs at seven in the evening. Because a gym is the covenant that fits the rent per square metre, on a floor plate nobody else wants at that price.
I have made the argument against decisions like that one, in rooms like that one, and lost. I had a view about how a street gets lived in. The other side had a rent roll. The rent roll wins that argument every time, and it should, because it is the only side of the table holding evidence.
So stop arguing and bring a number. If frequency is the product, measure the product.
Two instruments. Both boring enough to survive a legal review.
Visit Yield. Verified entries per square metre per year, read at the turnstile, reported quarterly next to NOI. Every membership operator already has this number to the minute. It is better data than any retailer's turnover declaration. No landlord I know has ever asked for it.
The frequency clause. A share of plinth rent indexed to delivered visits instead of delivered turnover, with a floor. If the club produces the footfall that the rest of the plinth and the flats above it monetise, the club stops paying for the privilege of producing it.
One refinement, because the gradient is the real risk. Two hundred thousand visits a year at 39 euro a door is not the same infrastructure as two hundred thousand at 25 euro a month. Visit Yield cannot see the difference. The frequency clause can: make the relief a function of visits delivered and the price of entry together, so the cheaper the door, the larger the discount. That single line decides whether your plinth produces a third place or a members' lounge.
The obvious objection is correct. Visits do not pay rent. A club at 3,000 members and 25 euro a month cannot carry prime retail. But that is the argument, not the rebuttal. The gym is not a tenant in the plinth. It is the reason the plinth has traffic. CBRE's own work with operators puts the ceiling at occupancy costs of around 20 percent of club revenue. Below that it works. Above it the club closes and takes the footfall with it.
Visit Yield is a crude number. It counts twenty minutes on a treadmill the same as two hours on a Sunday. It says nothing about who came, or whether anybody spoke. It is a proxy for frequency, and frequency is a proxy for belonging. I would rather argue about a crude number in an IC pack than keep having no number at all.
08 / What I am changing
From now on, every XD run I do asks for the access data of every membership operator in and around the asset before it asks for the rent roll, and records the price of entry next to it. Every plinth advice I write carries a frequency clause, and every portfolio brief opens with the question of whether the owner could issue a membership of its own. I will publish the first Visit Yield benchmarks once I have ten assets, so the number can be attacked properly.
One prediction, and this one sits on me rather than on the market. Before the end of 2028, a European plinth lease will be signed with rent indexed to verified visits. I intend it to be one of mine. I will come back in January 2029 and collect that receipt whether I have earned it or not.
Wellness real estate reached 876 billion dollars in 2025 and is forecast at 1.8 trillion by 2030. Of the 910 branded residence schemes worldwide, carrying a 33 percent price premium, 79 percent wear a hotel brand. The remaining fifth is where sport and wellness are arriving now. The premium is not the logo. It is the calendar, the operator, and the fact that somebody is responsible for whether Tuesday evening exists.
On a Tuesday in November, at ten past nine, thirty people stand outside a padel hall in the rain, waiting for a court to clear. They talk. Nobody sold them an evening. They bought a habit, and the evening came with it.
That is infrastructure. On the drawing it says fitness, 400 square metres.
Sources: KVK/ANP via Misset Horeca, 27 January 2026 (Dutch cafes). KNLTB and EY, Padel in Cijfers 2025, 31 March 2026 (industry body). Deloitte and EuropeActive, European Health & Fitness Market Report 2026 (industry body). British Sauna Society, own map, February 2026. Athletech News, 12 May 2026 (Bathhouse). Global Wellness Institute, 12 May 2026 and November 2025 (wellness real estate, spa sector). ClassPass survey of around 900 of its own members, August 2026. CBS StatLine (loneliness, 2025) and CBS Statistische Trends, 11 December 2025 (association membership). Gallup, 20 August 2026 (US drinking). Garrett Fagan, Bathing in Public in the Roman World, 1999, on the Notitia Urbis Romae Regionum. Dickens's Dictionary of London, 1879 (coffee taverns). YMCA Year Book and Official Rosters 1921, Annual Statistical Review. Amsterdam op de Kaart; Tokyo Metropolitan Government; Swim England / ukactive; Mulier Instituut; History Workshop and Stir to Action. Oldenburg, The Great Good Place, 1989, accessibility criterion as summarised by Jeffres et al., Applied Research in Quality of Life, 2009. CBRE Research, Fit for Business, December 2025. Life Time investor releases, 24 February and 30 July 2026. City AM, 16 September 2025 and mainsights.io, 2 February 2026 (Third Space, asking price, not a closed deal). Allsop, 25 November 2024 (UK gym yields). Savills, Annual Report Branded Residences 2025/26. Les Mills and HFA via ABC Fitness, 2026 (industry figures). The 26.6 times against thirteen to sixteen times comparison is my own arithmetic, not a published finding.