America's Third-Largest Fitness Chain Has 2,634 Locations. A Gym Census Finds 242.

Two companies you can't see, invisible in opposite ways

This series has spent twenty-odd posts on a single question: when an industry consolidates, can you see it in the data? Usually the answer turns on names. A private-equity roll-up buys 400 dental practices, keeps every practice's name and website, and the consolidation becomes invisible — the listings look exactly like 400 independents.

Fitness has that problem too, and a second one that is stranger.

Purpose Brands was formed in April 2024 by a merger of equals between Orangetheory Fitness and Self Esteem Brands, the parent of Anytime Fitness. It is the classic case: the parent's name appears on zero storefronts, but every location is still legibly a gym. Count category=gym and you find all of it — you simply mis-attribute it to two unrelated companies instead of one.

Xponential Fitness is the strange one. It franchises five boutique brands, and almost none of their locations are filed under gym at all — which makes the company structurally invisible to anyone studying gyms, however thoroughly they look.

Owner	US listings	Tagged `gym`	% in gym	Largest primary category
Purpose Brands	4,204	3,747	89.1%	gym (89%)
Planet Fitness	2,757	2,063	74.8%	gym (75%)
Xponential Fitness	2,634	242	9.2%	pilates_studio (46%)
LA Fitness	690	662	95.9%	gym (96%)
Crunch Fitness	603	519	86.1%	gym (86%)
Life Time	232	149	64.2%	gym (64%)
Equinox Group	206	132	64.1%	gym (64%)
Gold's Gym	169	146	86.4%	gym (86%)

Xponential is the only operator in the top eight whose largest category is not gym, and the only one below 64%. It sits at 9.2%.

Purpose Brands is 1.5x the size of Planet Fitness. Xponential is within 5% of it. Neither name is on a door anywhere in America.

Where Xponential actually lives

The company's five current brands each landed in a different corner of Google's category taxonomy, and only one of them — the smallest — sits mainly in gym.

Brand	US listings	Primary category	Tagged `gym`
Club Pilates	1,241	pilates_studio	10
Pure Barre	642	physical_fitness_program	225
StretchLab	505	fitness_center	0
YogaSix	224	yoga_studio	1
BFT (Body Fit Training)	22	personal_trainer	6

StretchLab is the cleanest illustration. It has 505 US locations and zero gym tags. Club Pilates, the largest boutique fitness brand in the country at 1,241 US studios, contributes 10. The categories are not wrong — a Pilates studio genuinely is a Pilates studio — but the cumulative effect is that a company with a Planet-Fitness-sized footprint leaves almost no trace in the place you would look for it.

Across all five brands the mass lands in pilates_studio (1,223), fitness_center (367), physical_fitness_program (358), yoga_studio (223) and wellness_center (145), with gym (242) a distant fourth.

It is worth being precise about what is and isn't unusual here. Large chains all pick up a long tail of incidental category tags — Crunch Fitness's 603 listings touch 30 distinct labels, more than Xponential's 15 — so the number of categories a company spans means little. What is unusual is where the weight sits. For every other major operator, the single largest category is gym and it holds 64–96% of their listings. For Xponential the largest is pilates_studio, and gym holds 9.2%.

What a gym census would conclude

Suppose you did the obvious study: pull all 58,520 US listings in category=gym, group them by owner, and report who dominates American fitness. Here is what you would find, against what is actually true.

Owner	What a gym census sees	Actual US footprint	Undercount
Purpose Brands	3,747	4,204	1.1x
Planet Fitness	2,063	2,757	1.3x
Xponential Fitness	242	2,634	10.9x

The census is approximately correct for the two operators whose brands read as gyms, and wrong by an order of magnitude for the one whose brands read as studios. That is the trap: the method does not fail loudly. It returns a plausible, internally consistent ranking in which Xponential places somewhere below Gold's Gym. Nothing about the output signals that a top-three operator has gone missing.

This is a distinct mechanism from the ones this series has catalogued before — name-and-website preservation, franchisee-owned domains, per-practitioner listings, per-service listings, sibling-category composition, closure lag. Call this one category displacement: not that an owner's brands touch many labels — every large chain does — but that the centre of mass sits outside the category a researcher would search. It is the only mechanism so far that defeats a complete category census rather than a sampled one.

The measurement problem underneath all of this

Every number above had to survive a counting method that took three tries to get right, and the failures are worth stating because they are not specific to gyms.

The category's own vocabulary saturates it. The dataset's q= search does OR-token matching, so q="Planet Fitness" matches Planet or Fitness. Within category=gym, the bare token Fitness returns 22,699 listings, and so does every two-word brand containing it. The clean test is a control query using a brand that cannot exist:

Query (category=gym, US)	Total returned
q=Fitness (bare token)	22,699
q=Qwertyuiop Fitness (does not exist)	22,699
q=Planet Fitness	22,712
q=Anytime Fitness	22,909

Relevance sort hides the contamination. Records matching both tokens rank above records matching one, so the head of the result set is 100% genuine and stays that way for pages. Sampling 500 records deep proves nothing — for Planet Fitness the results stay clean until record ~2,070, then fall off a cliff into LA Fitness and Fuel Fitness. Depth is not evidence when the sort front-loads the clean records.

Stopping at the first impure page undercounts. Our first pass ended the count at the first page that was not 100% brand matches. For Club Pilates a single interloper on page 9 truncated the count at 899 against a true 1,241 — a 28% undercount from one record. The fix is to page past the cliff and stop only after several consecutive empty pages.

Names collide; domains don't. Counting Row House by name returned 116 listings, of which 30 were apartment buildings. Counting by therowhouse.com returned 40. Every figure in this post is counted by website domain.

How we did this, and the caveats

Listings are not locations. A listing is a Google Business Profile; chains occasionally maintain more than one, and permanently closed sites can linger. Our counts land close to independently published figures — Club Pilates 1,241 against 1,400+ reported, Pure Barre 642 against 624, StretchLab 505 against 531, Planet Fitness 2,757 against a reported 2,611–2,732 — which is the accuracy band to read these numbers in, not to three significant figures.

Purpose Brands is our loosest number. The company reports 7,000+ locations globally with 46% outside the US, implying roughly 3,780 domestically; we count 4,204. The gap may be closed locations, multi-profile sites, or a dated public figure. It does not change the ranking — the company leads on any reading — but treat 4,204 as the top of a range.

Category tags are Google's, not ours. We did not reassign anything. The finding is precisely that the taxonomy, used as intended, disperses one company across 15 labels.

This is a footprint measure, not a revenue or membership measure. A 1,200-square-foot Pilates studio and a 20,000-square-foot big-box gym count the same here. Purpose Brands leading on locations does not make it the largest by members or revenue, and we make no such claim.

Count an industry yourself

The Google Business dataset behind this analysis covers 132M+ listings with category, website and geography on every record — enough to check who really owns a market. 2,000 free credits a month, no card.

Explore the Google Maps APISee pricing