The State of Recovery

What the sample is

Every figure is computed from Praxium’s own directory — the Published studio listings that power the site’s city hubs and studio profiles — read live at computation time. There is no survey behind it, no purchased panel, and no third-party market report. A listing’s fields come from the studio’s own website and its public business profile, and its services are matched to the modality taxonomy the rest of the site routes on. What that matching does and does not confirm is set out on how we verify; this release does not claim more than that page does.

Read the result for what it is: a large sample of a category, assembled by us, not a census. No government agency tracks “recovery studios” as a standalone industry, so nobody — including us — can state how many are operating in the US. What can be stated precisely is what is listed and Published in this directory, which is what every row here does. The directory changes continuously as studios open, close, and update, so each row carries its own as_of date rather than a single report date.

What is counted, and what is dropped

• Scope is the United States. The public directory is US-only by design, so the release carries no international rows rather than a partial international slice.
• Listings are deduplicated by name and city before anything is counted, using the same key the city hubs use. An operator carried under two records counts once.
• Records with no routable city are excluded — including the “multiple locations” placeholders a few chains publish, which belong to no single market and would inflate any city they were guessed into.
• Nothing below 3 listings is published. Not a price median, not a share, not a pairing. A modality with two listed studios in a city produces no row at all — neither in the CSV nor here.

That floor is the rule this release is built around, and it is the reason some obvious rows are missing. A median over two listings is not a median, it is those two numbers wearing a statistical hat; a “share” over two listings is a coin flip with a decimal point. Publishing them would make the file look more complete and make every figure in it worth less. Here the rule is enforced in the code that writes the file, so a thin slice is omitted rather than narrated into confidence it has not earned.

Known limits

• Price coverage is thin. 421 of 3,117 listed studios publish a starting session price at all. Where a median is shown, the number of listings behind it is shown next to it, and where fewer than 3 listings publish one, the cell is empty rather than filled.
• Price is stored per listing, not per service. A studio’s starting price is attributed to each modality it offers, so a multi-modality studio contributes the same figure to several rows. That is an approximation, and it is the reason these are labelled median listed price rather than a price for the service itself.
• Self-selection. Studios that publish a price are not a random sample of studios; they are studios that chose to publish a price. Read the medians as directional.
• A share of listings is not a share of the market. Every percentage here has the directory as its denominator, which is stated in the column definitions and in the column name.
• None of this is a Praxium result. These are observations about businesses in a directory, not outcomes produced by Praxium, and not benchmarks for what anything should cost.