Methodology
How we evaluate recovery gear
This is our headline promise: a method you can check. Everyone in this category says they tested twenty units. We have not lab-tested any of them, and we say so — here is what we do instead.
0
Units we claim to have lab-tested
0
Sponsored placements accepted
0
Free products accepted for coverage
Those zeros are the point, not a gap. We would rather publish that we haven't tested something than pretend we have. What we bring instead is a consistent, reproducible method applied the same way to every product.
1. How we source the picks
We start from what people actually search for and what Amazon actually stocks — because a recommendation you cannot buy is useless. We build a longlist for each category (the well-known names, the enthusiast favorites, the value options), confirm each one is currently sold, and then cut it down. Nothing gets a spot because a brand asked; brands cannot buy a spot, because we accept no sponsored placements and no free product in exchange for coverage.
2. The scoring rubric
Every product is scored out of 10 against a small set of criteria chosen for the category. For a massage gun those are usually power (amplitude and stall force), noise, ergonomics, battery and value; for a compression system, pressure and zones, presets, comfort and value; for a wearable, sensors and accuracy, battery, the subscription reality, and total cost of ownership. The overall score is the mean of those metric scores, shown to one decimal place, and we publish the sub-scores on every card so you can see what the number is made of.
The score is a judgment from documented research — published specs, manufacturer manuals, and aggregated owner reviews. It is not a measurement we took in a lab, because we do not have one. This is also why you will never see an aggregate star ratingfrom us in a page's structured data: we have no customer reviews to aggregate, and inventing one would be a lie.
3. The specs we compile, per category
Recovery gear is unusually easy to evaluate on paper because the numbers are published. We compile them into one comparable table per roundup: for massage guns, amplitude (mm), stall force (lbs), speed range, measured or rated decibels, battery and weight; for compression boots, pressure range (mmHg), chambers, presets and whether it runs corded or on a battery; for foam rollers, density, length, diameter and surface pattern; for wearables, the sensor set, battery, and what each subscription locks away. Where a maker does not publish a number, we write "not published" rather than guessing.
4. The math nobody else shows
Two derived figures are our differentiator. For anything with a motor or a heater — a chiller cold plunge, a sauna blanket — we show the cost-to-run: the power draw multiplied by the US average electricity rate, so you know the monthly figure before you buy, not after. For subscription wearables we work out the total cost of ownership over two years, because a cheaper device with a monthly fee often costs more than a pricier one that owns its data outright. The arithmetic is on the page so you can check it with your own local rate.
5. The "don't buy this if" rule
Every product we recommend carries a line telling you who it is notfor. It is the fastest way we know to prove we are on your side rather than the retailer's, and no competitor page we studied does it. Every roundup also names at least one pick to skip.
6. How prices work
Every price on this site is pulled live from the Amazon Product API and stamped with the date it was fetched. If our data is more than 48 hours old, the number disappears and the button falls back to "Check price on Amazon" — we would rather show you no number than a stale one. We never type a price into the page by hand. Amazon's price at checkout is always the one that counts.
7. Health claims and updates
Anything physiological — cold exposure and inflammation, percussion and soreness, sauna and circulation — is cited to reputable secondary sources and framed as general information, not medical advice. Roundups are reviewed at least quarterly and whenever a major new model ships; every page shows a visible last-updated date. If we get something wrong and you tell us, we correct material errors within 48 hours and note the correction. See our editorial policy for the full standards, and contact us to flag anything.