Everything below is written so that it can be checked against the pages it produced. A method that cannot be held against its own output is a statement of intent, not a method.
The rule the rest of it hangs from
No figure appears on this site without its source beside it. Not in a note at the bottom, not implied by a logo — beside it, in the same table row or the same sentence.
That is a practical rule rather than a pious one. A figure with a source attached can be disputed, traced and corrected by anybody who reads it. A figure without one can only be believed. Where a manual and a shopping listing disagree, the manual is what gets printed here. Where neither publishes anything, the field is left empty and the page says the field is empty — because once a plausible substitute has been typeset, nothing on the page distinguishes it from a real measurement.
Fitment before everything, including price
The first question on every page here is whether the thing will physically work with what a reader already owns and with the room it will stand in. It comes first because it is the question that produces returns, and because it is the one a photograph can never settle.
In practice it is a handful of millimetre figures. The bore a plate is cut for, Ø 30 or Ø 50, which meet at no point. The spacing between holes up a rack upright, and whether it tightens where a bench press needs it. The loadable sleeve length a bar offers before a collar runs out of room. The architrave depth a doorway bar has to bear on. The pin diameter a cable attachment needs. Where a maker publishes none of it, the page records that it was not published — which is itself a finding about the listing.
The headline figure, and the condition that makes it mean anything
Second comes whatever the listing shouts, carried together with the thing the listing usually omits. A load rating with whether it is static or dynamic, or with the fact that the maker did not say. A band's resistance in kilograms alongside whether that is a setting or a point on a curve. A cable station's stack mass kept beside the pulley ratio that halves it at the handle. A treadmill's horsepower with peak or continuous attached to it, or the note that neither word appeared.
When a maker prints the number and omits the condition, the omission is what gets recorded. That is not pedantry: on this shelf the condition is frequently the difference between two products that quote the same figure.
Then the four that decide whether it is still in use in five years
- The part that fails, and whether it is sold on its own. Bench upholstery, trampoline bungees, treadmill belts, cable assemblies, selector mechanisms. Every family has one. The page names it and says whether a spare exists, because that single fact separates a product with a service life from a product with a component life.
- What the box turns out not to contain. The collars a bar is cut for and ships without. The filling a training bag does not include. The masonry fixings a wall-mounted frame needs. The pump an exercise ball assumes. Anything requiring a second purchase before it works was never the price on the label, and the page adds it back.
- What the room has to provide. Ceiling height against a frame plus a bar overhead. Floor construction against a fixing pattern. Rear clearance behind a treadmill. Crated dimensions against a staircase turn. These are published often enough to compare, and they eliminate products rather than ranking them.
- Whether it can be bought where the reader is. Every reference is confirmed store by store before a page goes up, and each page lists the markets that had an offer. Excellent equipment that nobody stocks in your country is not a recommendation.
Three sources, and which one wins when they disagree
For anything structural — bores, hole spacings, thread sizes, shaft diameters, fixing patterns, what an attachment states it fits — the manufacturer's own documentation is taken first and a marketing headline last. The order inverts for anything printed in the largest type on the box, for the obvious reason that the largest type exists to sell.
Ratings and markings come second, and they are treated as claims to re-confirm rather than as permanent properties. A maker revises a pad, a bearing or a supplied collar without touching a model name, and a standard is revised on a schedule of its own.
Owners a year or two in come third, and for one question they have no substitute. No specification sheet records whether a selector dumbbell still locks after a winter, whether a bungee has gone slack, whether a bench seam has split, or whether anyone still answers the support address. Where a judgement rests mainly on one of these three, the page says which. Where none of the three answers, the page says so instead of reasoning its way to a number.
Positions within a family, never a score
Nothing here carries a mark out of ten. A single score would have to trade a power rack against a bottle of liquid chalk, and there is no honest exchange rate between those two things. What appears instead is a position inside a family, with the sentence that put it there printed beside it.
Families are ordered from the modest upward rather than from the expensive downward. Nearly everyone arrives with a figure in mind and one specific gap to fill; a page that opens on the dearest item on the shelf is answering a question almost nobody came with.
The pull this whole arrangement resists
Affiliate income leans towards whatever is expensive, recently launched and pleasant to write about. What holds against it is arrangement rather than good intentions: no family has ever been reordered for earnings, several pages here end by recommending a scrap of plywood or a replacement pad instead of a new machine, and no position has ever been exchangeable for anything.
One distortion is specific to this subject and is better named than quietly resisted. Equipment on this shelf is sold on capability figures nobody will ever reach. Build a page around those and it finishes by recommending the largest number in every family, because a bigger number can always be argued for. Where the honest answer is that an upgrade changes nothing a reader would notice — a heavier rating on a rack whose base is too narrow to use it, a bar rated far beyond what its sleeves will physically hold, a stack a 2:1 pulley halves before it reaches the hand — that is the answer the page gives.
One bias cannot be arranged away. What is covered here is what can be bought through the Amazon stores of the United States, the United Kingdom, Germany, France, Italy and Spain. Equipment sold only direct, or only through commercial suppliers, is under-represented as a result, and where that shapes a conclusion the page says so.
Nothing on this site is calculated by us
This shelf is sold on numbers, and a number is the easiest thing in a listing to repeat without thinking. One rule covers every page carrying one: every mass, load rating, bore diameter, stride length, flywheel mass, belt size, resistance figure, thickness, density, hole spacing, pulley ratio and quoted noise level on this site is copied from the manufacturer, in the unit it was published in, with its source named — and not one of them is recalculated, converted or inferred here. Where a maker publishes nothing, the page says nothing was published.
The pulley ratio is where that rule pays for itself. Dividing a stack mass by a ratio takes a second and the answer looks authoritative on a page. It would also be our number rather than the maker's, and there would be nothing in the typography to say so.
What a published load actually commits to
Everything above rests on a single distinction, and that distinction is the reason these pages sound less confident than the ones they sit beside in a search result: a published load records what a maker put in writing; it does not undertake an outcome. Where a named standard sits behind it, that standard describes a test — a defined mass, applied in a defined way, for a defined duration. Where none is named, which is most of this shelf, the figure is the maker's own. Neither is a statement about a particular frame, a particular floor or a particular failed repetition. Treat either as a promise and you sell more equipment, at the cost of misleading somebody about what they bought.
The two things this site will not write
Both limits are fixed, and both are better stated plainly than left to be worked out. Nothing here — no rack, no safety catch, no belt, no strap — is described as making anything safe. A page reports what the maker specifies for its own product: the load cited, what that load covers, the construction, the fixing pattern, the inspection asked for and the replacement interval. The rest belongs to those instructions, because whether equipment holds depends on how it is assembled, what it is fixed to and what arrives on it — and no listing knows any of that.
And no page here says what anybody should lift. A load rating is a property of a bench. What a body should be asked to do is not printed on any box, is not ours to state, and is the one recommendation this site is structurally unable to make honestly.
The point at which a page stops and a manual starts
None of this substitutes for the manufacturer's instructions. For a rack's anchoring, for the torque on a bolt carrying a loaded bar, for a cable's inspection interval, for the substrate a wall fixing assumes and for the sequence a frame is assembled in, the page points at those documents rather than rewording them — even though a page that stops there earns nothing.
One further limit is repeated on every page it touches instead of being stated once and assumed: whether two things fit is settled by a figure or a list the maker published, never by how they look side by side. A bore decides which plates go on a bar. A hole spacing decides where a J-cup lands. An architrave decides which doorway bar can hang from it. A pin diameter decides which attachment meets which cable. Where a maker publishes none of it, the page reports the answer as unpublished rather than deducing it from a photograph or a product name.
Questions readers send in about this
Has any brand ever paid for a position on this site?
Never, in any form — money, loaned equipment, an arrangement of any kind. If that ever became untrue, the page it touched would say so in its own body text, where somebody reading that page would actually see it, rather than in small type at the foot of the site.
Where do the numbers on these pages come from?
Three places, and every page names which one it leaned on: what a manufacturer puts in writing, what a named standard defines when one is cited, and what owners consistently report after a year or two of use. A load rating or a stride length therefore appears here as something somebody wrote down, with their name attached.
How can something cheaper finish above something dearer?
Because the order follows the job. A bench whose pad is wide enough at the shoulders gets trained on for years; an expensive one measuring 22 cm across is worked around on every single press. The equipment that fits the room and the body is the equipment that stays in use.
Do pages here ever conclude that nothing should be bought?
Regularly, and those pages earn nothing. Measuring a ceiling, sliding plywood under a rack leg, checking a door architrave before trusting a bar to it, wiping a bench down after a session — each of those does more than most purchases, and each of them is free.
Why will this site not say whether a rack will hold?
Because no published document supports that sentence. A load rating is what a maker put in writing and it is rarely tied to any named test. What happens under a failed repetition depends on the repetition, the frame and the floor. So the figure gets printed, its condition gets stated or its absence recorded, any standard cited gets named, and the page stops where the evidence does.
Why will this site not tell me what weight to buy?
Because that is not a property of the product. A dumbbell’s mass is published and comparable between brands; what a particular body should be asked to lift is neither published nor ours to state. The pages here give the range, the increment and the mechanism, and say plainly where a published figure stops being enough.
Last reviewed 21 September 2026