How we choose picks

Every criterion below is computable from published documentation. None of them requires touching a product. Each pick is scored 0-10 on each criterion relative to the best published figure in that page's field, and the overall score is the mean of those, rounded to one decimal place. A score is therefore a ranking within a field on a stated criterion - not an absolute grade.

Weighting is published per page, not hidden. On standing-desk stools fit range carries the most weight, because the page's entire question is whether a stool reaches a desk. On drafting chairs adjustability leads, because the foot ring is the category's reason for existing. On chairs for back pain it is published lumbar travel. Each page's "How we ranked these" section says which and why.

Criterion 6 is computed but not always scored. Most makers publish no warranty term, and scoring its absence would penalize non-disclosure twice - once under criterion 5 and again under 6. So where a term exists we compute and display the figure per pick; where a page's field is mostly silent on warranties we score the five criteria that apply. The criteria used are listed inside each pick's score breakdown, so you can always see which five or six produced the number.

A product is excluded before it is scored if any of these apply. Exclusions are published with the reason, on the page where the product would otherwise have appeared.

1. It is not currently purchasable. Every product on this site was verified through a live retailer lookup at build time. A product we cannot confirm is on sale does not become a pick.
2. The manufacturer publishes no figure for the page's governing dimension. On seating that is usually the seat-height range: if we cannot assess fit, we cannot recommend it. Two stools sit on an excluded list for this alone, one of them with an 850 lb capacity we would otherwise have found remarkable.
3. Capacity headroom is negative for the user band the page addresses.
4. A capacity claim is made with no stated standard or test basis and nothing else on the listing is checkable. A number on its own is not a specification.
5. The published specs are internally contradictory across the manufacturer's own pages.
6. It is a product we have a relationship with but have not tested. One brand in this category is a hands-on testing partner of ours. Nothing of theirs appears anywhere on this site - no picks, no comparison rows, no "also consider" mentions - and nothing will until we have actually tested it. At that point it will be labeled as the only thing here we have sat on.

Every fit figure on this site comes from one module, computed from one published table: the British adult percentile data in Stephen Pheasant and Christine Haslegrave's Bodyspace: Anthropometry, Ergonomics and the Design of Work. Because it is one module, the fit tables, the per-pick fit columns and the comparison tables cannot disagree with each other. Because the source is named, you can check it.

We interpolate between the published percentile points rather than applying a single ratio of stature. The popliteal-to-stature ratio drifts from about 0.245 at the short end of the data to 0.263 at the tall end, so one multiplier systematically under-predicts for tall people - the group the market already serves worst.

The posture targets these functions serve are OSHA's, not ours: the chair height is appropriate when the entire sole of the foot can rest on the floor with the back of the knee slightly higher than the seat of the chair, and the seat pan should support most of the thigh without contact between the back of the user's knee and the front edge of the seat pan. Both are quoted from the OSHA Computer Workstations eTool, linked in Sources below.

Every price on this site comes from a live lookup against the retailer, is stored with the timestamp of that lookup, and is refreshed daily. Three rules follow, and they are enforced in code rather than by editorial discipline:

• A price older than 48 hours disappears. Past that the reader sees "check price" instead of a number. A component on this site physically cannot render a figure that has gone stale.
• No live price means no number, ever. Not a remembered figure, not a range, not "around". The fallback is "check price".
• Every displayed figure carries its date. Currently: last verified September 30, 2026, 79 products carrying a live price.

The one exception, stated because it matters: where we discuss a manufacturer's published list price or MSRP as a spec figure, it is labeled as such and cited to its source. It is never rendered through the price layer and never shown on a buy button.

• Scores derive only from the criteria in section 2. If a score appears anywhere on this site, the rubric that produced it is one click away and the per-criterion breakdown is on the page.
• Every roundup carries at least one explicit "skip this" - a pick we recommend against, with the spec-based reason. This is structural rather than editorial: every product in our registry carries a required trade-off field, so a shortlist with no exclusions is not possible to build here.
• Re-ranking is triggered by a specification change, a newly qualifying product, or a price move that changes cost per warranty year. Never by a commission rate. Where a cheaper option is better for the buyer it is the higher pick, including when it earns us less.
• Pick counts are never stated in body copy, because the count comes from the data. That is not a style choice - it means a new pick can be added without rewriting a paragraph around it.

Published rather than buried, because it decides whether this site is the right source for your question.

• All-day comfort over an eight-hour day.
• Foam breakdown and compression over months or years.
• Build quality, rattle, and mechanism durability.
• Assembly difficulty and packaging.
• Caster behavior on carpet versus hard floor.
• Noise.
• Whether a seat that fits on paper feels right to your body.

• Invent a figure. Not a spec, not a price, not a rating, not a review count. If it is not published somewhere we can cite, it is not on the page.
• Claim testing we have not done, or imply it with photography. No hands-on-a-lever shots, no workshop settings, no "in our testing".
• Emit our score as structured data. The visible score is ordinary page content; it is never marked up as a product rating. We do not run a test lab, so doing so would contradict this page.
• Publish testimonials or a rating for ourselves. This brand launched in 2026 and has no customer base. A page with no testimonials is correct; a page with invented ones is not.
• Give medical advice. There is no clinician here and nothing on this site is medically reviewed. Back-pain and sciatica pages describe mechanism - pelvic tilt, seat-pan pressure, published angles - and stop. Products in that space sometimes carry "doctor recommended" in their titles; that is a marketing phrase, not a credential, and we do not repeat it as one.
• Borrow authority. No "experts agree", no "studies show" without a named study, no implied review panel.

Product specifications are cited per product, on the page, linking the manufacturer's own listing. These are the sources behind the method itself. Currently 42 pages on this site carry product picks, and every figure on each of them traces to one of these or to the product listing named on the page.

• Pheasant & Haslegrave, Bodyspace: Anthropometry, Ergonomics and the Design of Work, as tabulated by RoyMech - every percentile figure the fit module uses.
• OSHA eTool, Computer Workstations - Chairs - the posture criteria the arithmetic serves.
• OSHA eTool, Computer Workstations - Good Working Positions.
• ANSI/BIFMA X5.1, General Purpose Office Chairs - test summary - the 275 lb occupant basis behind most chair ratings.
• ANSI/BIFMA X5.11, Large Occupant Office Chairs - test summary - the 253-400 lb standard, and the only one specifying a minimum seat width.
• BIFMA, Releases First Large Occupant Office Chair Standard.

Found an error? We would rather hear it than not. Tell us and we will check it against the source and date the correction on the page - which is the policy, not a courtesy.