How a guide is researched

A typical page on this desk follows five steps. The point is a number a reader can re-measure, tied to a public source, with the partner link kept off the explainer until the end.

1. Topic selection. We keep a short list of search intents that are already purchase or fix decisions: under-16-inch seats, the tape method, tailbone and calf symptoms, footrests, seat depth, and cylinder swaps. The list is reviewed when a guide goes stale, not when a keyword tool spikes.
2. Source collection. Tier 1 is OSHA, HSE, CCOHS, NIOSH, or a clinical reference on the NCBI Bookshelf. Tier 2 is a specialised reprint such as the Openshaw and Taylor PDF hosted by Oregon State, and it is used on at most two articles. The working list is the sourcing standard described in the editorial policy, with URLs in docs/authority-sources.md.
3. First draft. Yara Haddad writes the page and puts a named source on every numeric claim, with the access date. Worked examples use fixed statures so the arithmetic can be checked. They are not presented as patient histories.
4. Editorial review. Linh Doan checks the calculator, the tables, and whether a sentence overclaims. She is not the author. Medical red flags are sent to a clinician instead of being given a product.
5. Update cadence. Every guide shows a published date and an updated date and is re-read every six months, or sooner if OSHA, HSE, or CCOHS changes the posture page it relies on. Listing copy on Amazon is not frozen here. The reader re-reads the minimum on the day she buys.

Drafting assistance from software is used for research organisation and for line edits. A human decides the clearance convention, the examples, and whether a sentence ships. The declaration in the page source is “human-curated, AI-assisted research.” That is also the line in the editorial policy.