# Product dimensions across two country pages: observation protocol

Recorded before opening sampled product pages. Pilot: LACK 30449908 US/GB; the LACK family is excluded from the sample. Category cards (including their abbreviated sizes) were visible during sampling. This is a prospective analysis plan after a feasibility pilot, not a blinded or externally preregistered study.

Question: For 30 public IKEA product listings selected from three US categories, can the displayed overall product dimensions be reconciled with the exact same article on a GB page without guessing?

Sampling: first ten main result cards visible in each frozen public web-reader category response; ignore alternate-color links. Side tables excludes LACK; retain the next eligible main cards. Retain combinations and nonrectangular products as potential unknowns. No substitutions for unavailable counterpart pages, missing axes, or inconvenient results. The categories are side/end tables (10717), bookcases (10382), and dining chairs (25220). This is a convenience sample from one publisher, not independent observations across manufacturers or an industry estimate. Cards and rankings may differ between retrievals; the frozen sample table determines membership.

Collection: use the linked US page, then replace only /us/en/ with /gb/en/ for the initial counterpart lookup. A different article code is not the same item. Missing, redirected, or unconfirmed same-article pages remain unresolved; no name-only substitutions. Record URLs, article ID, variant label, observed date, named overall product axes and displayed units. Ignore shipping packages, seats, and other partial dimensions. Use public web-reader content, which may be cached; retrieval date does not establish source update time. Preserve unavailable retrievals and contradictory raw-access redirects as access limitations.

Comparison: exact rational conversion (one inch = 25.4 mm), same named axes only. Depth stays depth; never relabel it length. Compare common named overall axes only, requiring height plus two horizontal axes for a complete three-axis decision. A diameter is retained as diameter and does not imply width and depth. Where the toolkit's length/width/height contract cannot represent the named axes, use the separately published study calculation; do not silently alter tool input. Keep exact equality and a preselected 2 mm per-axis display-comparison threshold as separate results. The 2 mm threshold is an analytical sensitivity choice, not a manufacturing/installation tolerance or evidence of an error. Do not optimize it after seeing results.

Decision: hold for identity or missing comparable three-axis evidence; otherwise exact match, within 2 mm only, or difference exceeding 2 mm. Report counts over all 30 and over the comparable subset. Describe display differences without declaring a physical measurement wrong. Publish observed numerical facts, source links, machine-readable rows, deterministic calculation and correction log. Do not republish source images, reviews, descriptions, or full pages.

Validation: reread both measurement sections for each complete comparison, recompute fractions with an independent rational-arithmetic implementation, and verify that category totals and all outcome totals reconcile to 30. No traffic, citation, adoption or AI-model-performance claims follow from this study.