Comparing product dimensions across country pages: a 30-listing audit

Eight complete comparisons; twenty-two unresolved ↗

This audit asks a narrow question: can the displayed overall dimensions of the same article on two country pages be reconciled without inventing a missing fact? It is a convenience sample from one publisher, observed on September 24, 2026, US Eastern time. It does not measure physical products, model accuracy, retailer quality, or industry-wide error rates.

The full denominator stays at 30. A web-reader timeout is recorded as a retrieval gap. It is not counted as a missing product page, an unavailable product, or an incorrect measurement.

Small display differences remained after exact conversion ↗

The calculation uses exactly 25.4 mm per inch. The table reports the largest absolute difference across each product’s three named axes. “Within 2 mm” is an analytical classification selected before collecting the full results. It is not a manufacturing tolerance, installation recommendation, or proof that either displayed value is physically correct.

The eight complete comparisons contain 24 axis comparisons. All 24 differ at zero tolerance; all 24 are within 2 mm. The pattern is consistent with display rounding, but these observations do not establish how either page produced its measurements.

Source: HOLMERUD side table — IKEA US measurement section; HOLMERUD side table — IKEA GB measurement section; OLSERÖD side table — IKEA US measurement section; OLSERÖD side table — IKEA GB measurement section; BRIMNES bookcase — IKEA US measurement section; BRIMNES bookcase — IKEA GB measurement section; BILLY bookcase combination — IKEA US measurement section; BILLY bookcase combination — IKEA GB measurement section; HEMNES bookcase — IKEA US measurement section; HEMNES bookcase — IKEA GB measurement section; BILLY bookcase with height extension unit — IKEA US measurement section; BILLY bookcase with height extension unit — IKEA GB measurement section; PINNTORP White Chair — IKEA US measurement section; PINNTORP White Chair — IKEA GB measurement section; SKOGSTA chair — IKEA US measurement section; SKOGSTA chair — IKEA GB measurement section

HOLMERUD: retain both original values ↗

For article 40541421, the observed US page lists length 31 1/2 in, width 12 1/4 in and height 20 1/2 in. The GB page lists length 80 cm, width 31 cm and height 52 cm. These convert to US values of 800.1, 311.15 and 520.7 mm, versus GB values of 800, 310 and 520 mm.

The absolute differences are 0.1, 1.15 and 0.7 mm on the same named axes. A strict equality rule returns a difference; the declared 2 mm comparison rule admits all three. Neither result tells an agent which page is authoritative or whether an item fits a particular space. Preserve both source records.

Source: HOLMERUD side table — IKEA US measurement section; HOLMERUD side table — IKEA GB measurement section

An unknown result needs a specific reason ↗

Twenty records lack a usable measurement section from at least one sampled country URL through this retrieval method. Some initial page reads succeeded while subsequent reads timed out; the retained observations use the complete measurement sections that were actually returned. Source update times were not available. A cached reader observation is not a fresh origin verification.

The other two holds are GLADOM 10534002 and BORGEBY 80619880. Both retrieved country pages report diameter and height. The predefined three-axis test does not invent length and width from diameter. That makes these records outside this comparison contract, not defective products.

The download retains each attempted URL, observed values, article identity check and retrieval status. Missing values remain missing. No records were replaced after results were seen. A future collection may resolve gaps; it must be published as a new observation with its own date and method.

Source: GLADOM tray table — IKEA US measurement section; GLADOM tray table — IKEA GB measurement section; BORGEBY side table — IKEA US measurement section; BORGEBY side table — IKEA GB measurement section

How the sample and rules were fixed ↗

The sample follows the first ten main result cards in frozen category responses for US side/end tables, bookcases and dining chairs. Alternate-color links were ignored. The LACK family was excluded after a feasibility pilot; the next eligible side-table cards filled its places. The exact 30 selected URLs are retained because category ordering and displayed variants can change.

The analysis protocol was committed locally before opening sampled product pages. Sample URLs were frozen after the initial link-opening step, which exposed some measurements. This is a prospective analysis plan after a pilot, not a blinded experiment or an externally preregistered study.

The initial counterpart lookup changes the country portion of the sampled URL from US to GB. A shared name is insufficient: complete comparisons require the same article number in both retrieved page texts and the same overall product axes. Shipping-package measurements, seat measurements, ranges and missing axes are not substituted. Depth stays depth.

Collection and source review were AI-assisted. The author’s publication is responsible for the method and interpretation. The complete numeric comparisons were checked against their retained source sections and independently recalculated using rational arithmetic.

Recalculate the observations locally ↗

Save the observation JSON and the calculation script in the same directory. Run node reproduce-dimension-study.mjs product-dimensions-study.json with Node.js 22 or later. The script reads the saved facts and returns the totals and every comparable axis. It makes no network requests and cannot verify whether a source page has changed.

This study calculation preserves depth as depth. The separate browser dimension checker accepts length, width and height only; do not rename an observed axis merely to fit its contract. The study script and the general checker answer related but different input formats.

Saved results and corrections ↗

The saved result contains eight complete comparisons, zero exact matches, eight comparisons within 2 mm only, zero exceeding 2 mm, and 22 holds. The subset of eight is the denominator for numeric agreement; 30 is the denominator for coverage. Neither supports an industry estimate.

Initial publication: September 24, 2026. No post-publication corrections are recorded. Report a correction with the article number, country URL, measurement section and observation date. Future corrections will retain the original observation and explain the change.

What this changes in an agent workflow ↗

Keep retrieval status, identity, measurement basis, named axes and conversion results separate. A tool can calculate a precise number from the wrong object. An agent also needs a useful way to stop when a required source or axis is unavailable.

This audit gives another reader checkable observations and a calculation, not evidence of tool adoption or traffic growth. The useful next test is whether a real workflow can resolve one of these holds without guessing and preserve the supporting source.

Cite this guide

George Kelly. Comparing product dimensions across country pages: a 30-listing audit iamgeorgekelly. Updated September 24, 2026. https://www.iamgeorgekelly.com/field-guide/product-dimensions-country-page-study

Keep the source notes and example labels with an excerpt. For a provider requirement, follow the original documentation in Sources & scope.

Choose your next task ↗ · Browse the source directory ↗