mezu

How Mezu Converts Clothing Sizes

mezu compares brand sizes through body measurements in centimeters. The goal is a cautious closest-size estimate, not a promise that a garment will fit the same in every style.

mezu covers tops and bottoms for Zara, H&M, Mango, and Uniqlo in the current live pair set; it does not cover shoes or dresses.

Measurement-first conversion

A size label is only a label. When a chart has enough sourced data, mezu maps the source brand size to its body measurements, then finds the nearest target brand size using the measurements that matter for that category.

  • Tops compare bust or chest first, with waist or hips included when both charts have it.
  • Bottoms compare waist first, with hips included for women when both charts have it.
  • The primary bust, chest, or waist measurement must exist in both charts. Missing secondary measurements are disclosed.
  • If any comparable source range extends beyond the stored target chart, the result is “No reliable match”. This conservative rule applies to our stored coverage, not every size the retailer sells. Use your own measurements in the finder when a letter covers too wide a range.
  • Published ranges stay visible. Overlap gives a zero interval gap, but overlapping ranges can still describe different bodies. A chart row with a single point is not an exact garment-fit claim.

About the data

The live pages currently use 16 gender/category charts from 4 brands. The 15-brand curation registry also contains unpublished records; those are excluded from the live counts below. Numeric source links identify the market, category, measurements, version, and verification date. Methodology links are labelled separately and do not count as numerical proof.

Official brand or retailer size charts are preferred. Some official chart URLs are bot-protected, geo-routed, or only visible after product-page rendering, so a direct static fetch is not always enough to capture the numeric table. When that happens, mezu records the reachable official source context and treats secondary corroboration as weaker evidence, not as a high-confidence substitute.

H&M uses Indonesia-specific charts. Mango uses the selected US categories. Uniqlo women use the cited FW2025 body-chart revision, while men’s bottoms are scoped to a named product. Zara’s numeric data remains secondary and medium confidence. Check the product guide when shopping in another market or collection.

Confidence tiers

  • High confidence (10 live charts) means the relevant chart was checked against an official numeric source.
  • Medium confidence (6 live charts) means the relevant values are corroborated or official product-page data supports the chart, but the evidence is not as complete as a master body chart.
  • Low confidence (0 live charts) can be stored for curation, but low-confidence charts do not generate pair pages.

A comparison inherits the weakest confidence of the charts and rows it uses, including peer brands on brand pages. Confidence grades the evidence; it is not a probability that clothing will fit.

How chart rows are ranked

For each shared axis we measure the gap between published intervals, using a point when only a single value is available. For ranking, each gap is multiplied by its axis weight; the squared results are summed, divided by the sum of weights, and square-rooted. This score orders chart rows; it is not a physical gap on a single body measurement. Tops use bust/chest weight 1, waist 0.6 for women or 0.5 for men, and hips 0.6 when available for women. Bottoms use waist 1 and women’s hips 0.8. Equal scores keep the target chart’s published order.

The comparison label uses the largest individual gap: up to 1.5 cm means closest chart values, up to 4 cm means nearby chart values, and larger gaps require checking measurements. These are display thresholds, not experimentally validated fit tolerances. The page names the limiting axis and does not display a fit percentage.

Why recommendations stay approximate

Brand charts change, garment cuts vary by collection, and stretch or fabric can change how a piece feels. mezu keeps the recommendation cautious and shows the underlying measurements so shoppers can compare the chart before buying.

Data changelog

This log is generated from the `updatedAt` and `lastVerifiedAt` fields on the live brand records. Historical source references keep their own earlier dates.

  1. H&M, Mango, Uniqlo, and Zara data records were updated and source-checked.