How to fix size chart inconsistency across styles with data
Inconsistent size charts push shoppers to order several sizes. A practical sequence for auditing garment measurements, using returns data and closing the loop with design and fit.
KEY TAKEAWAYS Summary by the editors
- Zalando states that size and fit can account for up to half of returns in European online fashion, and that return rates in jeans can reach 65%.
- Size inconsistency starts upstream: a 2025 study in the Journal of Textile and Apparel, Technology and Management notes that brands build size charts according to their own understanding, with no standard chart.
- Returns data with reasons such as too small or too large is the cheapest signal for finding styles whose labelled size does not match how they fit; Zalando uses purchase and return behaviour to produce size flags that show when an item runs small or large.
- Zalando reported that its size and fit solutions prevented 8% of size-related returns in 2025, which indicates realistic gains from tooling rather than a complete fix.
- Fixing the chart is a design and technical-development task as much as a data task: where flags reveal a grading or pattern error, the garment specification must change, not only the label.
To fix size chart inconsistency, start by comparing the measurements you intend each size to have with the garment measurements you actually produce, then use returns and customer fit feedback to find styles whose labelled size misleads customers. Correct the specification where the garment is wrong, and publish accurate guidance where it is not. Data shows where to look, but it does not remove the need for fit sessions and supplier checks.
Why do size charts become inconsistent across styles?
Charts diverge for ordinary reasons. Different blocks, fabrics and suppliers produce different fits under the same label, and charts are often copied between styles by habit. A 2025 paper on plus-size body shapes in the Journal of Textile and Apparel, Technology and Management notes that the lack of a standard size chart complicates fit identification and that brands build their own charts. It also cites earlier work finding that plus-size bodies do not grow proportionally when sizes are simply graded up. Shoppers, the paper adds, judge fit from model photos that may not reflect their own bodies.
The commercial effect shows in returns. Zalando's June 2026 account of its size technology says return rates in European online fashion can reach about 50%, with size and fit accounting for up to half of them, and that jeans can reach 65%. These are company figures for its own market, but the direction is widely accepted: fit uncertainty makes shoppers buy several sizes and send most back.
What data do you need to find the problem?
| Data | Question it answers | Typical weakness |
|---|---|---|
| Garment specification and measured samples | Does the garment match the intended chart? | Production tolerances are rarely recorded for finished goods |
| Returns with reason codes (too small, too large) | Which styles run small or large against their label? | Free text reasons and inconsistent coding |
| Purchase patterns by size (same style, several sizes) | Which styles trigger bracketing? | Needs customer level identifiers |
| Customer fit feedback and reviews | Is the problem length, width, or an area such as calf or hip? | Sparse for new or low volume styles |
| Body measurement data, where customers have chosen to provide it | Does the chart reflect the customers you actually have? | Consent, bias and small samples |
How do you audit size charts step by step?
- Build a master table of every style, size label, size chart reference, block, supplier and fabric, so that you can see which styles share a chart.
- Compare each chart with garment measurements from fit samples and, where possible, from finished production. Mark differences larger than your internal tolerance.
- Calculate size related return rates by style and size, and compare them with the category average. Use enough orders per size to avoid chasing noise, and set a minimum volume before a style is judged.
- Look for patterns: a single size that returns far more than its neighbours often signals a grading jump, while a whole style returning for the same direction (too small) suggests a block or label issue.
- Review the flagged styles with design and technical development, then decide for each: change the garment specification, change the label, or add a clear fit note on the product page.
- Re-measure after the next production run and track whether size related returns for the style move.
How can returns data produce size flags?
Zalando describes using machine learning on brand data, purchase and return behaviour, customer fit feedback and the insights of trained fitting models to produce size flags, which indicate when an item likely runs small or large, and size recommendations, which suggest the most likely fitting size. It says these foundational solutions cover about 70% of its assortment, and that return insights are shared with brands. You do not need that scale to benefit from the logic. Even a spreadsheet comparing the proportion of too small and too large returns by style and size is a useful first flag.
Treat the output as a prompt for investigation. A style that runs small may be accurate to its chart, and the chart may be the problem. Equally, a fabric with stretch can change how customers choose a size without anything being wrong.
Should you use customer body data to rebuild charts?
It can help, with care. Zalando says its Body Measurement tool, which uses two photos or a phone video, has been tried by more than 1.5 million customers, and that it has built what it describes as the largest anonymised body measurement dataset from European fashion customers, now used in size charts. A smaller retailer is unlikely to collect data at that scale, but can use aggregated, consented measurement data or public anthropometric surveys. Just Style reported in July 2023 that Zalando's measurement tool, launched for women's tops and dresses in Germany, Austria and Switzerland, cut size related returns by 10% compared with items without size advice. Body data is sensitive personal data in many jurisdictions, so collect it only with clear consent, minimise what you keep and aggregate it before using it for chart design.
Who should own size charts?
Give one team accountability. In many companies design owns the intended fit, technical development owns the specification, merchandising owns the product data and e-commerce owns the page. Without a named owner for the chart as a product data object, errors fall between teams. A simple governance rule helps: no style goes live until its chart reference has been checked against the approved garment specification, and a monthly review looks at the highest return styles.
For wholesale brands, the same discipline benefits retail partners, who inherit your sizing data on their own sites. Providing consistent measurements in product data feeds reduces the chance that a partner's customers return your product for avoidable reasons.
Frequently asked questions
Why do the same size labels fit differently across brands and styles?
Brands build their own size charts, patterns and grading rules, and there is no standard chart. Different fabrics, blocks and suppliers also change the finished garment. The result is that a size label is only a rough guide unless it is backed by consistent garment measurements.
How do I know if a style's size chart is wrong?
Compare its garment measurements with the chart, then check returns by reason and size. If a style returns noticeably more often than its category for the same reason, for example too small, review the block, grading and label before blaming customers.
Can AI fix size charts automatically?
Not on its own. Machine learning can flag styles that run small or large and recommend sizes, as Zalando describes, but correcting a chart often means changing the garment specification or the supplier process. Treat AI as a way to prioritise investigations.
How much can better size guidance reduce returns?
Published figures vary and come mainly from company reports. Zalando says its size and fit solutions prevented 8% of size related returns in 2025 and that its Virtual Fitting Room pilots reduced return rates by up to 40%. Expect your results to depend on category and data quality.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
