Most fashion returns are about fit, not taste. A customer who picks the right size keeps the garment; a customer who guesses sends it back and rarely reorders. Every return costs the shipping both ways, the handling, and often the sale, because a garment that has been out and back is not always resellable at full price. A size advisor is the least glamorous tool in Fashion Try-On and, for a brand that ships, probably the one with the clearest return on effort.

What an advisor does

It asks for a few measurements, compares them with the garment's size chart and recommends a size with a confidence note. That is it. No magic, no body scan, just fewer guesses. The confidence note matters: a customer between two sizes should be told so, with a hint about the cut, rather than given a false certainty that turns into a return.

Where to place it

Next to the size selector, before the add-to-cart button. If it appears after the decision, it is a warning; before, it is help. We have seen advisors buried in a size-guide modal two clicks away, and the usage numbers were what you would expect. Put it where the hand already is.

Keep the questions short. Height, weight and one or two garment-specific measurements are enough for most categories. Every extra field halves the number of people who finish.

Feed it real charts

The advisor is only as accurate as the chart behind it. Measure your garments flat and enter the numbers; do not copy a supplier's generic table. Two shirts labelled medium from two suppliers can differ by several centimetres at the chest, and a generic chart averages that difference into a wrong answer for both. Measuring forty garments takes an afternoon. It is the same afternoon that fixes most of the return rate.

  • Measure flat, on a table, garment buttoned or zipped as worn.
  • Record the measurements that decide fit for the category: chest and length for tops, waist and inseam for trousers, bust and hip for dresses.
  • Enter them per size, per garment. If two garments share a pattern, they can share a chart; if not, do not pretend.
  • Note the cut in words: slim, regular, relaxed. The confidence note uses it.

Try-on and the advisor together

The try-on image shows what the garment looks like on a body; the advisor says which size to order. They answer different questions and both reduce returns for different reasons. A customer who has seen the drape on a model close to their build and has been told the size is less likely to order two sizes to compare. If you register your own models from full-body photos, you can show the garment on more than one build, which makes the picture honest and the recommendation credible.

Measure the effect

Track return rate on items where the advisor was used versus not. Analytics next to the catalog show usage; your order system shows returns. Even a few points of difference pay for the effort many times over. Give it a full return window before judging, because returns arrive weeks after orders, and compare like with like: the same items, the same season, advisor used against advisor ignored.

What not to expect

An advisor does not fix returns caused by colour, fabric feel or a photo that oversold the garment. It does not fix a size chart that is wrong. And it does not work for customers who will not enter a measurement; some will not, and for them the try-on image and a clear written chart remain the tools. The point is not to eliminate returns; it is to remove the ones that were never about the garment.

A rollout that does not disrupt the shop

  1. Pick the ten items with the highest return rate.
  2. Measure them flat and enter real charts.
  3. Switch the advisor on for those ten, placed before the add-to-cart button.
  4. After one return window, compare. Extend to the rest of the catalog if the number moved.

Frequently asked

Does the advisor store customer measurements? It uses them to make the recommendation. Your privacy policy should say what you keep; the platform does not use customer inputs to train anything.

Can the advisor work without a chart? It can fall back to a generic table, but the recommendation is then only as good as the table, which is the problem you are trying to solve. Enter real measurements.

Is the advisor charged per use? The advisor and analytics sit alongside the catalog rather than in the per-operation cost table; the pricing page lists what is charged, and the advisor is not a generation.