Everything you'd want to know before you install.
A detailed look at how AI Size & Fit Advisor works, why we built it the way we did, and the thinking behind the features above.
§ 01
Nobody knows their measurements
A size chart asks for bust and hip measurements the shopper has never taken. They guess, order two sizes and send one back. AI Size & Fit Advisor starts from what everyone knows: their height and weight.
§ 02
Two paths, one result
With "Height & weight", the shopper also gives their gender when the chart is unisex, and their build (slender, average, full). The module estimates their measurements with a regression on body mass index, then compares them with your chart. With "My measurements", the shopper uses a tape measure, guided by a silhouette that highlights the current measurement. Tape values take priority, and height and weight fill in the ones left empty. Estimated values are flagged and the confidence score drops accordingly.
§ 03
A result that leads to the order
The recommended size appears on a garment label, with the confidence score, the neighbouring size to consider, an explanation in the shopper's language and a per-measurement breakdown (good fit, snug, loose). The "Select size M" button picks the combination. If that size is out of stock, the shopper sees it and the button offers the available alternative. The profile is remembered: on later visits the product page button shows "Your size: M" straight away.
§ 04
Charts you control
Each chart targets all products, categories, brands or specific products, with a clear priority: product, then brand, then category, then all products. Six ready-made templates give you a starting point. The "Model fit" setting tells the advisor when an item runs small or large, and the tester checks the chart with a height and a weight before you save.
§ 05
Measure the effect on returns
Each recommendation is linked to the order that follows, with the size actually ordered. Merchandise returns and credit slips are then taken into account. The performance panel compares the return rate of orders placed in the recommended size with that of other sizes, and lists the most advised products.
§ 06
Optional AI, protected budget
The deterministic engine is enough to produce the recommendation. The AI layer (Mistral, OpenAI or Anthropic) refines the choice and writes the explanation, but its answer is discarded if the size does not exist in the chart. A per-visitor hourly quota protects your API budget. Only the measurements needed for the calculation are sent, without name or email.
There are no reviews yet.