A Size Recommendation Is Not a Measurement: What AI Needs to Guide Shirt Fit
A shopper usually does not ask, “Which label should I buy?” The real question is more human: “Which size will sit cleanly at the shoulder, give me enough room at the chest and still feel comfortable after lunch?”
That difference matters. An AI system can recommend a size, but it cannot turn an unreliable label into a reliable measurement. A strong fit assistant begins with evidence and ends with an honest level of confidence.
Why the size label is only the starting point
M, L and XL are convenient names, not universal dimensions. Even within one wardrobe, a regular-fit shirt and a closer-cut occasion shirt can feel different in the same labelled size. A recommendation therefore needs garment-level information: chest, shoulder, sleeve and length measurements; the cut; and any notes about stretch or ease.
This is closer to a recommendation problem than a simple lookup. Google’s overview of recommendation systems describes a common three-stage pattern: candidate generation, scoring and re-ranking. In a fashion context, the “candidates” might be available sizes, the score might reflect likely comfort and the final ranking should apply rules such as stock availability and minimum ease. See the Google for Developers recommendation systems overview.
The four inputs that make guidance useful
1. A trusted reference garment
For many shoppers, the most practical input is a shirt they already like. Measure it flat, then compare those dimensions with the size chart for the new garment. This avoids pretending that a body measurement and a finished-garment measurement are interchangeable.
2. Fit preference
“Fits” can mean sharp, relaxed or deliberately oversized. An assistant should ask whether the shopper wants room for movement, plans to tuck the shirt in, or prefers a neater silhouette. The answer can change the recommendation even when the measurements stay the same.
3. Product-specific facts
The system should retrieve the actual size options and verified product details, not invent them from the image. For example, the Regal White Cotton Men's Shirt can be used as a real product-page reference, while the final size choice should still follow the measurements shown on that page and the shopper’s preferred ease.
4. Feedback with a clear meaning
“Returned” is not enough. The useful signal is why: collar tight, shoulder narrow, sleeve long, or simply a change of mind. Those reasons should not be collapsed into one negative label.
Confidence should be visible
A responsible answer might say: “Based on your reference shirt, L appears closest; confidence is moderate because sleeve length is missing.” That is more useful than a confident guess. When two sizes are close, the system can explain the trade-off and direct the shopper to the size chart.
Testing also matters. Precision and recall expose different kinds of error in classification systems, so a fit tool should not be judged only by the percentage of recommendations that led to a purchase. It should also track exchanges, reason codes and whether certain body or fit profiles receive weaker results. Google’s classification metrics guide explains why one headline accuracy number can hide important mistakes.
Men’s and boys’ sizing must remain separate
An adult shirt model should not be casually reused for boys’ kurtas. Children’s sizing often references age or height, while adult menswear commonly uses garment or body dimensions. Growth allowance and parent preference introduce different decisions. Separate data, language and review rules are essential.
A practical TRYBUY.IN approach
AI-assisted fit guidance is most valuable when it reduces uncertainty without disguising it. Start with verified garment measurements, ask one or two useful preference questions, compare against a familiar garment and provide a fallback when the evidence is incomplete.
Browse TRYBUY.IN men’s shirts, then use the listed size information as the source of truth for your final choice.
FAQ
Can AI guarantee that a shirt will fit?
No. It can narrow the choice, but bodies, preferences and garment construction vary.
Should I submit a photo for a size recommendation?
Measurements from a well-fitting garment are usually more dependable and less intrusive than a photo alone.
What if I fall between two sizes?
Compare the shoulder and chest first, then choose according to the fit you prefer and the product’s exchange terms.
Shop TryBuy Kurtas
Explore these men’s kurta picks from TryBuy. Check each product page for current sizes, prices and availability.
- Men’s Beige Multicolor Floral Printed Kurta
- Men’s Khaki Slim Fit Cotton Kurta
- Men’s Sky Blue Regular Fit Cotton Kurta
Shop all TryBuy Men’s Kurtas — browse the full ethnic kurta collection.