Building a Clothing Size Recommendation Product
A clothing size recommendation product has to answer a hard question: which size will fit this person in this garment, given the store's own chart. Maqasi, the AI-assisted fashion sizing app I built and launched, is an example of the approach. The design challenge is less about the model and more about making the recommendation understandable and honest.
The core problem is inconsistent charts
Every store measures garments differently. Two shops can label the same size with different measurements, and a single letter means little across brands. A useful product reads the store's own chart and compares it with the user's measurements, rather than assuming a universal size scale.
Inputs must be simple and honest
Users rarely know all their measurements. Ask for the inputs that change the recommendation most, explain why each one matters, and let people choose a simpler path when they are unsure. Avoid asking for information that does not change the result.
Explain the recommendation
A bare size label invites doubt. Show the reasoning: which measurement drove the choice, where the garment runs small or large, and how the fabric's stretch affects the fit. A short explanation builds trust more than a confident single answer.
Show uncertainty
Some garments and body shapes are harder to predict. Make the confidence visible, and suggest the next step, such as checking the store's chart for a specific measurement or choosing a size range. An honest 'between two sizes' is better than a false certainty.
Design the moment before purchase
The product is useful when it helps the user decide before they buy. That means a clear path from the product page to the recommendation, and a way to return to the same garment later. Users should be able to compare two options without restarting.
Build the data carefully
Chart data is messy. Store names, size labels and measurement units vary, and some charts change over time. Build a process for updating and checking charts, with a way to flag suspicious data before it reaches users.
Measure the outcome that matters
The useful measure is whether the user completes the sizing flow and whether they keep the purchase. Track completion and the points where people leave. Qualitative feedback about the explanation is as important as the numbers.
A product checklist
- Store chart read, not assumed
- Inputs limited to what changes the answer
- Reasoning shown with each recommendation
- Uncertainty stated clearly
- Chart data checked and updatable
- Completion measured step by step
A sizing product earns its place by reducing doubt at the moment of decision. Keep the explanation visible, the uncertainty honest and the data current.
Next step
Get help improving conversion
Send me the page or flow that loses people. I will tell you where the friction is and what to change first.
Open the conversion page