Every apparel brand knows this question: "Which size should I get?"
It shows up hundreds of times a week, and most brands file it under support. It's really a sales opportunity. Someone asking about sizing is ready to buy and needs just enough confidence to click "Add to Cart."
Answer quickly and you make the sale. Answer late and they leave, probably for good.
In this post I'll show how to automate sizing questions with AI in a way that raises conversions instead of frustrating customers.
Why Sizing Questions Are Different
By the time someone asks "which size should I get?", they've already:
- ●Found your product
- ●Liked it enough to consider buying
- ●Checked the size chart (and it didn't help enough)
- ●Decided to reach out instead of bouncing
This person is at the bottom of the funnel, deciding rather than browsing. Confidence is the only thing between them and a purchase, and traditional support can't deliver it fast enough.
If your average response time is 4 hours, that customer has probably:
- ●Closed the tab
- ●Found a competitor with clearer sizing
- ●Gotten distracted and forgotten
- ●Talked themselves out of it
Research shows that response time is the #1 factor in converting pre-purchase inquiries. Speed counts for more than the quality of the answer.
An instant "good enough" answer beats a perfect answer 6 hours later.
Why Size Charts Don't Work
"But we have a size chart!"
Every brand does, and the sizing questions keep coming, for four reasons:
Size charts require translation. "Chest: 42 inches" means nothing to someone who doesn't know their chest measurement. They need guidance more than data.
Size charts don't account for preferences. Some people want fitted, others want relaxed. "True to size" depends on what "true" means to that person.
Size charts can't compare. "I'm usually a Medium at Nike, what am I here?" A size chart can't answer cross-brand questions.
Size charts don't build confidence. Even after checking a chart, customers are often unsure. They want someone to confirm their choice.
The brands that convert well treat sizing guidance as a conversation.
What Good Sizing Guidance Looks Like
Before automating anything, look at how great human support handles sizing:
Step 1: Gather context
- ●What's your usual size at [similar brand]?
- ●How do you like your [product type] to fit: fitted or relaxed?
- ●What's your height/weight? (if customer volunteers)
Step 2: Process against knowledge
- ●This item runs small/large/true
- ●Customer reviews mention [fit detail]
- ●The fabric has [stretch/no stretch]
Step 3: Make a confident recommendation
- ●"Based on what you've told me, I'd recommend the Medium"
- ●"If you prefer a looser fit, size up to Large"
Step 4: Reduce risk
- ●"If it doesn't work, returns are free and easy"
- ●"Most customers with your measurements are happy with Medium"
That's a decision tree: inputs (customer preferences) → processing (product knowledge) → outputs (recommendation). AI can run it.
How to Automate Sizing With AI
This is the practical implementation.
Step 1: Audit Your Current Sizing Conversations
Export your last 500 sizing-related support tickets. You'll see patterns in:
- ●Most common products asked about
- ●Most common customer inputs (height, weight, usual size, preference)
- ●Most common concerns (too tight, too long, returns)
- ●How your team actually responds
This becomes your training data.
Step 2: Build the Knowledge Base
For each product (or product category), document:
- ●Fit profile: Runs small / true / large
- ●Key measurements: And what body types they suit
- ●Fabric behavior: Stretch, shrink after washing, etc.
- ●Comparison to other brands: "Similar to Nike Medium, smaller than H&M Medium"
- ●Customer feedback summary: What real reviews say about fit
It doesn't need to be perfect. It needs to capture what your best support rep knows by instinct.
Step 3: Train the AI
The AI needs to:
- ●Ask qualifying questions: Size at other brands, fit preference, product type
- ●Match against product data: Fit profile, measurements, fabric
- ●Deliver a confident recommendation: "I recommend Medium" instead of "check the size chart"
- ●Handle objections: "What if it doesn't fit?" → Explain easy returns
- ●Know when to escalate: Edge cases, frustrated customers, unusual requests
Step 4: Deploy With Confidence Building
Be open that it's AI, and frame it well:
✗ "Chat with our bot" ✓ "Get instant size recommendations"
Customers should see it as a helpful tool, not as a way to keep them from human support.
Step 5: Measure and Improve
Track:
- ●Resolution rate: What % of sizing questions does AI fully handle?
- ●Escalation rate: What % need human follow-up?
- ●Conversion impact: Do customers who engage with AI sizing convert higher?
- ●Return rate: Does AI guidance reduce size-related returns?
Use escalated conversations to improve the AI. Every edge case it can't handle today is training data for tomorrow.
What Results to Expect
Based on our implementations:
- ●Volume reduction: 70-85% of sizing questions handled without human intervention
- ●Response time: From hours to seconds
- ●Conversion lift: 10-20% on products where sizing is a major friction point
- ●Return rate: Often decreases (better guidance → fewer wrong size purchases)
- ●Support team impact: Freed to handle complex issues, VIP customers, and revenue-generating work
Common Mistakes to Avoid
Mistake 1: Using a generic chatbot
Generic chatbots say "check our size chart" or "here are our measurements." Customers have already seen the size chart, so that deflects the question without helping. The AI has to be trained on YOUR products with YOUR fit knowledge.
Mistake 2: No escalation path
Some customers want a human to confirm, and some situations really are complex. When the AI can't handle one, the handoff to human support has to be instant and seamless.
Mistake 3: Hiding the human option
Customers shouldn't feel trapped, so "Talk to a person" should always be one click away. Oddly enough, customers trust the AI more when they know they CAN reach a human.
Mistake 4: Not training on real conversations
AI trained on your size chart will repeat your size chart. AI trained on your actual support conversations will respond like your best rep.
Mistake 5: Forgetting the confidence layer
A recommendation on its own isn't enough, because customers need confidence to buy. "We have free returns if it doesn't fit" eases purchase anxiety more than another measurement.
The Bottom Line
"Which size should I get?" is a conversion opportunity sitting in your inbox. Every one of those questions comes from a customer ready to buy, if you can give them confidence fast enough.
Automate it with AI that understands your products, makes confident recommendations and responds instantly, and let humans handle the edge cases. You get more sales, fewer tickets, and customers who feel helped.
Want to turn sizing questions into conversions? We build AI that handles product guidance for eCommerce brands: size, fit, shade, and more. See how it works