Fashion: size verification in 30 seconds — does it really reduce returns?
In apparel e-commerce, 20-30% of operational cost is size-related returns. AI Voice Agent during confirmation can verify fit in 30 seconds. Question: when does this exchange make sense? We show where yes, where no.
AI size verification is not magic — two simple questions (height/measurement + preferred fit), comparison with manufacturer's size chart, and a suggestion. Works best for categories with stable sizing (T-shirts, hoodies, basic pants). Poor fit for shoes and vintage clothing. Pilot 01-03/2026: returns dropped from 28% to 19% in 90 days.
- Size returns: 20-30% of PL fashion operational cost (Statista 2024).
- AI verification in 30s = 2 questions + size chart + suggestion.
- Works best with stable sizing categories. Poor for shoes.
- CallBotAgent pilot: -9 percentage points returns in 90 days (28% → 19%).
Size returns: 20-30% of fashion cost
Statista 2024 for PL fashion market: average online return rate is 28%, of which 60-75% are size-related. Roughly every fifth order returns due to fit. Each return costs 12-35 PLN operationally (return logistics + handling + inspection + restocking).
Traditional answers: extensive size charts (partial work), virtual try-on with AR (expensive), free returns (worse economics). Missing piece: 30-second conversation the customer must already have during COD confirmation.
Two questions, one size chart, one suggestion
AI Voice Agent during confirmation (after item check + address) asks a brief verification sequence. Total 15-30 seconds, doesn't extend the call uncomfortably.
Sequence is the same for all clothing: first question verifies if customer knows their size in this brand. Second (conditional) asks for reference measurements. Suggestion based on manufacturer's size chart from store catalog.
- 1Q1: 'Do you know your size in brand X?' If 'yes' → confirms order. If 'no/unsure' → goes to Q2.
- 2Q2: 'Height and chest measurement?' AI compares with manufacturer's chart from catalog.
- 3Suggestion: 'I recommend size M, but if you prefer looser fit — L works too'. Customer accepts or modifies. Size change = Shopify update.
Pilot 01-03/2026: -9 percentage points size returns
Pilot ran in 1 PL fashion store (casual + basics, avg cart 165 PLN, 60% COD / 40% prepaid). Sample: 412 confirmation calls in 90 days.
Baseline (pre-AI): 28% returns (70% size-related). After deployment: 19% returns at day 90, with 52% size-related. -9 pp overall returns, -18 pp size-specific.
What didn't work: unstable sizing categories — shoes (-2 pp), vintage/recycled (0 pp), swimwear (-1 pp). Conclusion: AI size verification works for 60-70% of fashion catalog, not all.
Size chart must be in catalog — without it AI won't suggest
Only technical requirement: products in Shopify catalog must have a size chart in metadata.size_chart field — per product or per product line. Product Catalog (Starter+) auto-syncs this with our AI.
Verification config in panel: toggle 'Size verification' per product category. The rule 'ask only if customer hasn't confirmed size in order notes' is frozen — you don't want to ask 100% of customers, since some know exactly what they're buying.
Four things to remember
Size returns = 60-75% of all fashion returns. Each 1pp reduction is 12-35 PLN cost per order.
30 seconds in call = real return reduction. Pilot showed -9 pp in 90 days for stable sizing categories.
Don't ask 100% of customers. Rule 'only when not confirmed in order notes' avoids friction for sure customers.
Requirement: size chart in catalog. Without chart AI has no basis for suggestion. Sync via Shopify metadata.size_chart.
Fashion store?Check return reduction on your own orders.
1,000 trial credits (~20 confirmation calls) · full access to your chosen plan · cancel anytime. Card required · no charge during trial · auto-activates on day 8.