Pour Moi Adopts AI Fitting Tech in Chinese Lingerie Market

H2: Why Virtual Fitting Isn’t Just Hype—It’s a Conversion Lifeline in China

The Chinese lingerie market hit ¥38.2 billion in retail value in 2025 — up 12.4% YoY — but conversion rates for online lingerie purchases remain stubbornly low at just 1.7% (Updated: November 2026). That’s less than half the 3.9% average for apparel overall. Why? Sizing uncertainty. Returns due to fit issues account for 34% of all lingerie returns on Tmall and JD.com — nearly double the rate for casual wear. Consumers don’t abandon carts because they dislike the lace; they bail because they’re unsure whether ‘M’ means ‘medium’ or ‘maybe-not’.

Enter Pour Moi — the UK-based mid-market brand known for inclusive sizing and direct-to-consumer agility. In Q3 2026, it quietly soft-launched an AI-powered virtual fitting tool across its Tmall Global flagship and WeChat Mini Program. Not as a gimmick, but as a targeted response to one pain point: size confidence. And unlike Victoria’s Secret’s earlier AR try-on experiments (which prioritized visual flair over precision), Pour Moi’s system was built from the ground up for Chinese body metrics, local return behaviors, and mobile-first shopping habits.

H2: What It Actually Does — And What It Doesn’t Claim To Do

Pour Moi’s AI Fitting Tech isn’t scanning your living room with your phone camera. It doesn’t ask users to strip down or upload selfies. Instead, it deploys a lightweight, three-step guided input flow:

1. Height, weight, and current bra size (with optional verification via bust/waist/hip measurements in cm) 2. Preferred fit preference (e.g., "tighter band, fuller cup" or "softer underwire, wider straps") 3. Past purchase feedback (pulled from Tmall order history if consented, or manually entered — e.g., "last M felt tight in band, true in cup")

That data feeds into a proprietary sizing model trained on 127,000+ anonymized fit outcomes from Pour Moi’s global customer base — but reweighted using 2025–2026 fit feedback from 18,400 Chinese customers. The output? A single recommended size per style — plus a confidence score (low/medium/high) and a concise rationale (e.g., "87% match for 'Luna Plunge Bra' — based on 92% of similar body profiles selecting size 75C").

Crucially, the tool *does not* promise perfection. Its on-screen disclaimer reads: "AI recommendations improve fit accuracy — but individual preferences and fabric stretch vary. When in doubt, size up." That transparency builds trust where competitors overpromise.

H2: Real Impact — Not Just Vanity Metrics

Within eight weeks of rollout, Pour Moi’s Chinese digital channels saw measurable shifts:

- Cart abandonment dropped 22% on product pages with the AI widget enabled (vs. control group without) - Average order value (AOV) rose 14.3%, driven by higher add-to-cart rates for matching panties and shapewear bundles - Return rate for bras fell to 26.1% — down from 33.8% pre-launch (Updated: November 2026) - Net Promoter Score (NPS) among users who engaged with the tool: +41 (vs. +18 for non-users)

These aren’t vanity spikes. They reflect behavior change. One Beijing-based customer support log from October 2026 noted: "Customer returned 'Luna Plunge' in 75B after AI recommended 75C — but then purchased two more styles *using the same recommendation*, citing 'the explanation made sense'. No refund requested."

That nuance matters. It signals that the tool isn’t just reducing friction — it’s building category literacy. In a market where terms like 'underwire', 'quad-cut', or even 'cup depth' lack standardized Chinese translations, the AI’s plain-language rationales (“This style has deeper cups to hold fuller tissue — 75C fits 91% of customers with your measurements”) serve as micro-education.

H2: How It Compares — And Where It Fits in the Competitive Landscape

Pour Moi isn’t alone in experimenting. But its execution stands apart from peers — both global and domestic. Victoria’s Secret launched an AR bra try-on on its China app in early 2025, but adoption stalled at <5% of session users due to device compatibility limits and no integration with actual size databases. Intimissimi’s 2024 WeChat tool asked only for height and weight — too crude for bra sizing — and delivered generic advice like “Try M”. Etam’s 2025 pilot used third-party SDKs and suffered from inconsistent measurement capture across Android OEM skins.

Meanwhile, domestic players like La Vie En Rose and Hope rely heavily on live-streamed fit consultations — effective for high-touch sales but costly to scale. Triumph uses a PDF size chart with QR-linked video demos — helpful, but static. Pour Moi’s approach bridges the gap: algorithmic precision, localized logic, and human-readable context.

The table below compares core implementation traits across six major brands active in the Chinese lingerie market:

H2: Why This Works Now — And Why Timing Matters

Three converging forces made 2026 the right inflection point:

1. Platform maturity: Tmall’s 2025 API update enabled secure, consented access to anonymized order history — critical for training feedback loops. Prior to that, brands relied on self-reported post-purchase surveys (<22% response rate).

2. Consumer readiness: 68% of Chinese lingerie shoppers aged 22–35 say they’d trust an AI size recommendation *if it explained its reasoning* — up from 41% in 2023 (Updated: November 2026). The shift reflects growing comfort with algorithmic assistance in finance, health, and travel — not just entertainment.

3. Operational pragmatism: Pour Moi didn’t build this in-house from scratch. It partnered with Shanghai-based FitLogic — a B2B sizing AI startup serving 14 lingerie and athleisure brands in APAC — and co-trained the model using Pour Moi’s proprietary fit taxonomy. That kept development time under 14 weeks and cost under $320,000 — far less than Victoria’s Secret’s estimated $2.1M AR initiative.

H2: Caveats — And What Comes Next

Let’s be clear: AI fitting won’t replace fit consultants, nor will it erase cultural barriers overnight. Some limitations are structural:

- It works best for standard bra silhouettes (plunge, full cup, balconette). It struggles with highly engineered styles like molded t-shirt bras with internal seams or convertible multiway designs. - It assumes consistent manufacturing tolerance — but supply chain variability across factories (even within the same brand) means a 75C from Batch A23 may differ subtly from Batch B09. Pour Moi mitigates this by tagging batches in its CMS and adjusting confidence scores accordingly — but it’s a band-aid, not a fix. - It doesn’t address subjective preferences: one customer’s “firm support” is another’s “uncomfortable pressure”. The tool surfaces fit patterns, not taste.

Next, Pour Moi plans to layer in fabric-level insights — e.g., “This microfibre cup stretches 12% more than satin — consider sizing down if you prefer structure.” That requires integrating material specs from its ERP into the AI pipeline — a Phase 2 rollout scheduled for Q1 2027.

H2: Tactical Lessons for Other Brands

If you’re evaluating similar tech, skip the vendor pitch decks. Start here:

- Start with returns data, not algorithms. Pour Moi’s first step wasn’t choosing a tech stack — it was auditing 18 months of Chinese return reasons. 73% cited “band too tight” or “cup too shallow”. That dictated the feature set: band/cup decoupling logic came before anything else.

- Localize the logic, not just the UI. Translating “underband” to “下围” isn’t enough. You need to know that Chinese consumers associate “tight band” with “support”, not “discomfort” — so the AI’s language had to reflect that framing.

- Design for low intent, not high commitment. Most shoppers won’t spend 90 seconds on a fitting quiz. Pour Moi’s flow takes <22 seconds on average — because it defaults intelligently (e.g., pre-fills height/weight from past orders) and lets users skip non-critical fields.

- Measure what moves the needle — not what’s easy to track. Pour Moi tracks “recommendation acceptance rate” (how often users select the AI-suggested size) and “post-recommendation cross-sell lift” — not just “widget impressions”.

H2: The Bigger Picture — Beyond Bras

What Pour Moi is doing sits at the intersection of three macro trends reshaping the Chinese lingerie market:

1. The rise of ‘functional intimacy’: Shoppers increasingly view lingerie through performance lenses — breathability, moisture-wicking, posture support — not just aesthetics. AI fitting becomes the gateway to explaining *why* a particular construction delivers those benefits.

2. The erosion of brand-as-proxy-for-fit: Legacy trust in names like Triumph or Etam is weakening. In 2025, 57% of Chinese shoppers said they’d try an unknown brand *if the fit tech gave them confidence* (Updated: November 2026). Technology, not heritage, is becoming the new credibility layer.

3. The quiet consolidation of fit intelligence: We’re seeing early signs of ecosystem play. FitLogic (Pour Moi’s partner) now shares anonymized, aggregated learnings across its client base — meaning Pour Moi benefits from Hunkemöller’s Dutch fit data, and Scala gains insight from Bendon Lingerie NZ’s Pacific Rim patterns. It’s not open source — but it’s a pragmatic alternative to every brand reinventing the wheel.

None of this replaces craftsmanship, design intuition, or ethical sourcing. But in a market where 41% of first-time buyers abandon after one confusing size interaction (Updated: November 2026), AI fitting isn’t about replacing humans — it’s about removing avoidable friction so the human elements — fit, feel, confidence — can finally take center stage.

For teams building their own solution, our complete setup guide walks through data sourcing, vendor evaluation, and KPI alignment — with templates adapted from Pour Moi’s internal rollout playbook.