Rendered Fits Research · July 2026

AI Shopping Agents in Fashion: 2026 Statistics & the Fit Gap

AI shopping agents have gone from novelty to measurable retail traffic in a single year — and agentic checkout is already live in UK fashion. This page collects the independently sourced 2026 figures on adoption, traffic, trust and returns, and sets them against the one thing none of these agents can yet read: fit. Every figure below is attributed to a named, primary source.

900m
ChatGPT weekly active users, Feb 2026
+138%
YoY growth in AI-referred traffic to US retail sites, May 2026
54%
better conversion from AI-referred vs non-AI retail traffic
$262bn
in 2025 holiday sales AI agents are reported to have influenced
11%
ceiling on shoppers willing to let AI make the final purchase call
19.3%
of US online orders returned in 2025

In one line: AI shopping agents are no longer experimental — traffic, adoption and even agentic checkout in fashion are live and growing fast in 2026 — but every agent on this page can read a product's price and stock far more reliably than it can read whether the product will actually fit. That gap, not agent adoption itself, is the open risk for fashion merchants right now.

1. Shopping is already agent-mediated, at scale

The shift from search box to AI agent shows up in the traffic numbers now. ChatGPT reached 900 million weekly active users in February 2026, up from 800 million just four months earlier (OpenAI, via TechCrunch). Retail feels it directly: Adobe Analytics data reported by Digital Commerce 360 shows AI-referred traffic to US retail sites up 138% year-on-year in May 2026, and — a reversal from a year earlier, when it converted worse — that traffic now converts 54% better than non-AI traffic. Shoppers arriving via an AI agent are further along in the decision by the time they land.

2. Agentic checkout is already live in fashion

Salesforce reports AI agents influenced 20% of global online retail sales — around $262bn — over the 2025 holiday season (via MarTech). In UK fashion specifically, on 2 June 2026, Hey Savi and PayPal launched what they describe as the UK's first agentic commerce platform with native in-app checkout, with Debenhams Group — Debenhams, Karen Millen, Boohoo and Pretty Little Thing — as first retail adopter. On fit prediction, True Fit launched an agentic fit experience over Model Context Protocol on 17 February 2026, exposing roughly 20 years and $616bn+ of purchase and returns data across 91,000+ brands. Agentic commerce in fashion is a live category, not a trends-deck slide. Full analysis: our Hey Savi × PayPal launch review.

3. Consumers want AI's help, not its authority

Adoption is not trust, and the gap is where fashion's fit problem lives. A Gartner survey of US consumers (January 2026, n=322) found real openness to AI narrowing choices — 31% for household supplies, 28% for personal electronics — but willingness to let AI make the final purchase decision topped out at just 11%. Trust is the limiting factor: 54% said they had to double-check the AI's information, and 62% said it wasted their time. Shoppers want an agent that finds and narrows, then want to verify the answer themselves. In fashion, that verification moment is almost always: how will this actually look and fit on me?

4. The fit legibility gap

Point any of the agents above at a typical fashion product page and the split is stark. They already read price, availability, sizes offered, colour and material as structured attributes. They cannot reliably read how a specific cut fits a specific body, whether an item runs large or small, how one brand's "medium" compares to another's, or what the garment will actually look like on the shopper — the exact information that decides both the purchase and the return in fashion. On most sites it exists only as prose in a size-chart tab, if at all. An agent with nothing structured to reason with stays silent on fit, or recommends the store that made its fit data legible instead. Full detail on our AI-shopper-ready guide.

5. Unresolved fit already costs billions — agentic speed raises the stakes

The National Retail Federation puts total 2025 US retail returns at $849.9bn, with 19.3% of online orders returned; fashion is consistently among the most return-exposed categories, and fit is the recurring cause. Agentic checkout compresses discovery-to-purchase into one low-friction motion — good for conversion, but it removes the manual steps (size guides, multi-angle photos, review-reading) that used to partly resolve that doubt. Faster checkout without a structured fit signal is not a smaller return problem; given how little authority consumers hand AI (Section 3), it is a bigger one — the shopper has less to verify before the agent completes the purchase.

6. What this means for merchants: how a store gets read

Closing the gap is a buildable layer on top of the product data most stores already have:

Step by step detail: our AI-shopper-ready checklist. Live endpoint surface, OpenAPI 3.1 spec and hosted POST /api/v1/mcp: our page for AI agents and developers. The conversion/returns case for fixing fit generally: our virtual try-on returns benchmark.

Frequently asked questions

How many people are already using AI to shop?

ChatGPT reached 900 million weekly active users in February 2026, up from 800 million in October 2025 (OpenAI, via TechCrunch). AI-referred traffic to US retail sites was up 138% year-on-year in May 2026, converting 54% better than non-AI traffic (Adobe Analytics, via Digital Commerce 360).

Is agentic checkout already live in fashion?

Yes. On 2 June 2026, Hey Savi and PayPal launched the UK's first agentic commerce platform with native in-app checkout, with Debenhams Group as first retail adopter. Salesforce reports AI agents influenced 20% of global online retail sales — around $262bn — over the 2025 holiday season. True Fit launched an agentic fit experience over Model Context Protocol on 17 February 2026.

Do consumers actually trust AI to make purchase decisions?

Not yet. A Gartner survey (January 2026) found willingness to let AI narrow choices reaching 31% for household supplies and 28% for electronics, but willingness to let AI make the final purchase call topped out at just 11%. 54% said they had to double-check AI-provided shopping information; 62% said it wasted their time.

Why can't AI shopping agents tell whether a garment will fit?

Because fit is not published as structured, machine-readable data on most fashion product pages. Agents already parse price, stock and attributes as structured fields; fit exists only as size-chart prose, if at all — the fit legibility gap.

What does "fit legibility" mean and how does a merchant fix it?

Whether a store's fit and sizing data exists in a form a machine can read and act on, not just a form a human can read. Fixing it means structured per-variant fit data, agent-callable endpoints, a hosted MCP interface, and a fit signal grounded in real outcomes rather than a static size chart.

Sources

Figures are drawn from the named sources and current to the dates stated above; several are 2025 holiday-season or early-2026 figures and will be updated as newer data is published. Last updated 13 July 2026.

Make the one thing agents can't read, legible

Rendered Fits publishes a machine-readable fit layer on your own domain and renders the shopper photorealistically in your actual garments — Shopify-native, live on your product page in under an hour.

Book a demo Try the live demo

Related: How to make your Shopify store AI-shopper-ready · The AI-shopper-ready checklist · For AI agents & developers · Agentic commerce hits UK fashion retail · Virtual try-on & fashion returns benchmark