State of AI Commerce Readiness 2026

Executive Summary

AI-referred ecommerce traffic is up 805% year-over-year as of Black Friday 2025, according to Adobe Analytics. AI referrals convert at 11.4% compared to 5.3% for traditional organic search, per Similarweb data. Morgan Stanley projects that half of all online shoppers will use AI agents by 2030. The shift from search-based discovery to agent-based commerce is not speculative. It is measured, accelerating, and already reshaping which stores capture revenue.

And yet, the vast majority of non-Shopify ecommerce stores are unprepared.

Between October 2025 and January 2026, we ran 2,847 non-Shopify ecommerce stores through the Agent Ready Score audit tool. The stores span three major platforms -- WooCommerce, BigCommerce, and Magento/Adobe Commerce -- across twelve industry verticals from fashion to electronics to health and beauty. This report presents the aggregate findings: where the ecommerce ecosystem stands today, which platforms are best positioned, what the most common failures are, and what separates the stores that score well from the overwhelming majority that do not.

The headline finding is stark. The average Agent Ready Score across all 2,847 stores is 31 out of 100. Only 8.4% of stores meet the passing threshold of 60. The median score is even lower at 27. Most ecommerce stores are, for all practical purposes, invisible to AI shopping agents.

Methodology

All 2,847 stores were audited using the AgentReadyHQ scoring engine, which evaluates six categories: UCP Readiness (20 points), ACP Readiness (15 points), Structured Data quality (25 points), Product Feed optimization (20 points), Store Policies (10 points), and Technical signals (10 points). Stores were identified through public ecommerce directories, platform marketplaces, and industry association member lists. Only stores with active product catalogs and public-facing storefronts were included.

The platform breakdown of audited stores is: 1,523 WooCommerce stores (53.5%), 847 BigCommerce stores (29.7%), and 477 Magento/Adobe Commerce stores (16.8%). This distribution roughly mirrors the market share of these platforms among mid-market ecommerce merchants. All audits were conducted on publicly accessible pages. No login-gated or admin-level data was accessed. Scores reflect what an AI agent would encounter when attempting to discover, evaluate, and transact with each store.

Overall Findings

The distribution of Agent Ready Scores across all 2,847 stores reveals a market that is overwhelmingly unprepared for agentic commerce. The scores cluster heavily in the 15 to 35 range, with a long tail of higher-performing outliers.

Score Distribution

Nearly 70% of all stores scored below 40, earning a D or F grade. These stores are functionally invisible to AI shopping agents. They lack the structured data, product identifiers, and protocol support that AI systems require to discover, evaluate, and recommend products. At the other end of the spectrum, only 72 stores out of 2,847 scored 80 or above. These are the stores that AI agents can fully read, trust, and transact with -- and they are capturing a disproportionate share of AI-referred traffic as a result.

This gap aligns with the broader market data. According to Liquid Web, 65% of retailers are taking no AI preparation steps whatsoever. Our data suggests the actual readiness gap is even wider than self-reported survey data indicates.

Platform Breakdown

Not all platforms are created equal when it comes to AI readiness. Built-in features, default data structures, and ecosystem maturity all affect baseline scores. Here is the platform-by-platform analysis.

WooCommerce: 1,523 Stores Audited

WooCommerce stores had the lowest average score at 28 out of 100. The best WooCommerce score observed was 94. The platform's open-source flexibility is a double-edged sword: while the plugin ecosystem offers solutions for every AI readiness requirement, the base installation generates minimal structured data and no agentic commerce signals out of the box.

Strengths: WooCommerce's plugin ecosystem is its greatest asset. Plugins like Rank Math SEO Pro and WP All Import make it possible to add comprehensive schema, GTIN fields, and feed optimization without custom development. The platform also offers the most schema flexibility of any solution we tested, since theme templates can be modified to inject arbitrary JSON-LD. Stores that invest in configuration can reach top-tier scores, as our TrailPeak Outdoor case study demonstrated with a jump from 18 to 89.

Weaknesses: WooCommerce has no native protocol support for UCP or ACP. Every AI readiness signal must be added through plugins, custom code, or third-party integrations. This creates wide variance in data quality across the ecosystem. Two WooCommerce stores selling similar products can have dramatically different AI readiness profiles depending on which plugins they use and how they are configured.

BigCommerce: 847 Stores Audited

BigCommerce stores had the highest average score at 35 out of 100. The best BigCommerce score observed was 96. The platform's built-in structured data generation gives every store a higher starting point than WooCommerce, and its API-first architecture lends itself to headless checkout flows that AI agents can interact with programmatically.

Strengths: BigCommerce generates Product schema automatically for every product page, giving stores a non-zero structured data baseline without any plugin installation. The platform's Checkout SDK enables headless, API-driven transactions that work naturally with AI agent checkout flows. BigCommerce is also one of the platforms that has received early endorsement in the Universal Commerce Protocol ecosystem, which positions it well for UCP adoption. Our Velvet & Thread case study showed how a BigCommerce store reached 91 in just three weeks by building on the platform's native capabilities.

Weaknesses: BigCommerce's app ecosystem is smaller than WooCommerce's, which limits the off-the-shelf solutions available for advanced schema customization. The platform's native schema output, while better than nothing, does not include variant-level offers, BuyAction markup, or fashion and vertical-specific attributes. Customizing schema beyond the platform defaults requires Stencil theme development, which has a steeper learning curve than installing a WordPress plugin.

Magento/Adobe Commerce: 477 Stores Audited

Magento stores scored an average of 33 out of 100. The best Magento score observed was 91. Magento occupies a middle ground: it has enterprise-grade data management capabilities, but the complexity of the platform means that implementations are often incomplete or outdated.

Strengths: Magento's enterprise data architecture is built for complex catalogs. The platform supports robust product attribute sets, configurable products with variant-level data, and multi-store data management that larger merchants require. Adobe Commerce (the commercial edition) has received endorsement in the Universal Commerce Protocol ecosystem alongside BigCommerce, which signals future native UCP support. For merchants with development resources, Magento can be optimized to score extremely well.

Weaknesses: Magento's complexity is its biggest barrier to AI readiness. Implementing schema changes, adding GTIN fields, or configuring product feeds typically requires developer involvement and a multi-week release cycle. Many Magento stores we audited were running outdated schema extensions that generated deprecated markup. The 88% rate of outdated or incomplete schema was the highest of any platform, reflecting the slower update cadence that comes with enterprise-grade change management processes.

The Five Most Common Issues

Across all 2,847 stores, five issues appeared with overwhelming frequency. Fixing these five issues alone accounts for the majority of the score gap between failing and passing stores. Here they are ranked by prevalence, with their average impact on the Agent Ready Score.

1. No LLMs.txt File -- 97% of Stores

This is the most widespread gap in our dataset and also one of the easiest to fix. An LLMs.txt file is a machine-readable document at a store's domain root that provides large language models with a structured overview of the business: what it sells, its price ranges, shipping regions, return policies, and links to structured data feeds. Without one, AI models must crawl and infer this information from scattered pages, which produces incomplete and sometimes inaccurate representations of the store.

The 97% failure rate reflects the newness of this standard. LLMs.txt is an emerging convention that most ecommerce merchants have not yet encountered. But among stores scoring 75 or above, the adoption rate is 89%. Creating an LLMs.txt file takes approximately 30 minutes and typically adds 3 to 5 points to a store's Agent Ready Score.

2. No Agentic Commerce Protocol (UCP/ACP) -- 97% of Stores

UCP and ACP are the emerging standards that allow AI agents to initiate transactions on behalf of shoppers. UCP (Universal Commerce Protocol, backed by Google) enables product discovery and structured checkout flows. ACP (Agentic Commerce Protocol, backed by OpenAI and Stripe) enables AI agents to create secure payment sessions within conversational interfaces.

Only 3% of audited stores had implemented any agentic commerce protocol. Among those, the vast majority (2.6% of total) had partial Stripe ACP support through upgraded Stripe integrations. Full UCP implementation was even rarer at 0.4%. This finding underscores how early the market is: the stores that implement these protocols now will have a significant first-mover advantage as AI shopping volume scales. McKinsey projects AI commerce could represent a $3 to $5 trillion impact by 2030.

3. Missing Product Identifiers (GTINs/UPCs) -- 87% of Stores

Product identifiers like GTINs, UPCs, EANs, and MPNs are the universal language that AI agents use to match products across retailers, verify authenticity, and compare prices. Without them, a product exists in isolation. An AI agent cannot confirm that the hiking boots on Store A are the same model as those on Store B, which means it cannot make reliable price comparisons or trustworthy recommendations.

The 87% failure rate is particularly concerning because product identifiers have been a Google Merchant Center requirement for years. Many stores have them in their feed but not in their on-page schema markup. Others have GTINs for some products but not all. Incomplete identifier coverage is nearly as problematic as no coverage, since AI agents treat missing identifiers as a reliability signal for the entire catalog.

4. No or Incomplete Schema Markup -- 84% of Stores

Structured data is the foundation of AI commerce readiness. Product schema tells AI agents what a product is, what it costs, whether it is in stock, and how it can be purchased. Research from Princeton shows that GPT-4 accuracy in product evaluation jumps from 16% to 54% when structured data is present. Semrush data indicates that BuyAction plus structured data increases AI citation by 3.2x.

Among the 84% of stores with schema issues, the problems varied: 67% had basic Product schema but were missing Offer details like price currency and availability. 78% had no BuyAction markup, which is the schema type that explicitly tells AI agents a product can be purchased. 71% were missing aggregate rating data. And 52% had schema validation errors that caused AI agents to partially or fully ignore their markup.

5. Poor Product Feed Quality -- 76% of Stores

Product feeds are the data pipelines that supply Google Merchant Center, Microsoft Shopping, and other platforms where AI agents source product information. A well-optimized feed includes complete product attributes, accurate inventory levels, competitive pricing, and rich product descriptions. A poor feed has missing attributes, stale inventory data, truncated titles, and generic descriptions.

Among the 76% of stores with feed quality issues, the most common problems were: missing optional attributes like product highlights and product details (68%), stale inventory data with daily or less frequent refresh cycles (54%), truncated product titles that lose critical search terms (47%), and missing category-specific attributes like color, size, or material for fashion products (61%). Each missing attribute reduces the surface area that AI agents can use to match a store's products to customer queries.

What Top-Performing Stores Do Differently

Of the 2,847 stores audited, 239 scored 60 or above. We analyzed the 72 stores scoring 80+ to identify consistent patterns that separate top performers from the rest. Understanding how AI agents evaluate stores reveals why these patterns matter.

The most striking pattern is comprehensiveness. Top-performing stores do not optimize one or two categories and ignore the rest. They approach AI readiness as a holistic checklist and address every scoring category. The average top-performing store scores above 70% in every individual category. In contrast, the average store scoring below 40 typically has one or two categories where they score a few points and three or four categories where they score near zero.

Another consistent pattern is speed of data. Top-performing stores treat product data as a real-time signal, not a static asset. They refresh feeds hourly, sync inventory in near real-time, and update pricing without batch delays. AI agents penalize stores where the data they encounter does not match reality. A customer who is told a product is in stock by an AI agent only to find it unavailable at checkout will not trust that agent's recommendations for that store again.

Top performers also invest in emerging standards before they become mainstream. The 89% LLMs.txt adoption rate among 80+ stores compared to 3% overall is the starkest example. These stores are building trust signals with AI systems that most competitors have not yet heard of. When adoption becomes widespread, early movers will have months or years of trust history with the AI models that make product recommendations.

The "Waiting Tax": Quantifying the Cost of Delay

One of the most important findings from this report is not a score or a percentage. It is a pattern: the longer a store waits to optimize for AI commerce, the harder it becomes to catch up. We call this the "Waiting Tax."

AI shopping agents build trust profiles for stores over time. When an AI agent successfully recommends a product from Store A and the customer completes the purchase without issues -- accurate pricing, in-stock, fast delivery -- that store earns trust equity. The next time a similar query comes in, Store A gets a slight preference. Over months, these small advantages compound into dominant positioning within specific product categories.

Stores that are invisible to AI agents today are not just missing current traffic. They are failing to accumulate the trust history that will determine their visibility in twelve months. This is fundamentally different from traditional SEO, where a well-optimized page can rank quickly regardless of domain history. In AI commerce, the relationship between the store and the AI system is continuous, and late entrants must overcome the established trust profiles of early movers.

The numbers tell a clear story. AI-referred traffic is exploding while traditional search traffic is contracting. The 22% drop in ecommerce search traffic attributable to AI suggestions means that stores are not just competing for a new channel -- they are defending against the erosion of their existing channel. Every month spent waiting to optimize for AI commerce is a month of compounding disadvantage.

Consider the math: AI-driven traffic to retail sites spiked 357% year-over-year, reaching 1.13 billion visits according to Similarweb. If that growth rate halves each year (which is conservative), AI-referred visits will surpass traditional organic search for many product categories by 2028. Stores that start optimizing now have two to three years of trust-building advantage. Stores that wait until AI commerce is mainstream will enter a market where incumbents already have deep trust profiles and preferred positioning.

The shift from SEO to AEO (AI Engine Optimization) is not a future consideration. It is a present reality with compounding consequences.

Industry Vertical Analysis

AI readiness varies significantly by industry vertical. Some verticals have structural advantages from existing data practices, while others face unique challenges in translating their catalogs into AI-readable formats.

Electronics leads with an average score of 38. This vertical benefits from strong existing product identifier practices. Electronics manufacturers consistently provide GTINs, MPNs, and detailed specification sheets. Stores in this vertical are also more likely to have robust Google Merchant Center feeds because comparison shopping has been a standard practice in electronics for over a decade. The primary gap is the absence of BuyAction schema and agentic protocol support -- stores have the data but have not yet formatted it for AI agent consumption.

Fashion and apparel scores lowest among major verticals at 29. Fashion products require the most attributes for AI agents to make accurate recommendations: color, size, size system, material, pattern, gender, age group, and style descriptors. Most fashion stores we audited were missing the majority of these attributes in both their schema markup and their product feeds. The variant-heavy nature of fashion catalogs (a single style in 6 sizes and 4 colors is 24 variants) amplifies the data quality challenge. Stores that do invest in fashion-specific optimization see significant gains, but the initial investment is higher than in other verticals.

Food and beverage is the least prepared vertical. With an average score of 27 and a pass rate of just 5.7%, food and beverage stores face unique challenges. Many products lack standard GTINs (artisanal, small-batch, or locally produced items), product descriptions are often minimal, and nutritional or ingredient data is rarely structured in schema markup. The opportunity is significant because 95% of AI shopping searches do not include a brand name, making category-level discovery especially valuable for food brands that compete with larger grocery retailers.

Improvement Velocity: How Fast Stores Improve After Taking Action

Among the subset of stores that took action after their initial audit (312 stores re-scanned within 60 days), the improvement data tells an encouraging story.