Case Study: How Casa Moderna Tripled Their AI-Referred Traffic on Magento

The Challenge

Casa Moderna Collective is a contemporary home decor retailer based in Denver, Colorado, specializing in furniture accents, artisanal home goods, and curated design pieces. They operate on Magento 2 (Adobe Commerce) with approximately 1,200 SKUs and generate around $1.2M in annual revenue. Their core audience is design-conscious homeowners and interior designers, primarily in the 30 to 55 age range, who value craftsmanship and modern aesthetics.

In Q4 2025, Casa Moderna's founder Sarah Chen noticed something concerning. Organic traffic had been declining steadily for two months, and competitors she had never heard of were appearing in AI-powered shopping results. She tested it herself: she asked ChatGPT to recommend contemporary home decor brands for a living room refresh. The results were a mix of major retailers and smaller direct-to-consumer brands — but Casa Moderna was nowhere. She tried Google AI Mode with queries like "artisanal ceramic vases under $150" and "modern accent furniture for small spaces." Again, nothing.

Casa Moderna had strong organic search rankings on traditional Google, a loyal customer base, and high-quality products at competitive price points. The problem was that AI agents could not parse, understand, or interact with their Magento store. They ran the site through the AgentReadyHQ scanner in January 2026 and received an initial score of 15 out of 100 (F). Despite running on Magento 2, a platform with deep customization capabilities, virtually none of the AI-critical signals were in place.

Initial Audit Results

Here is the complete breakdown of Casa Moderna's initial Agent Ready Score:

Magento 2 ships with basic product schema through its built-in Rich Snippets configuration, which accounted for the 8 points in Structured Data. But the output was a bare-bones Product type with only name, description, image, and price. It was missing individual Offer objects for product variants, had no AggregateRating from their 2,400+ customer reviews, included zero GTIN or MPN identifiers across all 1,200 products, and had no BuyAction markup. The product feed had been set up for Google Shopping two years prior but had not been updated to include the attributes AI agents need for product matching. There was no LLMs.txt file, no agentic protocol implementation, and no structured policy pages.

The good news: Magento's architecture is built for extensibility. Unlike hosted platforms with limited customization, Magento gives merchants full control over schema output, API endpoints, and frontend templates. The challenge was knowing exactly what to build.

Week 1 — Foundation: Schema & Product Data (Score: 15 → 32)

The first week targeted the highest-impact gaps: structured data and product identifiers. These are the signals AI agents rely on most heavily when deciding whether to index and recommend a store's products.

1. Installed and Configured an Enhanced Schema Module

Casa Moderna installed the Amasty Rich Snippets extension for Magento 2, which replaces the platform's default schema output with a far more comprehensive implementation. The module generates complete Product schema with nested Offer objects for every configurable product variant, includes AggregateRating data pulled from Magento's native review system, adds brand, SKU, and identifier fields, and supports BuyAction markup. They configured the extension to output JSON-LD on every product page, every category page, and the homepage.

Beyond the extension, Casa Moderna's developer customized the schema output to include home-decor-specific attributes: material composition, dimensions, color family, room type, and style category. These custom attributes were already stored as Magento product attributes but were not being surfaced in the structured data. A small custom module mapped these attributes into the JSON-LD output. For details on implementing comprehensive schema markup for AI shopping, see our technical guide.

2. Began GTIN Assignment for Top Products

Zero percent of Casa Moderna's 1,200 products had GTINs assigned. This is common among home decor retailers, particularly those carrying artisanal and small-batch goods where manufacturers do not always provide UPC or EAN codes. The team started with their top 200 products by revenue, contacting manufacturers to source GTINs and registering their own GS1 prefix for private-label items. Magento's product attribute system made bulk assignment straightforward through CSV import. For products without manufacturer GTINs, they used MPN values and set the identifier_exists field to false in their feeds. The complete guide to GTIN product identifiers covers this process in depth.

3. Added AggregateRating to Product Schema

Casa Moderna had 2,400 customer reviews across their catalog, averaging 4.6 stars. None of this data was being surfaced in their structured data. They configured the Amasty module to pull review count and average rating directly from Magento's review tables and embed AggregateRating within each Product schema object. AI agents weigh review data heavily when making product recommendations, and 2,400 legitimate reviews across 1,200 products is a strong signal of a trustworthy merchant.

Week 2 — AI Visibility: LLMs.txt & Feed Optimization (Score: 32 → 55)

With the structured data foundation in place, week two focused on making Casa Moderna visible to AI models and expanding their product feed reach.

1. Created and Deployed LLMs.txt

Casa Moderna added an LLMs.txt file at their domain root that provides a machine-readable overview of their store for large language models. The file describes their product categories (furniture accents, wall art, ceramics and pottery, lighting, textiles, and decorative objects), price ranges ($35 to $1,200), shipping regions (continental US with flat-rate shipping), their artisan sourcing story, and links to their structured data feeds and API endpoints. For home decor stores, category descriptions and material information are particularly valuable because AI agents need to distinguish between mass-produced and artisanal goods when matching customer preferences.

2. Restructured Product Feeds for Multi-Platform Distribution

Casa Moderna's existing Google Shopping feed was two years old, contained only basic product data, and had not been updated with new products added in the past six months. They rebuilt the feed from scratch using a Magento feed extension and configured outputs for three platforms: Google Merchant Center, Microsoft Merchant Center, and the ChatGPT Shopping feed format. Each feed included the full set of AI-optimized product attributes: GTINs (for the 200 products that had them so far), brand, material, dimensions, color, room type, style category, product highlights, and high-resolution image URLs. Expanding from one outdated feed to three current, attribute-rich feeds immediately broadened their AI visibility.

3. Optimized Product Descriptions for AI Readability

Many of Casa Moderna's product descriptions were written in a lifestyle-marketing style that works well for human browsing but poorly for AI parsing. Descriptions like "Transform your space with the warmth of handcrafted beauty" tell an AI agent nothing about what the product actually is. The team rewrote the top 200 product descriptions to lead with factual, attribute-rich information: material, dimensions, color, intended room, and care instructions. They kept the evocative marketing language but moved it after the factual details. AI agents parse the first 200 characters of a product description most heavily, so front-loading structured information is critical.

Week 3 — Protocol Integration (Score: 55 → 71)

Week three addressed the agentic commerce protocols that allow AI agents to not just discover products but initiate transactions on behalf of customers.

1. Began UCP Implementation

Adobe Commerce has publicly endorsed the Universal Commerce Protocol (UCP), signaling that Magento merchants should prioritize this standard. Casa Moderna's developer implemented UCP endpoints that expose product catalog data, inventory levels, and pricing through a standardized API that AI agents from Google and other platforms can consume. Magento's REST and GraphQL APIs provided a solid foundation, and the UCP layer was built as a custom module that translates Magento's native API responses into the UCP-compliant format. This is where Magento's extensibility pays off: the platform's module architecture makes protocol integration cleaner than on hosted platforms where you cannot modify server-side code.

2. Added BuyAction Schema to Top Products

Beyond standard Product schema, Casa Moderna added BuyAction markup to their top-selling products. BuyAction explicitly tells AI agents that a product can be purchased directly, including the target URL for initiating a transaction and the accepted payment methods. They implemented this through a Magento layout XML update that injects a BuyAction JSON-LD block on product pages for items flagged as "AI-priority" in the admin panel. The flag was applied to their top 300 products by revenue and margin.

3. Created Structured Policy Pages

Casa Moderna had shipping and return information buried in a single FAQ page with no schema markup. They created dedicated CMS pages in Magento with structured data:

• /shipping-policy with OfferShippingDetails schema, clear delivery timeframes, and flat-rate pricing
• /return-policy with MerchantReturnPolicy schema, their 30-day return window, and free return shipping terms
• /warranty with product warranty details for furniture items

These pages were linked from the site footer, from individual product pages via a custom Magento block, and referenced in the LLMs.txt file. AI agents use policy information to build trust scores for merchants and to answer customer questions about shipping and returns during conversational shopping sessions.

Week 4 — Optimization & Testing (Score: 71 → 82)

The final week focused on extending coverage across the full catalog, integrating the remaining agentic checkout protocol, and validating that AI agents were actually finding and recommending Casa Moderna's products.

1. Extended GTIN Coverage to All 1,200 Products

The team completed GTIN assignment across the entire catalog. For artisanal products sourced from small makers without standard identifiers, they registered a GS1 Company Prefix and assigned GTINs through the GS1 system. This $250 annual investment gave them the ability to generate legitimate GTINs for any private-label or exclusive product. With full GTIN coverage, every product in their Magento catalog could be uniquely identified by AI agents across all shopping platforms.

2. Fine-Tuned Schema Based on AI Agent Testing

Casa Moderna tested their products against ChatGPT, Google AI Mode, and Perplexity Shopping. They searched for their own product categories and tracked which products appeared and which did not. Products missing from results were analyzed for schema gaps: some were missing material attributes, others had incomplete dimensions, and a handful had pricing data that was not being pulled into the Offer objects correctly. Each issue was fixed in the Magento admin and the schema module configuration. This iterative testing-and-fixing cycle is essential because different AI agents weight different signals.

3. Began Stripe ACP Integration

Casa Moderna processed payments through Stripe via Magento's payment gateway. They upgraded to the latest Stripe API version and enabled Agentic Commerce Protocol (ACP) features in their Stripe dashboard. This allows AI agents like ChatGPT to initiate secure checkout sessions where the customer confirms payment within the AI interface. The Magento integration required a custom payment method module that handles the agent-initiated session flow alongside the standard storefront checkout. This is more involved than on Shopify, where Stripe ACP is a toggle, but it also gives Magento merchants more control over the checkout experience.

4. Validated AI Visibility Across Platforms

By the end of week four, Casa Moderna's products were appearing in both ChatGPT shopping recommendations and Google AI Mode results. When Sarah searched for "modern ceramic table lamp under $200" on ChatGPT, Casa Moderna's Kyoto Table Lamp appeared alongside products from West Elm and CB2. On Google AI Mode, queries like "artisanal throw pillows handmade" surfaced three of their textile products. The shift from zero AI visibility to active product recommendations happened within the span of a single month.

Results After 30 Days

Casa Moderna completed all optimizations by the end of January 2026. Here are the measurable results after 30 days of the changes being live: