The search landscape is changing dramatically. More and more users aren't typing into Google—they're asking ChatGPT, Gemini, or Claude directly for product recommendations. If your e-commerce site isn't optimized for these AI agents to understand and recommend products, you're simply invisible to that 70% of new shoppers.
The data doesn't lie: only 32% of e-commerce businesses in Spain appear in AI search engine results, with an average visibility score of 39/100 according to Semrush. The rest are completely invisible.
But there are exceptions. Dortomedical, a specialist in orthopedic products, achieved 42/100 AI visibility—above the industry average. The result? Over 160 monthly mentions in AI chats and 88,900 users reached solely through AI-generated responses.
In this article we discuss:
- How to adapt your ecommerce so that AI can find you
- Dortomedical success story: From invisible to reference
- What your ecommerce site looks like when it's configured for Agent AI
- What's behind it: Technical architecture that AI understands
- Conclusions and next steps
1. How to adapt your ecommerce so that AI can find you
Optimization for AI engines (also known as GEO - Generative Engine Optimization or LLMO - Large Language Model Optimization) is fundamentally different from traditional SEO. It's not just about keywords or backlinks. It's about enabling AI agents to autonomously read, understand, and recommend your product catalog.
How do you know if you're doing it right?
- Your website needs to convert when users arrive from AI search engines: those coming from Google's ChatGPT or AI Mode are already highly qualified. The AI has already specifically recommended your product to them.
- AI needs to understand what you're selling: AI doesn't "see" your website like a human. It doesn't interpret pretty pictures or creative designs. It reads structured code.
- Get recommendations when comparing catalogs: When a user asks ChatGPT: "Where to buy quality orthotic insoles in Spain?", the AI compares catalogs in milliseconds.
- Appearing in direct replies: The holy grail of visibility in AI is the direct reply, when ChatGPT explicitly says, "I recommend this product because..."
- You have verifiable data and sources: Your content is backed by priority (verified reviews, mentioned certifications, cited studies and data, explicit guarantees, etc.)
2. Dortomedical success story: From invisible to reference
Who is Dortomedical?
Dortomedical is a Spanish ecommerce company specializing in orthopedic and health products: insoles, lumbar supports, knee braces, products for reduced mobility, and technical aids for everyday life.
How did they know they had to change their strategy?
As specialist SEO and GEO consultants, we identified a paradigm shift: while traditional SEO was becoming saturated, generative search engines were starting to recommend specific competitors in high purchase intent queries.
Dortomedical was losing a massive, albeit invisible, market share. We presented a strategic proposal to transform our website from "human-readable " into an " authoritative source for AI ." The client, aware that the future of e-commerce lies not in waiting for clicks but in proactively generating them, immediately accepted the challenge.
Strategy implemented: The architecture of success
To go from 0 to 42/100 visibility in just four months, we implemented a roadmap based on three technical pillars:
- AI-Ready Architecture Optimization: We perform deep code cleaning to eliminate technical noise. LLMs don't just browse, they process. If the code is dirty, AI ignores you.
- Structured Data Engineering: We implement advanced data schemas (JSON-LD) so that models like ChatGPT and Gemini understand without error what we sell, what certifications we have, and why we are the best option.
- Transactional and Semantic Content: We restructure content not only to rank for keywords, but also to address the user's search intent. We focus on clarifying critical doubts in the final stage of the funnel, enabling AI to say: " I recommend this Dortomedical product because it solves your exact problem ."
The results: the impact in numbers
- Massive reach: 88,900 users reached exclusively through AI-generated responses.
- Authority in chats: +480 monthly mentions on leading platforms such as ChatGPT and Gemini.
- Segmented visibility: We achieved a dominant presence in ChatGPT (179 mentions) and Gemini (168 mentions), placing us above the industry average.
3. What your ecommerce looks like when it's configured for Agent AI
An AI-adapted ecommerce site ceases to be a " static " website and becomes a trusted source for ranking in LLMs and multiplying your mentions.
This is how you can see those direct results:
- Guided, not solicited, sales: AI doesn't make random decisions. It decides autonomously and with total precision regarding the user's needs, offering a personalized proposal in a matter of seconds.
- Efficient Marketplace Management: If a product runs out of stock on Amazon, but is still available on your website, AI rebalances the stock and automatically adjusts prices to avoid losing that sale.
- Customer loyalty: AI ensures customers don't forget you. It detects patterns of interest and purchases, and delivers the perfect offers at the moment of peak purchase intent.
4. What's behind it: Technical architecture that AIs understand
For a language model (LLM) like ChatGPT or Gemini to recommend your products, simply having "good content" isn't enough. You need an infrastructure that facilitates crawling and data inference.
A. Second Generation Structured Data (JSON-LD)
AI doesn't "read" the web; it processes nodes of information. We've moved from basic labeling to hyper-specific labeling:
- Entity Linking: We connect each product (e.g., "patellar knee brace") with entities in Wikidata and DBpedia so that AI understands exactly what the product is, beyond the brand name.
- Trust Attributes: We implemented Review, Brand, Organization, and AggregateRating schemes. This allows AI to see that the product has user endorsements and verifiable certifications, a key factor in being the recommended option.
B. Conversational Content and RAG-Ready
LLMs use a technique called RAG (Retrieval-Augmented Generation). If your content is ambiguous, AI won't use it.
- Rich semantics: We write product descriptions focused on resolving doubts ("Is it suitable for sensitive skin?" or "How long should I use it per day?").
- Readable formats: Use of technical tables and lists of direct benefits that AI agents can extract in milliseconds to compare catalogs.
C. Preparation for Model Context Protocol (MCP) and APIs
The future of search is agentic: AI not only searches, but also performs actions.
- Optimized feeds: We create dynamic data sources that feed AI agents directly with real-time stock and prices.
- Technical documentation: The web routes were optimized so that AI "crawlers" (such as GPTBot) find the most relevant information first (FAQ, manuals, warranties), reducing the processing cost for the engine.
D. Performance and Technical Crawlability
Speed is not just for the user; it's for the efficiency of the generative search engine.
- Code cleanup: We remove unnecessary scripts and redundancies in the HTML so that the plain text (what the AI actually consumes) is clear and prioritized.
- Semantic navigation: URL structure and internal linking based on logical categories that help AI to autonomously map the entire Dortomedical inventory.
6. Conclusions
The Dortomedical case demonstrates that GEO is not a " future trend ," but an immediate competitive advantage. While most e-commerce businesses continue to optimize for a traditional search model, others are already capturing a new channel with better conversion rates and lower costs.
At Línea Gráfica, we don't just understand the code; we understand the business. We've spent over 20 years evolving alongside the market, and now we help companies translate their catalogs into the language of AI.
Is your brand invisible to AI, or is it already the recommended option? Don't guess. Let's talk.
