In this blog post
Organic traffic is falling at some brands by 40-70%, even as revenue holds steady or grows. The reason: a growing share of customers are reaching stores not through Google but through AI agents – ChatGPT, Gemini, Copilot. This article explains what Agentic Commerce is, how Shopify's Universal Commerce Protocol works, and what three things you can do today to make your store recommended by LLMs, not just found.
Key takeaways in 30 seconds (TL;DR)
- Agentic orders grew 15x year-on-year; average order value is 30% higher than from traditional channels.
- GEO (Generative Engine Optimization) means optimising to be recommended by LLMs – not just indexed by Google.
- Shopify's Universal Commerce Protocol is the new standard for agentic transactions, co-developed with Google, Amazon, and others.
- Three pillars of LLM visibility: structured product data, external user sentiment, and technical product page quality.
- Strong on-site SEO remains the foundation – GEO adds layers on top rather than replacing what already works.
- Brands that act now have a real first-mover advantage in categories where competitors have not arrived yet.
What is Agentic Commerce?
Agentic Commerce is purchasing done through AI in two ways: conversationally (the user asks an AI agent to find and buy something) and autonomously (an agent acts on the user's behalf, with permission, without active involvement at the moment of purchase).
Why it's already happening: Shopify data shows that orders attributed to agentic channels grew 15x between January 2025 and the end of Q1 2025. Average order value from agentic channels is 30% higher than from traditional channels. A customer arriving via an LLM recommendation isn't in the consideration phase – they've already been educated and are ready to buy.
What is GEO, and how does it differ from SEO?
GEO (Generative Engine Optimization) is optimising for visibility in generative AI engines: ChatGPT, Gemini, Claude, Copilot. Classic SEO optimised to be found. GEO optimises to be recommended.
LLMs partly rely on SEO signals – they often crawl pages in real time when answering queries. But when product data is available via API (through Universal Commerce Protocol), agents prioritise that structured data over crawled pages. Strong on-site SEO remains the foundation; GEO adds layers on top of it rather than replacing it.
Universal Commerce Protocol – the new standard for agentic commerce
Universal Commerce Protocol (UCP) is a communication protocol for agents to process commercial transactions. Where HTTP handles web pages, UCP handles the full agentic purchase journey: discovery, product data transfer, transaction processing. It was developed by Shopify together with Google, with Amazon, Etsy, and Walmart joining.
In practice: when a user finds a product in ChatGPT and decides to buy, they see your brand's checkout – exactly as it appears on your store – inside the chat window. Your business logic applies: payment methods, delivery options, discount codes. The order lands in your system like any other. Customer data stays yours.
Shopify Catalog and the Agentic Plan are available to stores not on Shopify. Merchants on WooCommerce, Magento, or BigCommerce can list products in Shopify Catalog without migrating. The cost is a commission on transactions completed through that channel only.
The three pillars of LLM visibility for your store
1. Structured product data
Data available via Shopify Catalog and UCP is the first and technically quickest layer to address. Required fields: title, price, image, URL, product ID. Recommended: specifications, use cases, awards and certifications, size guides as HTML tables (not JPEG images), product data in JSON-LD embedded in the page HTML. Content hidden under JavaScript-loaded accordions is less accessible to bots – expose it directly in HTML where possible.
2. User sentiment – reviews and external mentions
LLMs apply a trust hierarchy: highest weight goes to authentic user-generated sentiment (reviews, unboxing content, Reddit threads, industry forums). Rough weighting: 70% of influence comes from citations in credible external sources; 30% from a brand's own content. Context in reviews matters. A review that references a specific situation answers AI agent queries directly and gets cited.
3. Technical product page quality
LLMs process text better than multimedia. An inspirational product description serves conversion, but agents need it accompanied by a specification in table or list form, educational FAQs, use case descriptions, and awards data. JSON-LD structured data is a standard already built into most e-commerce platforms and CMSes.
Shopify Knowledge Base – controlling your brand's narrative
An AI agent doesn't necessarily draw only from your store – it may pull in information from external sources, including negative ones. Shopify Knowledge Base is an app inside the admin panel that lets you control how agents respond to queries about your brand.
👉 Watch the full conversation on YouTube
Frequently asked questions about Agentic Commerce
Does Agentic Commerce work for every product type?
Products with repeat purchase patterns (toothpaste, pet food, supplements) suit fully autonomous buying. Products requiring parameter comparison (electronics, sports equipment) sit squarely in the LLM advisory zone. Luxury and emotionally driven products are a weaker fit for agentic models – an AI chat on your own site is a better opportunity there.
My store runs on WooCommerce / Magento / BigCommerce – does this apply to me?
Yes. Shopify Catalog is open to any platform. You list your product data in the catalog without migrating; you only pay commission on transactions that complete through that channel.
Does this threaten my branding? What does the customer see at checkout?
When a customer buys through ChatGPT, they see your brand's checkout – with your logo, colours, discount logic, and payment methods. The order lands in your system. Customer data is yours.
How do I find out what LLMs are saying about my brand right now?
Start with a direct test: enter queries from different stages of the purchase journey into ChatGPT and Gemini. For ongoing monitoring: Brand Radar (Ahrefs), Brand24, and specialist LLM brand visibility tools.
Three steps to start – what to do this week
Step 1: Find out what LLMs are already saying about your brand and products.
Run queries at different funnel stages in ChatGPT and Gemini: top of funnel, middle, and bottom. Note which domains are being cited – that's your content and PR target list.
Step 2: Update your product data.
Add technical specifications, FAQs, use cases, awards, and certifications. Move size guides from images to HTML tables. Make sure key content is visible in HTML, not hidden behind JavaScript. Implement or audit JSON-LD. On Shopify: ask Sidekick to audit product data completeness.
Step 3: Run a technical crawlability audit.
Use a crawler-perspective tool to check which information bots can access and which is hidden. Prioritise product pages, category pages, and the homepage.
Key takeaways
Agentic orders grew 15x year-on-year; average order value is 30% higher than from traditional channels. GEO means optimising to be recommended by LLMs – not just indexed by Google. Universal Commerce Protocol is the new standard for agentic transactions, co-developed by Shopify with Google, Amazon, and others. Three pillars of LLM visibility: structured product data, external user sentiment, and technical product page quality. GEO work doesn't replace SEO – it's an additional layer on a solid foundation. Brands that act now have a real first-mover advantage.
