In short: More and more buying decisions start in ChatGPT, Perplexity or Copilot instead of Google. Whether your store gets recommended there is decided not by your ad budget, but by whether AI systems can find, read and trust your store. That is exactly what the AI Readability Scanner by Shopware and the Agentic Commerce Alliance tests on three layers: AI Discovery, Catalog Readability and Transaction Readiness. The good news: an existing Shopware store does not need to be rebuilt for this - but it does need to be technically prepared.

Visibility is shifting: from marketing to machine readability

For twenty years, visibility meant ranking on Google. We all know the tools - keywords, backlinks, content. AI search plays by different rules: an AI does not click through your menu and is not convinced by a beautiful hero banner. It reads structured signals - and what it cannot read does not exist for it.

That is the core thesis behind the AI Readability Scanner that Shopware launched together with the Agentic Commerce Alliance (an alliance for open standards in agent-based commerce): visibility is moving from marketing to machine readability. The scanner asks a current AI real questions about your store and measures what gets through - including a 0-100 score and a comparison against hundreds of other tested stores.

The three layers of AI readability

Layer 1: AI Discovery - is your store found at all?

Can AI agents reach your product URLs, sitemaps and content without obstacles? Surprisingly many stores fail at the basics here:

  • A robots.txt that blocks AI crawlers (GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot) across the board - often unintentionally through old "block all bots" rules or bot protection at the CDN level.
  • Sitemaps that are stale or full of variant chaos.
  • Content that only loads via JavaScript - many AI crawlers do not render JavaScript. The classic server-side rendered Shopware storefront has an advantage here; headless setups need to be careful.

Layer 2: Catalog Readability - does the AI understand your products?

Can an AI system parse and compare your products? This is structured data territory:

  • Product schema (JSON-LD) with price, availability, GTIN, ratings - since 6.7.9 Shopware supports JSON-LD natively behind a feature flag.
  • Product feeds: the Agentic Commerce sales channel from 6.7.10 delivers products as JSONL directly to AI platforms.
  • Attributes instead of prose: everything from my product data article pays in one-to-one here.

Layer 3: Transaction Readiness - can the AI stand behind a recommendation?

The underrated layer: before an AI recommends your store or an agent buys there, it checks trust and transaction signals - shipping and return policies, legal texts, reviews, clear pricing, clean checkout signals. Looking ahead, this is also where the emerging agent protocols belong (such as the Agentic Checkout Protocol around OpenAI or MCP-based integrations) through which agents interact with stores directly. That is still early - but the direction is set, and the Alliance is working on exactly these open standards.

What you can concretely do in Shopware - in this order

  1. Run the scanner. The test at agenticcommerce.shop is free and takes under a minute. You get a score, subscores and prioritized measures - there is currently no better free baseline assessment.
  2. Check crawl access: robots.txt, bot protection, CDN rules. Allow AI crawlers explicitly instead of hoping for default behavior.
  3. Activate JSON-LD (6.7.9 feature flag, tested on staging) and watch for consistency between schema, page and feed.
  4. Set up product feeds: get the Merchant Center feed clean, evaluate the Agentic Commerce channel.
  5. Complete trust signals: policy pages linked machine-readably, reviews embedded, prices and availability consistent.
  6. Scan again and measure. AI readability is not a one-off project but a feedback loop - just like SEO.

My assessment

You can dismiss all of this as hype. But the parallel to the early days of SEO is striking: back then, structured sitemaps and clean meta data also looked like pedantry - until the stores that were clean early lived off their head start for years. The difference today: the tools are there from the beginning, and your shop system already ships the necessary features. What is usually missing is just the consistent technical execution - and for an existing Shopware store, that is manageable. I will show what this looks like in a real project in an upcoming case study.

Tags:

Shopware Shopware 6 AI in E-Commerce SEO Agentic Commerce