Four New Audit Checks: Ready for Agentic Commerce and GEO

What's new?

The way people shop and research is increasingly shifting to AI systems like ChatGPT, Perplexity, and Gemini. These systems only recommend what they can clearly identify, compare, and evaluate. This is exactly where four new checks come in, which are now part of every audit.

If you have already audited your website: your previous results remain unchanged and comparable. The new checks will appear automatically during your next audit.

1. Product Identifiers

AI-powered shopping assistants match products using unique identifiers such as GTIN, MPN, or SKU, as well as the brand. This is the only way a product can be identified beyond doubt, compared with offers from other retailers, and recommended correctly. The check verifies whether these fields are present in your Product schema – on the homepage or on a recognized product detail page.

2. Complete Offer Data

An agent can only recommend a product and lead to a transaction if the offer displays price, currency (priceCurrency), and availability (availability) in a machine-readable format. This check evaluates the completeness of your Offer data – the foundation for transactional AI visibility in the emerging Agentic Commerce.

3. sameAs Entity Linking

Before an LLM or a search engine cites your brand, it must resolve it as a unique entity. The sameAs property in the Organization schema explicitly links your website to authoritative sources such as Wikipedia/Wikidata, LinkedIn, or Crunchbase. This is the strongest machine-readable signal for entity resolution. With three or more linked profiles, the signal is considered solid.

4. Structured Content Formats

LLMs break pages down into individual passages. Clearly structured blocks – tables, numbered lists, definition lists – are self-contained answer units that can be adopted directly as citations and are suitable for typical comparison and step-by-step questions. For this purpose, the check specifically selects a content-rich page – preferably a guide or blog article, otherwise an FAQ page, and as a fallback, the homepage – and checks only the main content, excluding navigation elements.

Why this selection?

We have evaluated the current state of knowledge on AI ranking and Generative Engine Optimization (including the Princeton GEO paper) as well as leading industry tools. The biggest gap for shops is no longer classic "schema present yes/no", but machine-readable depth: Can an agent uniquely identify, evaluate, and recommend the product? The four new checks close exactly this gap – and complement our existing strengths in crawler access, reputation, and trust signals.

What you can do now

• Add GTIN/MPN/SKU and the brand to your Product schema.
• Ensure that your Offer data includes price, currency, and availability.
• Include sameAs in your Organization schema with at least three authoritative profiles.
• Structure core content as tables or lists instead of pure body text.

Simply start a new audit – the four new checks are immediately part of your evaluation.