The shift to agentic commerce
AI can help shoppers find products, compare options and, in supported experiences, complete a purchase. Those are different steps with different requirements.
A useful starting point for your store is accurate, accessible product information. Explain what each product is, which needs it meets and how to buy it.
Shopping systems can draw on product pages, structured information, feeds and other sources. The mix varies by service. ChatGPT, for example, describes using structured metadata and third-party content, alongside the shopper's question and context. Read OpenAI's explanation.
Better product information gives you a stronger basis for testing. It does not guarantee a recommendation or a sale.
What information helps shoppers compare products?
Start with the facts needed to identify, compare and buy the item. Keep them consistent across your catalog, storefront and relevant feeds.
For Google, product-page markup and Merchant Center feeds can both supply product information. Google recommends using both where available.
Use the product-data quality checklist for a field-by-field review. Keep prices and availability current because accuracy matters, rather than changing dates to imply freshness.
Where shoppers use AI to find products
These experiences do different jobs. Product discovery, a cited research answer and an agent completing a task are not interchangeable.
Check the provider's current availability before planning around a feature. Country, account and merchant eligibility can affect access.
Match your product information to the shopping question
Use questions customers actually ask. These illustrative examples show which facts to check; they are not observed AI results.
If a fact is missing, confirm it from a reliable product source before adding it. Do not turn an assumption into a product claim.
Prepare your store for AI shopping
Start with accessible pages, accurate product information and consistent feeds. Structured data helps describe the product; it does not guarantee that an AI service recommends it.
Google says no special AI text file or special schema is required for its Search AI features. Follow its documented crawl, indexing and content requirements. See Google's guidance.
Keep discovery files such as llms.txt separate from crawler controls and checkout integrations. A file's presence does not prove that a provider uses it.
Commerce protocols have their own integration and eligibility requirements. For example, Google's Universal Commerce Protocol (UCP) checkout applies to participating merchants and eligible listings. Check the requirements.
OptAEO's readiness checks depend on your platform and configuration. Use the findings to identify work to do, rather than treating a passed check as provider approval.
Check your product's visibility
Choose a small set of relevant shopping questions and keep a record you can compare over time.
- Record the test. Save the question, service, date, market and important context such as budget or product requirements.
- Inspect the answer. Record whether your product appears, its position and source links where provided, and any incorrect facts.
- Fix the evidence. Review your product data and the pages behind the answer. Correct missing or conflicting information at its source.
- Repeat and compare. Use a consistent set of questions. Keep individual results in context; answers can vary between runs and accounts.
OptAEO's simulation coverage includes ChatGPT, Claude, Meta Muse, Grok Bot, Gemini and Perplexity. Simulation and citation tools are available on Business and Enterprise.
These tests use provider integrations. They do not reproduce every consumer app feature or shopping session. Meta Muse and Grok Bot simulations currently answer without a fresh web search, so they may not reflect your latest product changes. Google AI Overviews and Microsoft Copilot are described above as shopping experiences, not as supported OptAEO simulation engines.
Citations in OptAEO come from the simulation queries it runs. They show evidence from those tests. They do not measure citations from real shopper conversations.
Review your readiness score, tested mentions and source citations separately from referred visits and orders. A higher score alone is not proof of increased sales.
The AEO guide explains how Agentic Engine Optimisation (AEO), scoring and measurement fit together.
Getting started
Connect your store or import your catalog. Run a scan, review the issues and inspect the suggested fixes.
The free scan helps you see product-data gaps and fix previews. Applying fixes and using simulation or citation tools depend on your plan.
Start with the product facts you can verify. Then use testing to decide what needs further attention. Get a free scan ↗