BuyAction or OrderAction Markup So AI Agents Can Place Orders Directly
Tells autonomous shopping agents exactly which URL to call when a customer asks them to buy your product.
What this signal tests
We check whether your product pages include a Schema.org BuyAction or OrderAction descriptor pointing to a real purchase endpoint. This is machine-readable code, hidden from human visitors, that tells an AI shopping assistant the verb (buy or order), the destination URL, and the HTTP method to use when placing an order on a customer's behalf.
Why it matters for your visibility in AI
AI shopping agents are moving from suggesting products to actually completing purchases. When ChatGPT, Perplexity, or a future agent platform helps a customer buy something, it needs more than a description and price - it needs an unambiguous machine instruction for how to commit the transaction. BuyAction is the closest established Schema.org standard to that instruction. Without it, your products may be quoted by an AI but the agent will hand the customer back to a human-only checkout flow, and the sale often dies in that handoff. Sites that publish BuyAction today are positioning themselves to be the default choice when agentic commerce reaches mainstream volume, which most analysts now place inside the next twelve to twenty-four months.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| >=1 purchasable item has BuyAction/OrderAction potentialAction. |
How we test it
Our scanner fetches each product page, extracts the embedded JSON-LD structured data, and looks for a potentialAction property of type BuyAction or OrderAction. It then verifies the target field points to a real EntryPoint with a urlTemplate (the destination URL pattern) and an httpMethod (such as POST). At least one purchasable item must declare this for the page to pass.
Show technical detection method
Parse potentialAction on Offer/Product; require BuyAction/OrderAction with target.EntryPoint.urlTemplate and httpMethod.
If your site fails: how to fix it
- Identify which pages on your site represent something purchasable - products, plans, packages - and list the actual URL each one would be bought through if an agent placed an order.
- Ask your developer to extend the existing Schema.org Product or Offer JSON-LD on each of those pages with a potentialAction block of type BuyAction, including target as an EntryPoint object.
- Inside the EntryPoint, populate urlTemplate with the real checkout URL, httpMethod (usually POST) and contentType (usually application/json) so an agent has unambiguous instructions.
- Validate the resulting JSON-LD with Google's Rich Results Test and the Schema.org validator before deploying, and re-run the AI Ready Test scan to confirm a pass.
- For digital subscriptions, use OrderAction with priceSpecification so agents understand the recurring nature of the purchase, not just the one-off price.
Quick facts
| Maturity | EMERGING |
|---|---|
| Weight | medium |
| Category | Agent Actions |
Primary sources
Related signals
Frequently asked questions
Is this only useful when autonomous AI agents become mainstream?
No. Even today, when ChatGPT, Perplexity, and Gemini surface your product, having BuyAction makes your listing visibly machine-actionable, which improves the chance you are quoted in the first place. The deeper benefit lands once agent-driven checkout becomes common, but the indexing benefit starts immediately.
Does adding BuyAction commit me to building a custom agent checkout?
No. BuyAction can simply point to your existing checkout URL. You are documenting the destination, not building a new transactional endpoint. Building a true machine-to-machine purchase API is a more advanced step you can take later if agent traffic justifies it.
Will this matter in 2026 or is it years away?
It matters now. OpenAI, Anthropic, Google, and Stripe have all shipped agent-purchasing tooling during 2024 and 2025, and major retailers have begun publishing BuyAction. Early adopters are the ones being indexed first as agent traffic ramps.
Does this cost anything to implement?
The work is a one-time markup change, typically a few hours for a developer who knows Schema.org. There are no licensing fees and no per-transaction costs from Schema.org itself. The cost is purely the engineering time to add and validate the JSON-LD on each product template.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.