Shopping is moving inside the assistant. Buyers now ask ChatGPT, Perplexity and Microsoft Copilot to find, compare and even purchase products for them — and ChatGPT's Instant Checkout and the Agentic Commerce Protocol let an AI agent complete the sale without the shopper ever landing on your site. An agentic storefront is a page an AI agent can read, understand and transact against. Rankio's Agentic Storefront Readiness audit tells you, per platform, whether your pages are ready — and gives you the exact list of problems and fixes.
AI agents are becoming the checkout — and most sites can't be read by them
A person browsing your product page can forgive a missing price, a JavaScript-only "Add to cart", or a return policy buried three clicks deep. An AI agent cannot. When ChatGPT, Perplexity or Copilot try to recommend or buy a product, they parse structured data, not pixels. If your price, availability, identifiers and checkout path aren't machine-readable, the agent skips you and recommends a competitor whose data it can read.
This is a new, invisible failure mode. Your page looks perfect to humans and ranks fine in Google, yet it is unusable by the agents now mediating a fast-growing share of purchases. Agentic storefront readiness is the commerce layer of AI visibility — and it is measured in structured signals, not content.
- ChatGPT Instant Checkout (Agentic Commerce Protocol) lets agents buy directly from merchants
- Perplexity Shopping and "Buy with Pro" surface and rank products from structured data
- Microsoft Copilot shopping draws on the Bing shopping graph and merchant feeds
The signals that decide agentic readiness
Rankio extracts ground-truth signals directly from your HTML — it never guesses — then scores each against what agentic surfaces require:
| Signal | Why an agent needs it | Platforms most affected |
|---|---|---|
| Product structured data | schema.org Product JSON-LD with name, description, image and brand so the agent can identify the item | ChatGPT, Perplexity, Copilot |
| Offer data | Offer with price, priceCurrency and availability — the agent won't recommend what it can't price | ChatGPT, Perplexity, Copilot |
| Product identifiers | SKU and GTIN/MPN let the agent match your item across catalogues and feeds | Copilot, ChatGPT |
| Ratings & reviews | aggregateRating gives agents the signal they use to rank you against alternatives | Perplexity, Copilot |
| Return & shipping policy | MerchantReturnPolicy and OfferShippingDetails are required by ACP-style checkout | ChatGPT |
| Buy / checkout path | A machine-followable add-to-cart or checkout route and a canonical product URL | ChatGPT, Perplexity, Copilot |
| Agent crawler access | Allowing GPTBot, OAI-SearchBot, PerplexityBot and BingBot so agents can reach the page at all | ChatGPT, Perplexity, Copilot |
Scored separately for ChatGPT, Perplexity and Copilot
Readiness isn't one number. Each surface has different requirements, so the audit returns a distinct verdict — ready, partial or not ready — for every platform, plus the biggest blocker holding each one back.
| Platform | Agentic surface | What it depends on most |
|---|---|---|
| ChatGPT | Instant Checkout · Agentic Commerce Protocol · OpenAI product feed | Machine-readable Product + Offer, return/shipping policy, a checkout path, GPTBot access |
| Perplexity | Shopping · "Buy with Pro" · merchant program | Structured product data, ratings, clear price, PerplexityBot access |
| Copilot | Copilot shopping · Bing shopping graph · Microsoft Merchant | schema.org Product markup, pricing feeds, BingBot access |
An overall score, the problems, and the fixes
Every audit produces a decision-ready report, not a wall of raw data:
- Overall readiness score (0–100) summarising how agent-ready the page is across all three platforms
- Per-platform verdicts for ChatGPT, Perplexity and Copilot with a score and the top blocker for each
- Element checks — every signal marked present or missing, scored, with a concrete suggestion and the platforms it impacts
- An ordered list of problems, most impactful first, so you fix what unblocks the most agents
- Ordered recommendations you can hand straight to a developer
- Results by provider — compare how OpenAI, Google Gemini, Anthropic Claude and Perplexity each assess the same page
Because the audit runs across multiple models in parallel, you see where providers agree — a strong signal — and where they diverge, which usually points to an ambiguous or incomplete page.
SEO ranks you, GEO cites you, agentic readiness sells for you
| Discipline | Goal | Measured in |
|---|---|---|
| SEO | Rank your pages in search results | Positions, clicks, impressions |
| GEO | Get your brand cited in AI answers | Citations, AI Share of Voice |
| Agentic storefront readiness | Let AI agents read, compare and buy your product | Structured-data completeness, per-platform readiness |
They stack. Run the GEO Content Audit to become citable, then the Agentic Storefront Readiness audit to become transactable. Together they cover both halves of AI visibility — being recommended, and being buyable.
Frequently asked questions
See if AI agents can shop your site
Run the Agentic Storefront Readiness audit on your product page. Get a per-platform score for ChatGPT, Perplexity and Copilot, the ordered list of problems, and the fixes — free.