How big is agentic shopping in 2026?

Still small in absolute revenue (low single-digit billions globally) but growing fast. ChatGPT Shopping launched in 2024-2025 and expanded through 2026; Perplexity Shopping, Operator, and Apple Intelligence purchase flows added meaningful volume in Q1 2026. The trajectory matters more than current absolute size for brands planning their channel strategy.

Do I need to apply to each platform separately?

Yes for now. ChatGPT Shopping, Perplexity Shopping, Google Merchant Center, and Apple Business Connect each have their own merchant onboarding. The underlying product data (Schema.org, catalogue feed, GTINs) is shared, but the application and policy compliance per platform is separate.

What is the single highest-leverage step?

Clean Schema.org Product JSON-LD on every PDP plus a daily-refreshed Google Merchant Center feed. Without that foundation, MCP servers, agent applications, and review optimisation all have less effect. Most brands underinvest here.

Will an MCP server replace my product feed?

Not yet. The MCP server complements the feed; it provides live pricing and availability that batch feeds can't match, but agentic shopping platforms still ingest bulk catalogues via traditional feeds. Run both for the next 12-18 months as the patterns mature.

How to Win in Agentic Shopping (2026)

Why Agentic Shopping Matters in 2026

ChatGPT Shopping, Perplexity Shopping, OpenAI Operator, Apple Intelligence purchase flows, and emerging x402-based agent marketplaces are converting AI assistants into shopping channels. The buyer in 2026 is increasingly software, not a human clicking through a storefront. Brands that don't optimise for agentic shopping miss a fast-growing channel.

Step 1: Get Your Product Data Agent-Readable

Implement Schema.org Product JSON-LD on every PDP, publish a daily-refreshed catalogue feed, and connect to Google Merchant Center. See how to create agent-readable product feeds for the full step-by-step. This is the foundation; without clean product data, nothing else works.

Step 2: Publish an MCP Server with Product Tools

Expose search_catalogue, get_product, get_pricing, check_availability, and get_return_policy via MCP. Agentic shopping flows in OpenClaw, Cursor, Claude Desktop, and ChatGPT Shopping increasingly call MCP tools rather than crawling your PDPs. See the MCP server starter template.

Step 3: Apply to the Major Agentic Shopping Programmes

ChatGPT Shopping: Submit through the OpenAI partner programme; provide GTINs, pricing, returns, fulfilment data.
Perplexity Shopping: Apply through Perplexity's merchant onboarding; same data requirements.
Google Merchant Center + AI Overviews: Maintain feed health and resolve all warnings.
Apple Business Connect / Intelligence: Ensure brand and product entity match Apple's catalogue requirements.
Visa Trusted Agent / x402 marketplaces: Register if you operate at the right scale.

Step 4: Optimise Brand Description for AI Extraction

Agentic shopping assistants extract product descriptions, brand voice, and category positioning from your pages. Optimise:

Front-load the product's strongest differentiator in the first sentence of every PDP.
Use specific, claim-grade language (40-hour battery, 99.99% SLA) rather than generic descriptors (long battery, reliable).
Include comparison context for the agent ("vs Bose QC45" type framing helps agents categorise).
Add high-quality alt text on product images for vision-based agents.

Step 5: Make Returns, Warranty, and Fulfilment Discoverable

Agentic shoppers increasingly include returns policy and warranty in their answer. If your terms aren't published in machine-readable form, the agent defaults to industry norms (often worse than yours). Add:

Schema.org hasMerchantReturnPolicy on every Product.
Schema.org warranty property where applicable.
Plain-prose returns / warranty page linked from PDP.
MCP tool for returns policy lookup.

Step 6: Track How You Appear in Agent Sessions

Run a recurring agent-session test: instruct ChatGPT Shopping (or Operator, or Perplexity Shopping) to "find me a [category] with [criteria]". Note where your brand appears, how it's described, what the comparison set is, and whether the agent recommends you for the criteria you meet. Track this weekly.

Step 7: Win on Reviews and Ratings

Aggregate review ratings carry weight in agentic shopping rankings. Maintain accurate aggregateRating data, monitor for review fraud signals (which sometimes get your products downranked), and respond to bad reviews on Trustpilot, Google Reviews, App Store, Amazon, and category-specific sites.

Step 8: Prepare for Agent Payment Flows

Accept agent-authenticated payment flows where supported (x402, AP2 mandates, Visa Trusted Agent).
Provide clear pricing in one currency where the agent operates.
Make checkout flows agent-friendly (low friction, well-structured forms, no surprise fees).
Document any agent-specific rate limits or order-size caps in your developer / merchant docs.

What Not to Do

Don't rely on retail website crawling. Agentic shoppers prefer structured feeds and MCP tools; they sometimes skip pages entirely.
Don't expose stale or wrong pricing. A single bad transaction in an agentic flow damages brand trust fast.
Don't optimise only for one programme. The agentic shopping landscape is fragmenting; multi-channel presence matters.
Don't ignore the returns / warranty story. A "no returns" default applied by an agent because your policy was undiscoverable costs more than the data work to publish it.

How to Win in Agentic Shopping