How to sell on ChatGPT without Shopify or Etsy (2026 guide)

You do not need to move your business to Shopify or Etsy just to be discoverable through ChatGPT. Choose a route that fits what you sell, the website you already maintain and the current merchant requirements.

Reviewed October 1, 2026 · 7 min read

An independent retailer can pursue ChatGPT discovery without adopting Shopify or Etsy. A service business can also appear in conversational recommendations, although that is different from qualifying for a product shopping integration. Start by identifying which outcome you need.

This guide compares existing platform connections, direct integrations and hosted publishing tools. It also covers the public-page work a consultant, clinic or local service business can do without building a retail catalog.

How ChatGPT sources its shopping answers#

OpenAI's merchant page describes product data participation. Conversational search can also use public web information. These routes have different requirements:

• Public web discovery: maintain accessible pages that clearly describe the business and its offers. Search crawling and user-triggered requests have different behavior, so check the relevant controls.
• Product data integration: follow the current merchant application and feed specification. Publishing a file is not the same as having it accepted or displayed in shopping results.

OpenAI distinguishes organic answers from advertising. Neither paying for a publishing tool nor submitting a feed guarantees an organic recommendation. Judge a provider by the information and integrations it can actually deliver.

Merchant-owned checkout makes your existing purchase experience important. It does not mean all integration requirements have disappeared. Confirm the route currently available to your business before investing in a feed or checkout API.

Route 1: ride a platform integration#

Check whether your current commerce provider offers the integration you need. Ask which products and markets are supported, whether enrollment is required and how errors are reported. Do not assume a platform announcement means your own catalog is already live.

If you run a service business, a retail storefront may add little value. A clear offer page with scope, price basis, location and a booking or enquiry route may be the more appropriate foundation.

Route 2: build a direct protocol integration#

A direct integration can make sense when your team maintains the catalog and backend. First confirm platform participation and the exact contract. The work may include:

1. Mapping your catalog to the destination's feed schema and validating ingestion results.
2. Keeping prices, variants, availability and policy URLs current.
3. Implementing checkout capabilities only where the confirmed integration requires them.
4. Handling authentication, payment boundaries, order events and duplicate retries.
5. Testing success and failure paths in the supported test environment before accepting live orders.

Estimate the work against your existing systems, not a universal 'weekend' timeline. Ongoing monitoring and version changes need an owner. The ACP guide and protocol comparison explain the responsibilities behind the labels.

Route 3: use a hosted agent-commerce layer#

A hosted service can publish structured offers and provide supported actions while you keep your existing website. Evaluate its output, supported destinations and maintenance responsibilities. A public resource is useful only where a client can discover and use it.

Nexez publishes business listings and supported discovery resources from offer data, with checkout or scheduling for supported workflows. It can help organize the information, but it does not bypass another platform's merchant eligibility or guarantee a ChatGPT listing.

Compare the subscription, transaction costs, data portability and operational limits. Ask how updates propagate and what happens when a connected service fails. Review the current pricing and features against a concrete customer workflow.

The three routes compared#

The right route depends on your current systems and the destination's requirements. These are planning considerations, not guaranteed launch times:

Selling services when there is no "product" to feed#

Services need different information from physical products. A buyer may need to know the scope, location, duration, qualifications or available appointment times. Some services require a quote before there is a fixed price.

Describe the next useful action honestly. '90-minute massage, $140' can lead to a booking. 'Commercial HVAC inspection, quote after a site review' can lead to an enquiry. Do not invent a fixed price to appear in a comparison. See how agents book services.

For local services, state the actual service area and distinguish remote from in-person delivery. Keep visible details and appropriate LocalBusiness or Service markup aligned. The local AI search guide covers profiles and location information.

A practical starting checklist#

These improvements are useful whether or not you pursue a merchant integration immediately:

1. Review OpenAI's search crawler access separately from your training-crawler preference.
2. Give each distinct offer a useful page with scope, eligibility, location and a clear next step.
3. Explain the price or quote basis, including what can change the total.
4. Add appropriate structured data that matches the visible page, using the JSON-LD guide.
5. Test the enquiry, booking or purchase flow from a fresh session.
6. If you need a product feed integration, confirm eligibility before building it.
7. Inspect the published output and record a baseline for referrals and completed outcomes.

How you will know it is working#

Use three complementary signals, keeping their limits visible:

• Server logs show requests, but a user-agent string alone does not verify the caller or prove a recommendation happened.
• Identifiable referrals and campaign parameters show some visits. Missing attribution means the segment is not a complete count.
• Repeated customer questions show how your business is described. Record sources and errors rather than treating one answer as a stable rank.

The measurement guide explains how to connect these signals with enquiries and orders. Expand the channel based on demonstrated customer value rather than a promise that publishing a feed will produce immediate traffic.