Methodology and limitations
How FirstShelf evaluates AI-shopping readiness
FirstShelf runs a controlled readiness check on the listing information you submit. It identifies evidence relevant to example buyer questions and suggests truthful fields to review. It does not query live personalized ChatGPT results or control whether any external shopping system shows a product.
What we observe
- Submitted title, description, selected platform, optional URL, and optional images.
- Concrete product-selection constraints, such as product type, style, compatibility, format, materials, dimensions, intended use, variants, and delivery details.
- Whether submitted text or images support, leave ambiguous, omit, or do not supply evidence for a constraint.
- The relevant Etsy or Shopify listing field or area to review when FirstShelf has a versioned field definition.
An observed result is grounded in submitted text, a submitted image, or a known platform field definition. Missing data is labeled as missing or not supplied; FirstShelf does not invent product claims.
What we infer
A buyer-match state, priority, or suggested correction can include a model-generated interpretation. Those inferences may be wrong. Sellers review the evidence, classify a proposed fact as true, not true, or not applicable, and remain responsible for what they publish.
Controlled buyer-prompt readiness check
Buyer prompts are controlled examples generated from the submitted product category and can be reviewed by the seller. FirstShelf decomposes a prompt into constraints, maps the submitted listing evidence to those constraints, and offers specific, truthful corrective actions. A seller may reject an incorrect mapping or mark a constraint not applicable. This is not a live ChatGPT shopping query, a monitoring result, or a recommendation guarantee.
FirstShelf Score
Score version firstshelf-score-2026-07-22, effective 2026-07-22. Initial published descriptor for the Cloudflare audit workflow.
Product detail coverage
35%
Listing structure
25%
Product evidence
25%
| Dimension | Weight | What it means |
|---|---|---|
| Product detail coverage | 35% | Specific, relevant product facts in the submitted title and description. |
| Listing structure | 25% | How clearly the submitted listing organizes details for a reader. |
| Product evidence | 25% | Whether claims are specific, attributable, and supported by submitted information. |
| Platform fit | 15% | Whether submitted information fits the selected platform's listing conventions. |
When a visual audit is available, the combined score uses 40% content foundation, 30% visible image evidence, and 30% product and platform evidence. A score summarizes the submitted information; it is not a forecast of external visibility, traffic, or sales.
Platform fields and input methods
Field map version 2026-07-22. FirstShelf supports Etsy and Shopify listing URLs and manual entry in the authenticated Audit workflow. It does not connect marketplace accounts, publish changes, submit feeds, or claim partnership or endorsement.
Etsy
- Listing title: Etsy listing editor → Title
- Listing description: Etsy listing editor → Description
- Attributes: Etsy listing editor → Details → Attributes
- Listing photos: Etsy listing editor → Photos
Shopify
- Product title: Shopify admin → Products → Product title
- Product description: Shopify admin → Products → Description
- Product category and attributes: Shopify admin → Products → Category and product attributes
- Product media: Shopify admin → Products → Media
- Variants and options: Shopify admin → Products → Variants
External factors FirstShelf does not control
Product eligibility, catalog or feed access, availability, price, reviews, merchant quality, policies, region, personalization, competitor products, and freshness can affect external shopping results. ChatGPT selects product results independently and considers query context plus product and merchant information. OpenAI’s current merchant documentation says ChatGPT shopping is live for users in the United States and may expand; availability should not be treated as universal. Accessed 23 July 2026.
OpenAI: Shopping with ChatGPT Search · OpenAI: Power product discovery in ChatGPT
What FirstShelf does not promise
FirstShelf does not guarantee recommendation, citation, ranking, traffic, sales, or inclusion in any external AI-shopping experience. A copied suggestion is not evidence that a marketplace listing changed. Sellers remain responsible for reviewing factual accuracy, platform policies, and published listing changes.