Know where to start
Actions are sorted by the points they gain, so your time goes to the products and fields that matter most.
Feed optimization
Make your products easier for ChatGPT to recommend
ChatGPT recommends products it understands. Convrail scores every product from 0 to 100 on how complete its listing is, lists the fixes that gain the most, and drafts titles and descriptions written for an AI assistant. Rewrites go to the feed sent to ChatGPT only: your store pages never change.
- A 0 to 100 score per product
- AI rewrites for the feed, never your store
- The same rules as the feed validator
Nordic Oak LED lamp
Completeness score 58 92
Suggested title
NORDIC OAK LAMP !!! SALE Nordic Oak LED desk lamp, walnut base, dimmable
- +18 add the GTIN
- +10 complete the description
- +6 add an image
Applied to the feed sent to ChatGPT, never to your store
What changes for your store
Listings ChatGPT can reason on
Identifiers, a category, a real description and clear policies give the assistant something concrete to say about your products.
Faster rewrites, your store untouched
Drafted titles and descriptions follow the feed rules and go to the feed sent to ChatGPT. Your product pages stay exactly as they are.
Everything optimization needs, included
-
Completeness score
Every product scored from 0 to 100: 50 points for a valid row, the rest for the recommended fields.
-
Prioritized actions
The fixes that gain the most points come first, product by product.
-
AI-drafted titles
150 characters at most, never in full capitals, stating plainly what the product is.
-
AI-drafted descriptions
Plain text up to the 5,000-character limit of the feed: what it is, every stated feature, who it is for and how it is used.
-
Feed rules built in
Drafts follow the feed specification by construction and are validated again before you see them.
-
Nothing invented
A draft only states facts from your listing, its variants and its main image. It never adds what the product does not have.
-
Your store stays as it is
Rewrites live in the feed sent to ChatGPT. Your product pages are never edited, and a catalog resync does not erase the rewrites.
-
Context hints for ChatGPT Ads
For a category and your audience notes, 2 to 4 ad group context hints describing the buyer, the need and the situation.
-
See what is sent
For each variant, the exact fields OpenAI receives and what each one still needs.
Live in three steps
-
Install Convrail
Your products are scored as soon as the catalog syncs.
-
Start with the top actions
Open a product, see what it lacks and what each fix is worth.
-
Rewrite for the feed
Ask for a new title and description: they go to the feed sent to ChatGPT, and your store pages stay as they are.
Under the hood
For technical teams and agencies: how the score is computed, how drafts are produced and checked, and what the context hints cover.
The completeness score, explained
The score is deterministic and the same for every store. A valid row earns 50 points; each recommended field adds a fixed bonus. Actions are sorted by the points they would gain, so the list starts with what matters most.
| Criterion | Points | Why it matters |
|---|---|---|
| Row passes validation | 50 | An invalid row is never delivered, so nothing else counts until it is fixed |
| GTIN or MPN present | 10 | Identified products are matched to known catalogs far more reliably |
| Product category | 8 | Tells the model where the product belongs in a taxonomy |
| Description of 200 characters or more | 6 | The model reasons on meaning; a one-line description gives it nothing to reason on |
| Sale price when on promotion | 6 | Lets the assistant quote the price the shopper will actually pay |
| Additional images | 5 | More views, better product cards |
| Rating and review count | 5 | A strong trust signal when the assistant compares options |
| Variants described (variant_dict) | 5 | Sizes, colors and materials become answerable questions |
| Return window in days | 5 | A concrete policy the assistant can state |
AI-drafted titles and descriptions, for the feed only
The rewriting assistant produces a title of 150 characters at most, never in full capitals, and a plain-text description within the 5,000-character limit of the feed specification. It states what the product is, then every feature, specification, size and compatibility your listing gives, who it is for and how it is used. It reads the title, description, brand, category, variant options and main image, and writes in the language of your listing. Each draft comes with a short rationale.
- Drafts respect the feed specification by construction: the same rules the validator enforces are part of the prompt, and the output is validated again before it is shown.
- Rewrites are written to the feed sent to ChatGPT, never to your store. On WooCommerce you read the suggestion before applying it; on Shopify one click drafts and applies it, and the result is shown right away.
- The model provider is set by the operator. If it is unavailable, nothing is changed and the app tells you so.
- The draft only uses facts from your listing. From the image it keeps what is shown beyond doubt (shape, pattern, text printed on the packaging), never a material or a prop from the photo, and a product sold in several colors is not described in the color of its photo.
Targeting hints for ChatGPT Ads
ChatGPT Ads matches ads on meaning, not keywords. Give a category and a few notes on who buys and why, and Convrail suggests 2 to 4 context hints for your ad groups, each one describing the buyer, the need, the situation, the moment and the constraints. The hints apply to your whole store, not to a single product.
Frequently asked questions
Does Convrail change my product pages?
No. Rewrites apply to the feed sent to ChatGPT only. Your product pages stay as they are, and a catalog resync does not erase the rewrites.
Which AI model writes the drafts?
The provider is configurable by the operator. Prompts encode the feed rules (titles of 150 characters at most, never in full capitals, plain-text descriptions) and the output is validated before display, whatever the model.
Is my catalog used to train a model?
Convrail sends only what a draft needs (title, description, brand, category, variant options and main product image, never customer data) to the configured provider, under that provider’s API terms. Your catalog is not used by Convrail to train anything.
Why does a valid product still score 50?
Because validity is the floor, not the ceiling. The other 50 points reward the recommended fields that make a listing easier to match and describe: identifiers, category, a real description, ratings, variants and policies.
Related guides
Measure the ChatGPT channel this week
Convrail starts with a Free plan, and paid plans begin with a free trial. Leave your email and we send the install link when your spot opens.