Skip to main content
Tracked products connect your catalog to what AI models actually show shoppers. You upload each product with the ID you use for it in your own systems (a Shopify product GID, an ASIN, a SKU), and Promptwatch matches it against the products extracted from AI shopping responses by exact ID. Matched products tell you your catalog is surfacing; unmatched ones tell you where to aim your shopping prompts. You manage tracked products under Shopping → My products.

​
Add products with a CSV

​
Prepare the file

The CSV needs an id column and a name column; a description column is optional. Column headers are case-insensitive, and there’s a Download template link in the upload dialog if you want a starting point. The limits, checked before anything is imported:
  • 5,000 products per CSV
  • 5 MB file size
  • id up to 255 characters, name up to 500, description up to 2,000

​
Upload it

  1. Click Upload CSV.
  2. Drop your file in. The dialog parses it immediately and lists every problem at once, with row numbers, so you can fix the file in one pass instead of resubmitting per error. Duplicate IDs inside the file are flagged too.
  3. Click Import.
The import is forgiving where it should be: products you already track are skipped and counted, not raised as errors, so re-uploading your full catalog after adding new items is safe. The result toast tells you how many were added, how many were already tracked, and how many matched an extracted product right away.

​
Add a single product

Click Add product and fill in:
  • Product ID: the identifier from your catalog, for example gid://shopify/Product/1234567890. This is what gets matched against the product IDs AI platforms use, so use the same ID scheme your feed exposes to them.
  • Name: the display name.
  • Description (optional): a short description. This also feeds prompt generation, so a line about what the product is helps there.
Adding an ID you previously removed reactivates the old entry instead of creating a duplicate. The Promptwatch Add product dialog with the Product ID, Name, and optional Description fields for tracking a catalog item.

​
How matching works

A tracked product matches when its ID exactly equals the ID of a product extracted from an AI shopping response. Matching runs at upload time and again in the background afterwards, so a product you upload today still matches automatically when it first appears in a shopping response next week. The Match status filter splits the table into matched and unmatched, which is the practical to-do list: unmatched products either haven’t appeared in AI shopping results yet or appear under a different ID than the one you uploaded.

​
Edit and remove

  • Edit: name and description are editable. The Product ID is not, because changing it would break the match; remove the product and add it with the new ID instead.
  • Remove: stops tracking the product across AI shopping results. Nothing else is deleted, and you can add it again later.
Viewers can see the list but can’t add, edit, or remove products.

​
After you upload

Select unmatched products and click Generate suggested shopping prompts to create prompts designed to surface them in shopping responses; Setting up shopping prompts walks through that flow. Once prompts run and a product appears in a shopping response, its row flips to matched, and its appearances, positions, and merchants show up across the Shopping overview.