Max MorrishBack to projects

FeedPhantom

A browser-only scan that finds deleted Shopify products still listed in Google Merchant Center and writes the fix files.

Prototype. Screens and video come from a local copy running on sample data.

3 of 5 sample Merchant Center rows flagged as stale, each with a reason and an action

26 seconds on the local copy: loading the sample exports, scanning, adding one stale Merchant Center row by hand and scanning again (3 flagged, then 4).
Where
Personal project
Role
Product and build
Built with
HTML, CSS, JavaScript, Netlify Functions, Node.js scripts

What I made

I built FeedPhantom as a static site. Its scan compares a Shopify products export with Merchant Center diagnostics in the browser, flags each stale record with a reason and an action, and writes a report and two supplemental feeds that expire or exclude those items.

Later I added an AI-shopping readiness page and 12 help pages generated from one data file.

  1. ExportsShopify products CSV and Merchant Center diagnostics
  2. Browser scanmatches handles, product URLs and issue text
  3. Flagsseverity, reason and next action per stale record
  4. Fix filesreport, expiration feed and ads exclusion feed
The ghost feed scan. It runs in the browser on the merchant's own exports; no store login.

17checks passing in the project's verify script

20pages in the sitemap

12help pages generated from one data file

What came out of it

The scan on sample data

The sample is invented products on an example.com store. The scan read 5 Merchant Center rows, flagged 3 and gave a risk score of 9.

Ghost feed scan results on the built-in sample exports: 3 ghost candidates, risk score 9, 5 merchant rows read, and a table flagging 'Draft test candle' and 'Old blue vase' as high severity with reasons and actions, above three download buttons.
The scan on the built-in sample data: 3 of 5 Merchant Center rows flagged, each with a reason and an action.

The site

FeedPhantom landing page on desktop: the headline 'Find the product data AI shopping will misread.' beside the empty Ghost feed scan panel with Shopify and Merchant Center export inputs.
The landing page, with the ghost feed scan next to the headline.
AI-shopping readiness page: the headline 'Check if AI shopping can actually understand your products.' and an AI commerce scan panel with a store URL field.
The AI-shopping readiness page. Its scan needs the server function, so it is shown before a scan.

Help pages

Generated help page titled 'Check whether your Shopify feed is ready for ChatGPT shopping.' with a 'When this matters' box and a five-step fast path.
One of the 12 help pages generated from a single data file.

Where it's going

The full version is a product-data checker for Shopify merchants: a public scan shows whether AI shopping and Google can read a store's products, and the browser-only ghost scan finds stale feed records and writes the fix files.

  1. NowSample-data buildGhost scan, fix files, a readiness page and 12 help pages, on sample data.
  2. Readiness scan verifiedThe public scan reads a live store's homepage, product sitemap and product schema.
  3. Real store auditsMerchants scan their own stores and leave with a report of findings.
  4. Product-data checkerOne place to check and fix a store's product data for AI shopping.