Morrisons Grocery Price Scraper

Extract Morrisons product prices, availability, brands and ratings as clean structured JSON. Point it at a category, a list of products, or the whole catalogue. No account, no browser, no personal data.

Press Start and it returns 100 real products immediately — every input field has a working default.

What you get

One flat row per product. No nested objects, no arrays to unpick.

{
  "sku": "108413093",
  "name": "Morrisons Whole Cucumber",
  "brand": "Morrisons",
  "description": "Morrisons Whole Cucumber",
  "price": 0.99,
  "currency": "GBP",
  "pricePerUnit": null,
  "availability": "InStock",
  "itemCondition": "NewCondition",
  "ratingValue": 3.6,
  "ratingCount": 28,
  "imageUrl": "https://groceries.morrisons.com/images-v3/4b85987b-.../500x500.jpg",
  "productUrl": "https://groceries.morrisons.com/products/morrisons-whole-cucumber/108413093",
  "categoryPath": "Groceries > Fruit & Veg",
  "scrapedAt": "2026-08-11T09:14:22.187Z"
}

Field	Type	Notes
sku	string	Morrisons product id, stable across runs — use it as your join key
name	string	Product title
brand	string	Brand name
description	string	Short description
price	number	Current price as a number, not a string
currency	string	ISO code, GBP
pricePerUnit	number | null	Unit price where published
availability	string	InStock, OutOfStock, …
itemCondition	string	NewCondition
ratingValue	number | null	Mean customer rating
ratingCount	number | null	Number of ratings
imageUrl	string	500×500 product image
productUrl	string	Canonical product page
categoryPath	string | null	Breadcrumb trail, when crawled from a category
scrapedAt	string	ISO 8601 timestamp of extraction

Three ways to run it

Mode	Use it for	Cost profile
Category pages (default)	Assortment and pricing for a section	50 products per category page
Whole catalogue via sitemap	A full price file. Walks Morrisons' own published sitemap — 30,000+ products	One row per product, capped by maxItems
Specific product URLs	Daily price tracking on a fixed basket	Cheapest per run

maxItems is a hard cap on billable results, so it is also your spend cap.

Why not just build this yourself

You can. Here is the honest arithmetic before you do.

A scraper against a live retail site is 20–40 hours to build and a few hours a month to keep alive, because the source changes and yours breaks quietly. At a £60–100/hour loaded rate that is £1,200–4,000 to build, plus maintenance, plus your own error monitoring so you find out before your data does.

This actor costs $0.003 per product. A full 30,000-product catalogue sweep is about $90. A daily 500-product basket check is about $1.50 a day.

Below roughly 100,000 rows a month, the buy decision is not close. Above it, build — and I would tell you the same thing.

Compliance, stated plainly

This actor never returns personal data. Not "strips it" — never collects it. The output schema is a fixed allowlist of 15 fields, and rows are constructed from that list rather than filtered against a blocklist. If Morrisons begins publishing review authors tomorrow, those fields cannot appear in your dataset without a deliberate code change here. There is a test that proves it, and a runtime guard that refuses to push a row carrying a forbidden field.

Aggregate ratings are included, because a mean and a count are aggregates over people, not data about an identifiable person.

robots.txt is respected in code, not just in spirit. groceries.morrisons.com disallows /sso-login, /previewer/*, /api/ and /events/ for all user agents. This actor refuses those paths even if you pass them in as input. Everything it does crawl — category pages, product pages, and the sitemap — is explicitly permitted, and the sitemap is published by Morrisons themselves.

No login, ever. All data is read from pages served to an anonymous visitor.

Publishing notes (for the actor owner, not the buyer)

Pay-per-event pricing is configured in the Apify console, not in this repo. Create one charge event named exactly product — that is the event name src/main.js charges, and a mismatch means you ship for free.

Suggested price $0.003 per result. Do not undercut to $0.001: raising a price later requires 14 days' notice to existing users, so the floor you set is the floor you live with for a fortnight.

Running locally you will see Ignored attempt to charge for an event — the Actor does not use the pay-per-event pricing. That is expected and harmless; charging only exists on the platform.

Notes

Prices are read from the schema.org Product JSON-LD that Morrisons publishes for search engines, not from CSS selectors. That makes this actor unusually stable: structured data is a contract the site maintains deliberately, so it changes far less often than markup does. It is also why no browser is needed, which is why it is cheap.

Grocery prices are store- and time-sensitive. Treat every row as a reading at scrapedAt, not a permanent fact.

Pages that do not yield a usable product are skipped without charge. You pay for results, not attempts.