Case Study — Retail & Price Data / Live Product

Collecting prices from three supermarkets automatically, then doing the maths a shopper won't

Cheepaz is a live product: an automated pipeline tracks pricing across Coles, Woolworths and IGA, then works out the cheapest way to buy an entire shopping list.

The problem

Grocery prices move constantly and differ store to store, but nobody actually checks three supermarkets for every item on their list — so most shoppers just pick one store out of habit and overpay on whichever half of the list happens to be cheaper elsewhere. Solving that properly means someone (or something) has to track all three, continuously.

What was built

Cheepaz runs an automated collection pipeline that pulls current pricing across Coles, Woolworths and IGA on a running schedule, normalized into one catalog spanning 16 categories and hundreds of brands. On top of that sits a cost-optimization layer: given a shopping list, work out the cheapest total — from a single store, or split intelligently across all three — and surface it as a straightforward list builder, not a spreadsheet.

3 supermarketsources Coles · Woolworths · IGA Scheduled pricecollector runs without manual triggering Normalized productcatalog 16 categories, matched across stores Cost-optimizedlist cheapest single or split total

Why this is the hard part

Collecting three price feeds is the easy 20%. The actual problem is matching — the same product, named, packaged and categorized slightly differently by three separate retailers, has to be recognized as one thing before any cost comparison means anything. Get that matching wrong and the whole comparison becomes noise. It's the same discipline as any Systems Integration engagement: making disparate systems agree on what a single "thing" actually is before you can automate anything useful on top of it.

Where this pattern generalizes: the same matching-before-automating problem shows up anywhere a business pulls pricing, stock or supplier data from more than one source — see also multi-distributor parts catalog aggregation.

Got something like this running by hand?

The hard part here wasn't collecting prices — it was getting three different systems to agree on what counts as "the same product." That's the exact problem behind most messy internal data too.

Assess My Workflow
From $495, credited to any build