Coles vs Woolworths Product Matching: How to Build an Australian Grocery Price Comparison Dataset (2026)

Introduction

Coles vs Woolworths product matching is the step that decides whether an Australian grocery price comparison is trustworthy. Scraping two shelves is easy. Proving that a 1 kg tub on one site is the same item as a 1 kg tub on the other is the hard part.

Short answer: Coles vs Woolworths product matching links each Coles product to the same or nearest comparable Woolworths product, using brand, pack size, variant and barcode where available. Each pair gets a match type and confidence score, prices are normalised to unit price, and doubtful pairs are reviewed before analysis.

This guide covers why matching matters in 2026, the data fields you need, a step-by-step matching method, how to extend it to Aldi and IGA, and the compliance basics. It is written for pricing teams, FMCG brands and comparison apps rather than shoppers.

Why does Coles vs Woolworths price comparison need product matching?

Most people search "Coles vs Woolworths price comparison" as shoppers. Businesses ask a different question: across thousands of SKUs, where are the two chains priced differently, and how often do promotions change the answer? That question only works when every row compares like with like.

The stakes are high. In its final supermarkets inquiry report (released 21 March 2025), the ACCC estimates Woolworths has 38% of supermarket grocery sales nationally and Coles has 29%, with ALDI at 9% and Metcash (a proxy for independents) at 7%. Together, Coles and Woolworths hold about two-thirds of supermarket grocery sales.

The same report found that more than half of products purchased at Coles and Woolworths are on promotion, and that promotional practices can make value hard to judge. It also recommended that very large chains make application programming interfaces available with dynamic price information for third parties (same report).

Who uses a matched dataset?

• FMCG brands checking that their SKUs are priced and promoted consistently across both chains.
• Comparison and budgeting apps that need clean pairs before showing shoppers a "cheaper at" badge.
• Retail analysts and economists tracking price gaps, shrinkflation signals and promotion depth over time.
• Independent grocers and wholesalers benchmarking shelf prices against the two majors.

The 4 Australian grocery platforms that matter and how they differ

Coles and Woolworths are the core pair, but a complete dataset usually adds Aldi and IGA. Each one changes how matching works. Independent consumer testing shows why: in CHOICE's June 2026 basket survey, a 16-item basket without specials cost $68.60 at Aldi, $77.53 at Woolworths, $78.58 at Coles and $91.65 at IGA.

IGA's own Price Match terms show how retailers think about equivalence: they match on identical pack size, colour, flavour and variety, adjust different sizes pro rata, and fall back to the closest comparable product when no identical item exists. A good matching dataset makes that same logic explicit and auditable.

The 10 data fields you need for accurate matching

Matching quality depends on what you capture. These are the fields we recommend collecting from each retailer's public product pages:

• Retailer product ID. The stable product code shown in the product URL or page, used as the primary key per retailer.
• Product name. The full title as displayed, kept raw and also in a cleaned form for matching.
• Brand. Captured separately, with own-brand labels (such as Coles or Woolworths ranges) flagged.
• Pack size and unit. Split into a number and a unit (g, kg, mL, L, each) so sizes can be compared.
• Pack count. Multipacks such as 6 x 375 mL need a count field, not just a total volume.
• Variant attributes. Flavour, fat level, scent, colour or format – the details that make two items different.
• Shelf price and promo price. Both prices, plus the promotion label and any multibuy rule.
• Unit price. The displayed price per 100 g, 1 kg, 100 mL or each, as shown on the site.
• Category path. The retailer's own category breadcrumbs, used to narrow candidate matches.
• Barcode (GTIN) where public. When a public page shows it, it is the strongest identifier.

Barcodes help most, because the same branded item usually carries the same barcode at every retailer. They are not always shown on retailer pages, though, so most pipelines still need attribute-based matching as a fallback.

How to collect, clean and match the data

Coles vs Woolworths product matching works best as a staged pipeline rather than one big fuzzy-match step. Here is the method we use.

• Collect. Capture public product pages from each retailer on a fixed schedule, for one location or a set of postcodes, because prices and ranges can vary by store.
• Normalise. Clean names, standardise units (1000 g becomes 1 kg), split multipacks and separate brand from title.
• Block. Only compare items in the same broad category and brand group, which cuts false matches and speeds things up.
• Match. Use barcode first, then brand + size + variant rules, then text similarity on cleaned names for the remainder.
• Score. Give every pair a match type (exact, equivalent, no match) and a confidence score.
• Review. Send low-confidence pairs to a human reviewer and record each decision in a QA log.
• Compare. Calculate gaps on unit price as well as shelf price, with and without promotions.

What do match types and scores look like?

Ask for each pair to carry its match type, a confidence score and a QA log of reviewed decisions, so your team can filter by confidence before reporting; the exact outputs are agreed when we scope the project.

Common matching mistakes to avoid

• Matching on name only. "Full cream milk 2L" and "Lite milk 2L" look alike but are different products.
• Ignoring multipacks. A 10-pack and a 24-pack of the same can should not be treated as one price point.
• Mixing promo and shelf prices. Keep both, and flag the promotion type, or gaps will swing week to week.
• Assuming national prices. Prices and availability can differ by store, so record the location used.

Unit pricing and promotion strategy by category

Unit price is the fairest common currency when pack sizes differ. In August 2025 the federal government announced a consultation on strengthening the Unit Pricing Code, including options on readability, retailer coverage, consistent units of measure and civil penalties (Treasury media release). As of the Government's May 2026 response to the Senate committee, it was still consulting on options and had not announced final changes or a start date (Treasury, Government response, May 2026).

Until any changes take effect, your dataset should calculate its own unit price from pack size, rather than relying only on the displayed figure. The table below shows how strategy changes by category.

Use cases for a matched Australian grocery dataset

• Price gap monitoring. Weekly reports showing where Coles or Woolworths is cheaper on exact matches, by category.
• Promotion tracking. Measure how often matched products are on special at each chain, and how deep the discount is.
• Shrinkflation signals. Flag products whose pack size falls while the shelf price holds steady.
• Private-label benchmarking. Compare own-brand equivalents across Coles, Woolworths and Aldi baskets.
• Comparison app feeds. Supply clean pairs and unit prices to a consumer price comparison product.

A typical project we see: an Australian grocery comparison app needing a matched Coles and Woolworths dataset with weekly promotion updates, delivered as CSV. [Add result from delivery team]

Legal and compliance basics

Matching should only use publicly available product and price information. Respect each retailer's robots.txt, terms of service and sensible rate limits, and never log in, bypass security or collect personal data such as shopper reviews tied to individuals.

• Public data only. Prices, product names, sizes and categories shown to any visitor.
• Polite collection. Scheduled, throttled requests rather than heavy, continuous crawling.
• Watch for official feeds. If price APIs recommended by the ACCC become available, prefer them where they fit your use case.
• Clear labelling. Mark equivalent matches as judgement calls, not exact matches, in any published comparison.

Build vs buy: choosing a matching provider

Building in-house gives full control, but matching needs ongoing upkeep: ranges change, products are renamed and promotions rotate weekly. Buying a managed feed moves that upkeep to the provider.

When comparing providers, ask for a sample file with match types, a clear definition of "equivalent", the location used for prices, and how often the data refreshes.