Adding a Third Grocery Chain Without Breaking the Schema: Aldi Australia

Introduction

A grocery comparison application was already running on Woolworths and Coles data feeds. Adding Aldi Australia was not a new data project — it was an integration constraint. The new feed had to arrive in the identical schema and structure as the existing two, on a fixed twice-weekly schedule, or the client's downstream product-matching logic would need rework. Actowiz Solutions delivered Aldi category-level collection in JSON via Google Drive and Pull API, structurally interchangeable with the existing integrations.

The Problem: Onboarding Platform Three Is Not Like Building Platform One

When a comparison product launches, the first two data sources define the schema. Everything downstream — product matching, unit normalization, price-history storage, the comparison UI itself — is written against that shape.

By the time the third source is added, the schema is no longer a design decision. It is a fixed contract. And this is where most grocery data expansions go wrong: the vendor delivers a technically excellent Aldi feed with slightly different field names, a different nesting structure and its own way of expressing pack size. The client then spends weeks writing an adapter, and every future source repeats the cost.

This client stated the requirement up front and precisely: the Aldi feed must follow the same schema and structure as the existing Woolworths and Coles integrations, so downstream processing is unchanged.

That reframes the engagement. The deliverable is not data about Aldi. It is data about Aldi that is indistinguishable, structurally, from data the client's application already consumes.

Why Schema Parity Is Harder Than It Sounds

Three chains describing the same grocery aisle do not describe it the same way.

• Category taxonomies don't align. Aldi's category tree is not Woolworths' category tree. Aldi runs a much narrower assortment built largely on private label, where Woolworths and Coles carry deep branded ranges. A one-to-one category mapping does not exist, so the collection scope has to be defined against the client's target categories rather than any single retailer's navigation.
• Private label breaks brand-based matching. Comparison logic that keys on brand plus pack size works well between two full-range supermarkets. Introduce a chain where most products are exclusive private label and a large share of items have no cross-chain equivalent at all. The schema must accommodate "no match exists" as a valid, expected state rather than a data quality failure.
• Pack and unit expression differs. The same quantity is written differently across chains. Since the existing integrations already normalized units in a particular way, Aldi's expressions have to resolve into that same normalization — not a new, cleaner one.
• Attribute availability differs. A field that Woolworths exposes may not be published by Aldi. The parity requirement means the field still has to exist in the output, correctly marked as unavailable at source, rather than being dropped from the JSON. A missing key breaks a consumer expecting it; an explicit null does not.

What We Built

Collection design

Category-based traversal across the client's predefined Aldi Australia category set, running on a fixed schedule: Wednesdays at 08:00 and Saturdays at 07:00 AEST. Twice-weekly matched the cadence of the existing feeds and the client's own refresh cycle — grocery base prices do not require daily collection, and the promotional cycle is weekly.

Fixed clock times, not "twice a week", mattered here. The client's downstream job runs on a schedule; a feed arriving at an unpredictable hour is a feed that gets processed a day late.

Schema parity implementation

Rather than designing an output format, we worked backwards from the existing integrations:

• Field inventory — of the live Woolworths and Coles feeds — exact key names, nesting depth, data types, null conventions.
• Source mapping — each target field mapped to its Aldi source, or explicitly marked as unavailable at source.
• Normalization alignment — Aldi pack and unit strings resolved through the same normalization rules already applied to the other two chains, so a 500 g product reads identically regardless of origin.
• Structural validation — every delivered file validated against the schema contract before release. A structurally non-conforming file is not delivered; it raises an alert.

That fourth step is the one that makes parity real rather than aspirational. Schema drift in a JSON feed is silent — the consumer either ignores an unexpected key or throws on a missing one, and neither failure points back to the source.

Delivery

JSON, dual channel: written to Google Drive on the fixed schedule, and available via Pull API for the client's application to fetch on its own trigger. Dual delivery meant the client could migrate from file-based to API consumption at their own pace without a data-side change.

What Made It Work

• Read the existing contract before writing anything. The temptation on a third-source engagement is to deliver the best schema you can design. The client does not want the best schema; they want their schema. Field inventory first.
• Explicit nulls beat dropped keys. When a source doesn't publish a field, the parity-preserving answer is to emit the key with an unavailability marker. Dropping it saves bytes and breaks consumers.
• Validate structure, not just content. Price-sanity checks catch bad data. They do not catch a renamed key or a changed nesting level — which is the actual risk in a parity engagement.
• Fixed clock times, stated in the client's timezone. "Twice weekly" is a frequency. "Wednesday 08:00 AEST" is an integration contract.

Where This Pattern Applies

Any product built on multi-source retail data hits this problem at source three and every source after. It is the standard shape of work for grocery and retail comparison apps, price-comparison platforms, marketplace aggregators, and retail-analytics vendors expanding coverage into new chains or new countries.

Our current grocery and retail footprint includes Aldi Australia, Woolworths and Coles, Wegmans, Sam's Club, Costco, Tesco-class UK grocers, Metro Cash & Carry, BigBasket, DMart, JioMart and Udaan, across India, the US, UK, Australia and Southeast Asia.

Typical onboarding scope for a new chain: schema inventory, one category subset, two delivery cycles. It proves structural interchangeability before full category coverage is committed.

FAQ

Can a new retailer feed match the schema of feeds we already use?

Yes, and it should be specified as a requirement at the start rather than discovered at integration. The approach is to inventory the existing feed's exact field names, nesting and null conventions, then map the new source into that contract — including emitting keys for fields the new source does not publish.

How often should grocery price data be collected?

Twice weekly suits most grocery base-price and weekly-promotion tracking, since supermarket promotional cycles are weekly. Daily or multiple times daily is warranted for quick commerce, where dark-store inventory turns over intraday.

What happens when a retailer doesn't publish a field the others do?

The field remains in the output schema with an explicit unavailability marker. Removing the key breaks any consumer expecting it and makes "not published" indistinguishable from "collection failed".

Can data be delivered by both file and API?

Yes. Dual delivery — scheduled file drop plus Pull API — is common and lets a client migrate consumption method without a data-side change. Push API is also available where the client prefers to be notified.

How do you handle private-label products that have no cross-chain equivalent?

The schema treats "no match" as a valid expected state rather than an error. On discount chains where private label dominates, a large share of products legitimately have no comparator, and forcing matches produces worse output than acknowledging the gap.

Which grocery chains are already covered?

Aldi Australia, Woolworths, Coles, Wegmans, Sam's Club, Costco, UK grocers, Metro Cash & Carry, BigBasket, DMart, JioMart and Udaan among others, across India, US, UK, Australia and Southeast Asia. New chains are onboarded on request.