Go to market with DISTintel.ai: the trade map brand owners are missing

The asymmetry nobody designed, and everybody lives with

Australia’s off-premise trade is concentrated, sophisticated, and talks back to you in data. The major retail groups know exactly what sold, where and to whom, and they will sell you that view — on their terms, bounded by their own estate. Retail media has become a genuine planning input rather than a post-campaign activation line, and brand owners have built real capability around it.

The on-premise is the opposite. It is tens of thousands of individually-licensed businesses, reached through wholesalers who quite reasonably regard outlet-level sell-through as their own commercial asset. A brand owner looking at the on-premise sees depletions and trading names. Not premises. Not owners. Not change.

The half of the market that builds brands is the half nobody sells you data on.

That matters because the on-premise is where trial, ritual and recruitment happen. A house pour, a back-bar listing, a tap contract, an opening-night placement — these are the decisions that put a brand into someone’s hand for the first time. And they are made months before a venue trades, by an operator whose name is on a public licence application that almost nobody reads.

What the registers already contain

Every liquor licence in Australia is published. So is every application, every company, every director, every business name, every insolvency notice. The information is free. It is also effectively unusable in its native form — scattered across eight jurisdictions, five incompatible class vocabularies, and thirty years of inconsistent data entry, with no identifier joining a venue to the company that runs it.

The work is not access. The work is assembly, resolution, and keeping it current. That is what DISTintel does: 51,767 liquor licences across 63 normalised classes, geocoded to lat/long, postcode and ABS statistical area, resolved against a 21.7-million-entity national register, and monitored continuously. Applications are ingested every fifteen minutes. ASIC notices land on the same clock.

Five things a brand owner can do with it

These build on each other. The first is infrastructure and everything downstream inherits it; the last only becomes possible once the first four are running.

1. Fix the outlet master

   Most brand owners’ outlet masters are assembled from distributor files. They carry trading names rather than legal entities, stale addresses, duplicates, venues that closed two years ago, and no ownership at all. It is the foundation that field routing, national account management, distributor reconciliation and retail-media outlet matching all stand on — and it is usually the weakest data in the building.

   Replace it with the licensing regulator’s own record, resolved to ABN and ACN, with a crosswalk maintained back to your existing outlet IDs so it lands inside your systems rather than beside them. One correction, many beneficiaries.

   What this needs: your outlet master, and a match report. Nothing else.

2. See venues before they open, and the moment they change hands

   Every liquor licence application in New South Wales is posted to a public notice board. There are 4,595 live on it as this is written, up from 4,595 in July. For a wholesaler each one is a lead. For a brand owner it is something better: a placement window. A back bar gets specified long before a venue trades, and the earliest legal artefact of that venue’s existence is its licence application.

   Transfers matter even more. A licence transfer means the venue has changed hands, which means every incumbent pouring arrangement inside it has just gone back on the table. It is the single highest-conversion moment in the on-premise, and it is public.

   Also on the feed: condition changes, including extended trading hours — a late-night trade signal that skews hard to spirits and RTDs. And limited and single-function licences, which are effectively the state’s events market as a live stream: every festival, race day and one-off that will need a beverage partner.

3. Resolve the group behind the venue — then read it on a cadence

   A hotel group does not trade as a hotel group in any register. It trades as thirty-six companies, twenty-three licences and eighteen individually-named licensees, often with the freehold in a fourth structure again. The person who signs the licence is frequently an approved manager with no ABN; the entity that owes you money is the operator company behind them; the landlord may be a third party or may be the group itself, and which one it is tells you whether that venue is likely to change hands.

   Put one ABN in and the estate comes out — related companies, licensed venues, named managers, landlords, mapped. That is what makes a national on-premise agreement verifiable: the difference between believing an agreement reaches two hundred venues and being able to demonstrate that it reaches one hundred and seventy-four, before you price it.

   Run that same walk on a cadence rather than on demand and it stops being a one-off answer and becomes an economic read. Insolvency and administration notices, licence cancellations and surrenders, and new formations set against closures all land on the same records — so trade health reports by licence class and by postcode, month after month, either nationally or across only the estates you actually supply. Which territories are net-adding venues and which are shedding them is not a survey question. It is a count.

   Same walk, four uses: verify the contracted estate, find the whitespace inside groups you already supply, see consolidation in the register before it reaches the trade press, and get a monthly read on trade health per postcode — including a contagion flag when one insolvency inside a control structure puts every venue that structure reaches at risk.

4. Plan media against trade geography, not just population

   Media in Australia is bought against population, audience panels and retailer estates. None of those know where the licensed venues are. Once every venue carries a coordinate, a postcode and a statistical area, trade geography becomes a planning input.

   Build an opportunity index per postcode from venue density, venue-type mix, venue scale, formation and closure rates, catchment demographics, and your own depletions. Then use it where media is actually bought. Out-of-home is the clearest example: media owners sell roadside, bus and transit panels by audience count, and nobody sells them by stockist proximity. Ranking panels by the number of venues within walking distance that actually pour your brands is a different and better buying logic, and the data to do it exists today.

   Trade audiences too: licensees, approved managers and operator companies can be matched into LinkedIn, Meta and Google as business audiences for trade campaigns — by hashed identifier, with the hashing done before anything leaves the environment. Consumer targeting stays geographic: radius and postcode weighted by the index, so consumer spend concentrates where the liquid is actually available to buy.

5. Close the loop, so the map learns

   This only becomes strategic when it runs both ways. You hold outcomes we will never see — depletions, media flights, activation calendars, retail-media results. We hold structure you cannot buy. Joined at postcode, statistical area and venue rather than at person, each side keeps what it must and both sides learn.

   What comes back is spend-to-outlet-response at a geographic grain your mix models currently have to approximate, distribution movement separated into venue formation versus venue conversion, and an allocation recommendation that improves every quarter because it is being fed. This is also the part a competitor cannot replicate, because it is built out of your outcomes and our structure together.

   Never joined at the person. The loop runs on geography and premises. We do not need your consumer data, and we do not ask for it.

The line we will not cross

An alcohol brand cannot afford a clever data supplier. Register data is trade data: businesses, premises, and the people a regulator publicly names as responsible for them. It is not consumer data and we will not let it be used as though it were. The obligations here are real — the Privacy Act and the Australian Privacy Principles, the Spam Act, the ABAC responsible-marketing code, and each advertising platform’s own alcohol policy. We would rather say this before anyone asks.

“Then how do you help us target anyone?”

It is the fair objection, and the answer reframes the list above. The constraint on a big brand’s targeting has never been reach — you can already reach every drinker in Australia through platforms that know them far better than a licence register ever will. The constraint is knowing which geography is worth reaching, and that is the question nobody can currently answer for you.

We don’t sell you the audience. We tell you where to point the one you already have.

In practice that runs six ways, none of which moves a consumer record. Geography is already a targeting input — a ranked postcode and SA2 list loads directly into Meta, Google, DV360, radio and every out-of-home buy, and for a brand whose real constraint is availability, weighting by where the liquid can actually be bought beats guessing at demographics. Audience definition rather than audience data: we write the targeting spec and the platform builds the audience from its own consumer graph. Clean-room joins, where we contribute the trade and geography layer keyed on postcode and venue, the retailer contributes the consumer side, and neither party sees the other’s rows — the same governance model as any existing retailer data collaboration. Store-level retail media, where the networks already let you weight by individual store and nobody can currently tell you which ones deserve it. Trade audiences, the one place a list does change hands, because licensees and operator companies are businesses named by a regulator and targeted as businesses. And geo-holdout measurement — matched test and control postcodes, read against your own depletions, which is the cleanest incrementality method available and needs no consumer identity at all.

Where the data lives, and whose it is

Any brand owner sharing outlet-level commercial data with a supplier has a procurement conversation ahead of it whatever the idea is worth, so the architecture answers it first. Three tiers, three homes, three different owners — separated by design rather than by policy.

The register layer is public government data: licences, entities, directors, notices. It is our asset and it holds no customer data. It is also deliberately singular and international — one model, queried the same way in every market we cover — because a brand owner operating across dozens of countries does not want a different data structure in each of them. Fragmenting the register tier by jurisdiction would destroy the one thing that makes an Australian proof worth anything elsewhere.

Said before anyone has to ask: that means the bulk entity register — the 21.7 million ABR and ASIC records — is hosted offshore, in the United States, and our privacy policy names the region rather than leaving you to find it. That is a design decision, not a gap awaiting migration. It is public data, published by government for anyone to read, and it carries no residency obligation that would justify splitting it into eight national copies. What does stay in Australia is the data where residency actually bites: our Sydney tier holds the registers that carry genuine obligations, personal information among them — and your lake is Australian-resident from day one. Global layer for public data, local lakes for yours.

Your lake is everything you share with us — outlet master, sell-through, media spend by geography, contract scope. It sits in an Australian-resident private cloud data lake on dedicated tenancy, at rest and in processing, never commingled with another customer or with the register layer. It remains your intellectual property and your ownership from beginning to end. We hold it to do the work you asked for and nothing else, it is exportable on demand, and it is destroyed on exit to a stated timetable.

Agreed cohorts are the narrow exception, and they are named rather than assumed. De-identified, aggregated derivations of the join — defined in writing before anything is drawn, down to the fields, the aggregation grain and the minimum cell size — become ours, and they are what the platform learns from. That is the consideration that makes the loop worth running for both sides. We would rather write it down than let it happen quietly, because the version that happens quietly is the one that ends badly.

What we are not

We hold no consumer purchase data and we do not replace scan, panel or depletion measurement. Those sources answer what sold. We answer where it can sell, who controls that, and what changed. The two are complementary and we would rather be precise about it than oversell.

Coverage is also not yet national. New South Wales is deep. Victoria, South Australia, Tasmania and the ACT are live. Queensland is built but not yet running, and Western Australia and the Northern Territory are scoped work. Anyone claiming a finished national on-premise register today is describing an ambition, not a product.

Nor is every licence resolved to a company. 46,395 of 60,538 records carry an entity link — 77% — and the rate varies by register because the registers differ: the ACT reaches 94%, NSW and Victoria sit near 80%, South Australia manages 59% despite publishing no ABN or ACN at all, and Tasmania trails at 37%. That last one is a gap, not a rounding error, and we would rather name it than have you find it.

How this starts

Nobody should sign a data programme on the strength of a meeting. The sensible first step is small, fixed-price and falsifiable: give us your outlet master, and in about five weeks you get back a match report against the licensed universe — matched, unmatched, duplicated and closed, itemised — plus group resolution on your twenty largest on-premise accounts and a twelve-month backtest of the formation feed against venues you already supply.

Either that document changes how you see your own outlet data or it doesn’t. If it does, everything above becomes a straightforward conversation about scope. If it doesn’t, you have spent five weeks and found out cheaply.

And Australia is not really the prize; it is the proof. Every common-law market publishes a licensed-premises register — the United Kingdom through its licensing authorities, Ireland, New Zealand, and every US state through its own beverage-control board. The pipeline built for New South Wales is the pattern for all of them. A brand owner operating across dozens of markets is not buying an Australian dataset. It is buying a method.

Licence, party and entity counts are live platform figures drawn from public registers (NSW Liquor & Gaming and the Victorian, South Australian, Tasmanian and ACT liquor registers, plus the ABR and ASIC) as at 3 September 2026. 51,767 is the number of licences currently in force; 60,538 licence records are held in total, including lapsed and cancelled ones. Market-structure descriptions are our own reading of the Australian route to market. DISTintel is an information service — it does not provide credit, legal or financial advice.