Retailer advertising is trade spend until it proves otherwise

Every supplier to Australia's large supermarket groups has had the same conversation in the past two years. The retailer's advertising team, polite and well prepared, explains that it can now reach the supplier's customers precisely: through personalised offers to loyalty members, sponsored positions in app search results, banners on its websites and screens in its larger stores. It can show which shoppers saw an advertisement and which of them bought. It would like a share of the supplier's trade budget, and it would like a commitment before the next promotional calendar is set.

I understand the appeal, and I do not think the retailers are acting in bad faith. A loyalty programme holds better evidence about what people buy than any supplier has ever had, and a wine company that sells most of its domestic volume through two chains has every reason to want access to it. My argument is narrower. Money paid to a retailer's advertising network is trade investment under a new name, and it should be judged as trade investment is judged: by the gross profit it adds that would not otherwise have been earned. Until it shows that, it should not displace spending whose returns we can already measure.

The report that marks its own homework

The difficulty starts with measurement. The standard report from a retailer's network counts the sales made to shoppers who were exposed to an advertisement within a set period. Those are attributed sales, not incremental sales. A loyal buyer of our Barossa shiraz who sees a banner and then buys the wine she would have bought anyway appears in the report as a success. So does a shopper who saw an advertisement on Tuesday and bought on Saturday because the wine was on a price promotion we had funded separately.

When we first asked for the underlying method, the answers were partial. Neither chain would share shopper-level data, which is reasonable given their privacy obligations. One would not tell us the size of the comparison group it used, or whether it used one at all. We were being asked to pay for a channel whose results were calculated by the party being paid, from evidence we could not inspect. In any other part of our business, our finance team would not accept that arrangement for a moment.

There is also the question of where the money comes from. Most suppliers have no separate budget for retailer advertising. The money comes out of trade investment, out of marketing or out of margin. In our case the proposal was to move A$12 million, about a quarter of our domestic trade budget, away from price promotions and catalogue features. Those have well-known weaknesses, but we can measure them: we know, by wine and by store cluster, what a promotion adds in volume and what it costs in margin. Replacing a measurable return with an unmeasured one is not progress, however modern the channel looks.

A control group is a negotiating position

Our answer in 2025 was a test rather than a refusal. We committed A$4 million across five formats and asked for one condition: that each format be measured against a control. For loyalty offers that meant a randomly held-out group of members, matched to those who received the offer but sent nothing. For screens in stores it meant matched stores without the screens. Our finance team defined incremental gross profit before the first campaign ran, so that nobody could adjust the definition after seeing the results.

One chain agreed for three formats. The other preferred to rely on its own reporting, so we spent very little with it, and for two quarters we lost a preferred position in its promotional calendar. That cost us perhaps A$3 million of revenue. I mention it because a supplier who insists on evidence should expect to pay something for the insistence, and should decide in advance how much it is prepared to pay.

The results were mixed in an instructive way. Personalised offers to loyalty members and sponsored placement in search returned A$1.35 and A$1.20 of incremental gross profit for each dollar spent, better than the A$0.85 our average price promotion earned. Screens in stores, home-page banners and off-site campaigns built on the chain's data each returned less than A$0.60. The formats that worked were the ones closest to the moment of purchase, when a shopper is already looking for wine. The ones that did not resembled brand advertising sold at the price of performance advertising.

The control group did more than produce a number. It changed the conversation. Once one chain had seen our method, it began offering holdout testing to other suppliers, and the second chain agreed to test loyalty offers on the same basis the following year. A supplier in a concentrated market has few ways to influence how a retailer prices its services. Insisting on a shared method of proof is one of them, and it costs the supplier far less than refusing to take part.

What a board should ask to see

Directors of consumer companies will watch retailer advertising grow in their budgets over the next few years, often described as marketing innovation. I would suggest four questions. First, what share of the money is measured against a control, and what share relies on the retailer's attributed sales? Second, where did the money come from, and what happened to the return on the spending it replaced? Third, is the incremental return expressed as gross profit, after the cost of the product, rather than as revenue? Fourth, does management know which formats it would stop buying, and has it ever stopped one?

None of this is an argument against the channel. Within a few years the retailers' networks may well be among the most efficient ways for a supplier to reach a shopper who is deciding what to buy. But that efficiency will be earned format by format, and the supplier is the party with the clearest interest in measuring it honestly. A retailer that owns the shelf, the loyalty data and the report has every commercial reason to describe its network generously. Suppliers should listen, test and pay for what the test supports.

In FY2026 we moved A$7.5 million into the two formats that earned their place and reduced the depth of our price promotions to fund them. We spent nothing on the other three. Our trade budget is no smaller than it was; it is better evidenced. That is the standard I would ask any supplier, and any board, to apply before the next commitment is signed.