menu_open Columnists
We use cookies to provide some features and experiences in QOSHE

More information  .  Close

Local media is losing share because it is hard to buy and hard to analyse

22 0
02.11.2026

At Mamamia’s upfront this week, Mia Freedman told advertisers that 75 cents of every advertising dollar spent in Australia goes to the tech platforms, and asked them to give homegrown media a bigger part of the mix because it works. The explanation that has run through this upfront season, and that Tim Burrowes put in writing in Mumbrella’s Best of the Week, goes further: brands keep funding the platforms because their marketing mix model says it will improve short-term results by “a couple of basis points”. 

I agree with Freedman that local media works. But blaming “short-term ROI models” is completely wrong, and is the opposite of what the models have been saying. It is also the most convenient diagnosis available, because it is the only one that requires publishers to change nothing. There’s plenty of evidence that marketing mix models are telling brands to buy more local media, and that many local media publishers have room for further investment. The money keeps leaving because publishers have made advertising hard to buy and hard to analyse, not because the models say so. 

What the independent models say 

Earlier this year we published Shorting the Long, a read of two years of modelled outcomes across 83 enterprises in Australia and New Zealand. In 2025 those brands cut cinema, radio and free-to-air TV hardest. Those are the channels with the longest carryover, with adstock half-lives of eight to twelve weeks, and the share they lost went to search and social, which carry for one to two weeks. 

Enjoying Mumbrella? Sign up for our free daily newsletter.

We then measured each channel’s saturation: its spend divided by the point at which its returns start to diminish, so anything above 1.0 buys less with every extra dollar. Search was running at 3.66 times that point and social at 2.26. Most of the brands buying search, and most of the brands buying social, were spending more than double the efficient level. Free-to-air TV sat at 0.46, less than half-saturated, while remaining the single largest source of unit sales in the set at 18.1%. Cinema was at 0.27 and print at 0.13. 

That runs completely contrary to the claim that “short-term ROI doesn’t support local publishers”. It clearly does, and we’re one of the largest modellers in the country. Brands that leaned toward brand channels also grew unit volume about four times faster than those that leaned toward performance.

Obviously growing brands can afford more brand spend, but it points the same way. If the models were driving the shift to the platforms, the money would be flowing the other way. 

Mia Freedman at Mamamia’s 2027 upfront

Who builds the model matters 

There is a version of the MMM complaint I agree with. I said at Mumbrella360 in May that if the measurement’s free, ad sales is the product, about the free models Google and Meta give away through Meridian and Robyn. Ebiquity’s Ruben Schreurs made the same point at the same conference: those tools over-index on short-term metrics. 

There’s clearly a huge conflict when publishers and sellers of media inventory commission these studies. There’s clearly an issue with partnership programs being built without absolute transparency to the advertiser. I’ve been vocal on that, amongst others. 

But there’s also plenty of evidence that some of these studies line up with independent results. Seven’s cross-channel study with Prophet and BYD modelled more than 320,000 test drives against over $20m of media, and found linear TV, BVOD and out of home among the most cost-efficient channels, with efficiency gaps of up to 20 times across the mix. I find the result credible because it lines up with independent data. I would still treat it the way I treat Meridian: I treat a model commissioned by the party selling the inventory it measures as a sales tool until something independent confirms it. 

The principle cuts both ways, and it is about governance. Independence means the advertiser builds, governs and pays for the model, whoever supplies the data and whatever the model connects to. A model a seller commissions to measure its own........

© Mumbrella