Mutinex has rebuilt its GrowthOS marketing mix modelling platform with a significant expansion in the number, granularity and affordability of the models it can run as well as a further shift to providing marketers with real, business outcome-focused plans.
CEO and co-founder Henry Innis announced the sweeping product rebuild on stage at its Marketers & Money conference in Melbourne, saying that it would allow marketers to “put your target in, test the big bets and generate evidence around them”.
“This is GrowthOS 3.0. We’re moving away from being a measurement company and moving towards being your outcomes engine. And that’s business growth outcomes, not media outcomes,” said Innis.
The new GrowthOS 3.0 will support 16,500 linked models per business to build marketers’ plans, gather the evidence to support the impact of those plans on real business measures and to hit their targets. It will be available later this year. The models will work down to SA4 regions in Australia and Designated Market Areas in the US.
Traditionally, the number of models has been a limiting factor for MMM tools to demonstrate whole-of-business impact. Built one at a time, each model costs as much as the last and needs its own history, so most brands settle for a national read of a business they run by product and region.
GrowthOS 3.0, meanwhile, replaces that with a network growth model covering the whole business: a linked model for every product in every geography, each learning from the others. Thin product lines and new regions borrow from their neighbours instead of starting from zero, so they should get a read sooner and the cost per model falls as the network grows.
“We all know in the modern economy, budgets are being cut, targets are not. That’s the thing that keeps bouncing around my mind. Plausible ROIs are not evidence,” Innis told the crowd in Melbourne.
“Seeing high ROIs and all this sort of stuff doesn’t mean anything. Traditional MMM vendors are slow, they’re opaque. If you work with a traditional MMM vendor, it’s a bit like doing an archaeology dig. If you’re in that space, it’s very hard. It’s after the fact. It tells you want you want to hear.”
You’d be forgiven for thinking that this expansion in the number of potential models would significantly increase the cost of using GrowthOS 3.0. But according to Innis, the inter-relation of the models and their learning from each other, allows Mutinex to reduce the cost significantly.
“It will be $1,000 per model annually. That is exactly how much we are going to drive the cost down. If you do not commoditise your own business and your own pricing, I always think you’re at risk of getting out-innovated. This is a dramatic cost reduction in the accessibility of these hyper-granular models, which previously would have cost an absolute fortune,” Innis said.
Marketers will also be able to enter assumptions into GrowthOS 3.0 such as inflation, interest rates, consumer confidence, geopolitical events and weather, and see their impact on the sales forecast in every plan. When indicators shift, GrowthOS will recommend which outlook to plan for, such as a hard landing or a rate rise, and re-forecast.
When the forecast slips, it will separate what marketing drove from outside factors, and Atlas, a new map view, will show where the gap sits by region.
“For a long time, we have said that models need to be reproducible under different conditions. I think this is a critically important fact. How do you make something scientific? You reproduce it. Traditional MMM has not done that,” said Innis.
“I would argue we have not scaled enough to do that as well we could have to forecast total business outcomes accurately. And you have digital clone vendors. Frankly, I don’t want to rehearse reality. I want to see things happen in reality. That’s a really important part of how we’re thinking. Every single piece of what we do has to be tracked an accountable in reality.”
Marketers can set targets by product and quarter and see multiple ways to hit the target, rated by risk. Small tests can be run and these results can be fed back into GrowthOS to feed the model.
MAITE, Mutinex’s agentic intelligence layer, will answer questions on every page, with every figure cited to its source. MAITE can also create plans ready, easily digestible and actionable for finance teams, agency partners and leadership.
Also new is PartnerMax which gives a media publishing partner a version of the plan. Launched in partnership with News Australia, it allows media partners to plan against the brand’s targets while keeping the financials confidential.
Mutinex has tightened its partnership with Tracksuit, making the latter’s brand tracking data available within MAITE.
“Finally brand and financial data can sit alongside each other to answer mission-critical questions,” said Innis.
It is also working on integrations with Claude, ChatGPT, Microsoft Copilot and Google Gemini to allow GrowthOS to work with LLMs that many marketers are using day-to-day.
Biddable Outcomes, Mutinex’s agentic media buying capability, will move into production in 2027. Biddable Outcomes will use Mutinex’s campaign-varying marketing mix model to value each publisher, format and creative by its contribution to sales, and agents will set bids in-flight from that value. The brand and its agency keep full control of the platform.
Mutinex has piloted Biddable Outcomes with The Trade Desk since early 2026, with model values reaching live bids within about an hour of each refresh and no new tags required.

