From Radio to All-Digital: What Marketing Mix Modelling Revealed About Where Ad Dollars Actually Work
Table of Contents
The Problem with Gut-Feel Media Buying
For years, a national retailer with over 300 stores across Australia ran a mixed media strategy: radio in regional markets, a combination of digital and broadcast in metro areas, and a scattering of out-of-home and print depending on the season.
The logic was inherited rather than interrogated. Radio had “always been part of the mix” in regional areas. Digital was growing but hadn’t fully replaced traditional channels. Budget decisions were made based on relationships with media buyers, historical precedent, and a general sense of what felt right.
Then they started measuring.
What Marketing Mix Modelling Revealed
By implementing marketing mix modelling (MMM) across two years of sales and media data, the brand was able to isolate the incremental contribution of each channel to actual revenue.
The results were unambiguous:
| Channel | Approximate ROI |
|---|---|
| Paid Social (Meta) | 2–5x (varies by region) |
| Paid Search (non-branded) | Strong performer (with seasonal caveat) |
| Out-of-Home | ~4x |
| Programmatic/BVOD | High at low spend, saturates quickly |
| Radio | Below 1x |
Radio was consistently returning less than a dollar for every dollar spent. Not in one market. Not in one quarter. Across both metro and regional models, across the full two-year dataset, radio was the lowest-performing channel.
The Seasonal Caveat on Search
One important nuance the modelling uncovered: paid search showed very high ROI numbers, but the model flagged that search spend was heavily correlated with peak seasonal demand. The brand was spending most on search during periods when organic demand was already high, which inflated the apparent returns.
The recommendation wasn’t to cut search. It was to test search spending during off-peak periods to get a cleaner read on its true incremental value, separate from the seasonal tailwind.
This is a good example of why raw ROI numbers need context. A channel can look like a star performer while partly riding someone else’s wave.
The Decision: Go Fully Digital
Armed with the data, the brand made a significant strategic shift for their 2026 autumn/winter campaign: drop radio entirely in metro and national campaigns and go 100% digital.
This wasn’t a small tweak. Radio had been a fixture of the media plan for years, particularly in regional markets where it was seen as the primary way to reach customers. But the data told a different story. Digital channels, particularly paid social, were delivering multiples of the return that radio was generating, even in regional areas where the brand had assumed digital would underperform.
The decision was made collaboratively between the brand’s marketing team and their store network. Metro store owners, in particular, were consulted and supported the shift to digital-only.
Why Brands Struggle to Drop Underperformers
If radio was clearly underperforming, why did it take two years of data to make the change? This is a common pattern we see across brands:
Familiarity bias. “We’ve always done radio” is a powerful force. Channels that have been in the mix for years develop institutional inertia. Cutting them feels risky even when the data says otherwise.
Visibility bias. Radio feels tangible. You can hear the ad. Staff can hear it. Store owners can hear it. Digital ads are invisible to anyone who isn’t in the target audience, which can make them feel less “real” even when they’re driving more revenue.
Attribution gaps. Without a measurement framework like MMM, there’s no reliable way to compare radio and digital on a like-for-like basis. Brands end up relying on proxy metrics (reach, frequency, recall surveys) that don’t connect to actual sales.
Agency incentives. Media buying agencies may have established relationships with radio networks and planning frameworks built around traditional media splits. Shifting to digital can require renegotiation of these relationships.
The Role of MMM in Confident Decision-Making
What made this decision possible wasn’t just having data. It was having the right kind of data, presented in a way that made the trade-offs clear.
Marketing mix modelling provides:
- Channel-level ROI grounded in actual sales data, not clicks or impressions
- Saturation curves that show where each channel hits diminishing returns
- Budget optimisation scenarios that answer “what if” questions: what happens if we move 20% of radio budget to social? What if we increase total spend by 10%?
- Regional segmentation that reveals whether a channel performs differently in metro versus regional markets
For this brand, the regional segmentation was particularly valuable. The assumption had been that digital advertising wouldn’t work well in regional areas due to lower population density. The model showed the opposite: paid social ROI was roughly double in regional stores compared to metro. That finding alone justified the strategic shift.
How to Apply This to Your Media Strategy
1. Measure Before You Optimise
You can’t improve what you can’t measure. If your media decisions are based on anything other than sales-attributed data, you’re operating with incomplete information. MMM isn’t the only approach, but for brands with significant offline sales, it’s often the most practical one.
2. Challenge the Legacy Channels
Every media plan has at least one channel that’s there because it’s always been there. Identify those channels and subject them to the same ROI scrutiny as your newer investments. The results may surprise you.
3. Test in Phases
Going from radio-plus-digital to digital-only doesn’t have to happen overnight. This brand started by running the numbers, then consulted stakeholders, then made the shift for a specific campaign window. If the results validate the model, the change becomes permanent. If not, you have data to guide the next iteration.
4. Use Scenarios to Build the Case
Budget optimisation tools can model scenarios at the total level (“what if total budget increases 10%?”) and at the channel level (“what if we shift radio budget to Meta?”). These scenarios give stakeholders concrete numbers to evaluate rather than abstract arguments about channel effectiveness.
5. Revisit Regularly
Media effectiveness isn’t static. A channel that underperforms today might improve with better creative, different targeting, or changed market conditions. The model should be refreshed regularly so decisions reflect current reality, not stale data.
The Bottom Line
The shift from radio to all-digital wasn’t a leap of faith. It was a data-driven decision supported by two years of evidence. The brand didn’t abandon radio because digital was trendy. They abandoned it because the numbers showed, consistently and clearly, that every dollar spent on radio would generate more return if redirected to digital channels.
That’s the power of measurement. Not to confirm what you already believe, but to reveal what’s actually happening, and give you the confidence to act on it.
Seeda provides marketing mix modelling and budget optimisation tools that help brands understand channel-level performance and make confident media investment decisions. Get started with a measurement framework that connects spend to sales.