Marketing Mix Modelling Is Not a Magic Box: Lessons From a Media Agency’s Journey With Minute MMM

Sep 08, 2025

Introduction: A Real Story Behind the Numbers

Not long ago, one of our media agency clients reached out with a curious inquiry while using Minute MMM, our self-service marketing mix modelling (MMM) SaaS platform.

This agency had been exploring the tool for some time and genuinely appreciated its simplicity, flexibility, and the control it gave them over their own marketing data. They described it as “handy, easy-to-use, and surprisingly adjustable.”

But during one of their experiments, something fascinating happened: they adjusted the settings so that nearly every media channel contributed 100% impact to sales outcomes.

On the surface, it looked like a confidence move. But in reality, it opened the door to an important conversation about what MMM is, what it is not, and how agencies and brands can use it responsibly.

This article will take you through that story, what it reveals about MMM as a methodology, and the bigger lessons for both media agencies and brands.

Reviewing the user’s screenshot, I was stunned to see almost all media variable impacts set to 100%. In marketing mix modelling, this kind of assumption can be alarming, because it ignores the reality that no single channel—or even all channels together—can realistically account for 100% of business outcomes
When I first saw the screenshot from the user’s MMM setup—where nearly every media variable impact was set to 100%—I was genuinely shocked. It was the kind of moment that makes your jaw drop, because no realistic marketing mix modelling scenario should ever assume every channel drives 100% of results.

What Makes Minute MMM Different

Before diving into the specifics, let’s set the stage.

Minute MMM is designed as a self-service MMM tool that empowers marketers and agencies to measure and optimise their media investments without needing a PhD in data science.

One of its key features is a customisable setting that allows users to define their expectations about media variables and their impact on key performance indicators (KPIs).

In practice, this means:

  • If you believe search contributes 10% to your sales, you can set it that way.
  • If you’re uncertain, you can leave the system to provide a model-driven suggestion.

This flexibility is a powerful way to bridge data with human judgment and domain expertise. But it also brings responsibility.

The Reality Check: MMM Has No “Ground Truth”

One of the most important lessons in marketing science is this: MMM does not come with a built-in ground truth.

Unlike a physics experiment where gravity always pulls at 9.8 m/s², marketing systems are messy, interconnected, and ever-changing.

MMM works by creating a statistical approximation of reality. It can be shaped by:

  • The quality of input data
  • The assumptions you make
  • The expectations you set

That’s why the customisable expectations in Minute MMM directly reflect the truth of the discipline: MMM can be anything you make it.

And that’s both empowering and risky.

In the screenshot shared by my Minute MMM user, the media variable expected impact shares appear unusually elevated, with several values nearly touching 100%. This highlights a common challenge in marketing mix modelling: setting assumptions too high can create misleading insights about media performance
This is the screenshot I received from the Minute MMM user—showing media variable expected impact shares set unreasonably high, in some cases even reaching 100%. In marketing mix modelling, such extreme assumptions can distort results and lead to unrealistic outcome.

Why People Think MMM Is a “Magic Box”

Many marketers and even seasoned agencies fall into a common trap: they treat MMM as a black box.

The thinking goes like this:

  • Feed in your marketing spend data.
  • Press the button.
  • Out comes the “truth” about media effectiveness.

But that’s not how it works. MMM is not a vending machine that returns the answer you desire. It is a modelling approach that requires careful interpretation, judgment, and sometimes, humility.

When our agency client set every media channel to 100%, it wasn’t a failure. It was a reminder of how easy it is to slip into the mindset of “the model will prove whatever I need it to prove.”

The 100% Setting: What It Tells Us

So why would someone set every channel to 100%?

There are several possible reasons:

  • Optimism bias: Believing that every channel works equally well.
  • Pressure: Agencies often need to prove their value to clients.
  • Misunderstanding: Confusing potential impact with guaranteed impact.

Whatever the motivation, the result highlights a knowledge gap. No media channel can realistically be responsible for 100% of sales.

This is not just an academic point. It has real consequences.

The KPI sales decomposition from this marketing mix modelling (MMM) output highlights the impact of each media channel. When nearly all media variables are set to extremely high levels or 100%, the organic base (black bar) is pushed into negative territory, which defies realistic marketing science principles. In MMM, the organic base reflects natural, ad-free sales growth, and it cannot be negative for actual sales KPIs
This is the modelling output showing KPI sales decomposition by individual media channels. Because almost all media channels were set very high—or even 100%—the organic base (represented by the black bar) becomes negative, which is unrealistic. In marketing mix modelling, the organic base represents natural sales growth without advertising influence, and for KPI = sales, it can never be negative.

The Risks of Unrealistic Assumptions

When assumptions don’t align with reality, two risks emerge:

1. Business Risk

If a marketer or agency claims that one or more media channels deliver 100% of the sales, they create unrealistic expectations.

When the numbers inevitably fall short, the client relationship is strained. Trust is eroded. And in the worst cases, business partnerships collapse.

2. Reputation Risk

Agencies want to be seen as credible, trusted partners. But setting media impacts to extreme, unrealistic levels can make them look disconnected from reality.

Instead of being celebrated for hard work, they risk being questioned—or even replaced.

The Critical Role of Judgment

That’s why critical thinking is essential when using self-service MMM tools.

Every media assumption should be:

  • Justified with reasoning or evidence.
  • Grounded in business context.
  • Flexible enough to adapt when data tells a different story.

The role of MMM is not to flatter existing beliefs, but to challenge and refine them.

The share of spend compared to the share of media contribution indicates an exceptionally high ROI. Even with minimal budget spent on certain media channels, the reported impact seems 10x greater than expected, which is a clear red flag in marketing mix modelling, as results this extreme are unlikely to be realistic
This analysis of share of spend versus share of media contribution reveals an unbelievably high return on investment. Despite a limited budget allocated to the media, it appears to generate a tenfold impact—an outcome that is far beyond what’s realistically achievable in marketing mix modelling.

Independent Validation: Why Brands Look Elsewhere

In past articles, we’ve written about how brands often seek third-party MMM vendors to validate or challenge the insights provided by their media agencies.

These articles discuss how brands’ trust can erode when media metrics are tracked and measured improperly, often resulting in wasted marketing budgets and disappointing outcomes.

From Radio Waves to Landing Pages: Measuring 2GB Radio Ads with Google Analytics for a Sydney Apartment Launch

The $10 Million Lesson: When "Promising Media Performance" Masks a Revenue Disaster

The Pitfalls of Vanity Metrics: How an Online Flower Brand Wasted $500,000 on Misguided Marketing

This isn’t because they distrust agencies outright. Rather, it’s because they need reassurance that the story they’re being told matches a more neutral, independent perspective.

By turning to external providers, brands gain:

  • Clarity – a second opinion that either confirms or corrects assumptions.
  • Transparency – insight into the logic behind the numbers.
  • Confidence – the assurance that investment decisions are based on more than just optimistic reporting.

Agencies as Heroes, Not Victims

The point here is not to criticise agencies. In fact, it’s the opposite.

We want agencies to be seen as heroes of the media world—proving impact, demonstrating value, and earning the trust of the brands they work with.

MMM, when used responsibly, can be the tool that makes this possible. It provides the data backbone that shows a brand:

  • “Here’s where your marketing spend really makes a difference.”
  • “Here’s why the work we’re doing is driving results.”
  • “Here’s how we can optimise for even greater growth.”
Media agencies have the potential to be heroes by effectively planning, trading, and measuring ad campaigns. To achieve this, they must rely on accurate marketing mix modelling (MMM), proper tools, and realistic assumptions—ensuring their efforts truly deliver measurable impact
Media agencies can become heroes when their advertising campaigns are carefully planned, executed, and measured. However, this success depends on using the right tools and making assumptions grounded in reality—without proper support, even the best campaigns can fall short.

A Shared Mission: Supporting Agencies and Brands Alike

At More Than Data, our mission is not to take sides. Whether you are a media agency or a brand/SME, we are here to support you.

  • For agencies, we provide a platform to prove the value of your work.
  • For brands, we offer a transparent, independent measure of media effectiveness.

Ultimately, both sides want the same thing: clarity, confidence, and business growth.

Key Lessons From This Story

This agency’s journey with the 100% setting wasn’t a mistake—it was a learning opportunity. From it, we can extract three big lessons:

  • MMM is not a magic box. It requires thoughtful setup and critical thinking.
  • Assumptions matter. Setting unrealistic expectations can harm trust and credibility.
  • Transparency builds relationships. Independent validation strengthens both agencies and brands.

Conclusion: Facing Reality, Building Trust

Marketing is complex. Media impact is rarely clear-cut. And no model, however advanced, can replace the need for sound judgment and honest communication.

When used responsibly, MMM is not just a measurement tool—it’s a bridge of trust between agencies and brands.

That’s why at More Than Data, we celebrate both the curiosity of agencies experimenting with our platform and the brands who seek clarity. Together, we can create a marketing ecosystem where insights are realistic, investments are transparent, and growth is shared.