You keep using that word (MMM). I do not think it means what you think it means

MMMs have gotten a lot of attention lately. Some have called it a “renaissance,” but that is not quite accurate, since major brands have been using them for decades to understand how marketing and media drive their business.

However, with recent developments on the measurement front – including cookie deprecation and a wave of new vendors in the space (mostly software platforms that claim to solve or automate MMM) – there has been a clear expansion into the longer tail, including midsize and smaller brands.

With that expansion has come a lot of confusion, especially when some of these newer entrants weigh in on what MMM is and is not. One of the most basic areas of confusion is the simple definition of what an MMM actually is.

In practice, there are at least three different levels of response analyses that tend to get lumped under the MMM umbrella.

Level 1: Classic marketing mix modeling

The original commercial marketing mix modeling initiatives were developed in the early 1990s, primarily in the packaged goods space. There were academic and smaller-scale efforts before that to understand the impact of marketing and media, but they were nowhere near the sophistication of these early commercial MMM efforts.

Those early models for packaged goods were fully formed marketing mix models. They explicitly included variables for the four Ps: price, promotion, place (distribution), and product (including assortment, new products, etc.). The initial focus was to understand the incremental impact of marketing and media on the business, so “marketing mix modeling” was an appropriate name.

As the field developed, more and more entities – particularly advertising agencies – began to shift the definition toward “media” mix modeling, focusing on the impact of media rather than the broader marketing domain.

The assumption was that other forms of marketing (CRM, pricing, in-store, etc.) were not the main focus, and that value could be delivered by centering on media channels and their impact on the business.

One of the biggest assumptions baked into this shift was that the baseline (the non-incremental part of the business) could be accurately estimated with relatively few variables to explain the underlying nuances of that baseline – such as distribution and other factors that were part of the original marketing mix models.

Level 2: Media mix modeling in the open-source era

More recently, open-source approaches such as Meta’s Robyn and Google’s Meridian, along with a growing market of third-party modeling platforms, have pushed the “media mix only” idea even further. The promise is that clever algorithms can estimate the baseline without explicitly specifying many of the underlying variables, making it possible to build media-only models more quickly and cheaply.

In practice, this is not entirely true for every brand. It can work reasonably well in some circumstances, particularly for smaller, digitally focused businesses where media drives a higher percentage of business impact and the underlying base is relatively simple.

But for many brands, this approach glosses over important non-media drivers of the business and therefore limits both accuracy and usefulness.

Level 3: Business driver analysis (BDA) - beyond just MMM

In a third, more expansive definition, MMM evolves from focusing only on marketing to understanding the broader set of business drivers.

In the 1990s and beyond, as MMMs expanded into verticals such as retail, financial services, telecom, automotive, and e-commerce, it became increasingly important to include other factors in the model to account for the non-marketing forces that drive performance.

In concert with that, some of the early pioneers – notably Hudson River Group, started to target the C-Suite as the primary client; and their perspective was broader than just marketing and media. This evolution extended marketing mix models into what some was framed as full Business Driver Analysis (BDA).

This shift mattered for several reasons

1. Accuracy. Marketing and media usually explain only a portion of business performance in most verticals. Omitting major non-marketing drivers can materially undermine model accuracy. (BTW - I’ve heard some newer entrants claim that adding variables to models does not help accuracy, which is lunacy at best).

2. Value. These analyses are not inexpensive. By broadening the view to include more of what actually drives the business, the work becomes far more valuable to decision-makers. This is especially true if the model producer is a true business partner, not merely looking at media as part of the success factor.

3. Credibility. When presenting results to clients, it became essential to account for factors that contributed to business outcomes beyond marketing and media, and to call them out as reasons for changes in performance. Without accounting for them, results could reasonably be challenged.

As MMM evolved from a research-oriented exercise into a strategic management tool, these studies increasingly became C-suite tools to help manage the business and understand what was truly driving success.

Why definitions matter

These three concepts – marketing mix modeling (MMM), media mix modeling (mMM), and business driver analysis (BDA) – are very different in terms of the resources and expertise they require. Depending on the use case, the cost, timing, data requirements, and organizational implications differ significantly.

Clarifying which of these an organization actually needs is critical. Without that clarity, it is impossible to have meaningful conversations about methodology, structure, frequency, turnaround time, or even whether an off-the-shelf software solution is appropriate.

Unfortunately, these nuances are often lost on many new players in the space, especially those coming from a media, pure software or ad-tech background, and that is where much of the current confusion around MMM begins.

Here’s some practical advice:

• Be explicit about the question: Are you trying to optimize media, the total marketing mix, or the whole business?

• Ask vendors which of the three they actually provide, and what non-media drivers they include and how. Really pressure test their expertise and experience.

• Match ambition to maturity. Smaller, digital-first brands may start with media-heavy models, as long as they understand the limitations. Larger and multi-channel brands should push toward MMM or BDA, or they risk missing big drivers and getting misleading results.

• Use definitions as scoping tools: Before talking tech, timelines, and price, align internally on whether you need MMM, mMM, or BDA; this will force clearer choices on data, resourcing, and ownership.

• Treat “baseline” with skepticism: Any solution that hand-waves the baseline without explaining what sits in it and how it is modeled deserves extra scrutiny.