The MMM Ceiling: Why We Need Creative Mix Modeling

The Question That Started This

In a recent exchange with Avinash Kaushik—CSO at HMM, former Google digital evangelist, and best-selling analytics author—I posed a question that had been forming for months:

Avinash's response sharpened the challenge. He identified two structural problems that make creative measurement so difficult:

First, a "rows of data" problem—impressions by creative type are extraordinarily hard to get, and the industry's habit of producing thousands of creative variations with little cohesion to a singular concept makes it nearly impossible to find signal in the noise.

Second, an "Impact Horizon" problem—the challenge of reconciling the different time horizons of performance creative (short-term, immediate) and brand creative (long-term, cumulative, critical). Most measurement systems collapse both into a single window and lose the signal entirely.

He also offered a caution that anyone in this space should take seriously: when someone claims MMM has "solved" creative, be suspicious. Ask questions.

This article is the long answer to the question I posed. What is Creative Mix Modeling? Why can't MMM get there on its own? And what would it take to build it?

What Media Mix Modeling Does Well

First, credit where it's due. MMM is one of the most valuable tools in marketing measurement. It has earned its place for good reasons.

MMM uses econometric modeling to decompose business outcomes—typically sales—into contributing factors: television, digital, pricing, promotions, seasonality, distribution, economic conditions. By analysing two to three years of time-series data, it can estimate the incremental contribution of each channel and help brands reallocate budgets toward higher-performing media.

The best MMM implementations—from consultancies like Analytic Partners, Ekimetrics, and Nielsen, and from open-source tools like Google's Meridian and Meta's Robyn—do this with increasing sophistication. They account for diminishing returns, carryover effects, and interaction effects between channels.

MMM answers a genuinely important question: given a fixed budget, which channels should I invest in?

That question matters. But it is not the only question that matters.

The MMM Ceiling

The ceiling of MMM is not a flaw in the methodology. It is a boundary condition. MMM was designed to model media allocation. It was never designed to model creative contribution.

In a standard MMM, creative quality is not a variable. It is residual variance—the portion of the outcome that the model cannot explain through channel, timing, pricing, or seasonality. Some advanced implementations attempt to control for creative by adding binary flags (new campaign vs. existing campaign) or by including pre-test scores as inputs. But these are approximations. They do not measure what the creative work generated in the market. They measure whether its presence or absence changed the model's fit.

This limitation is well understood within the econometrics community. But it is rarely communicated to the CMOs and CFOs who rely on MMM outputs for budget decisions. The result is a systematic blind spot:

To use the metaphor from our exchange with Avinash: MMM optimises the fuel mix. It is silent on engine efficiency.

The Data Problem

The ceiling is reinforced by a data governance problem. MMM operates on media spend and sales data—both of which are typically held by the media agency or the brand's media team. Creative production cost—what was spent making the work—lives in a different system, usually with procurement, the creative agency, or the brand's production department.

MMM consultancies do not typically have access to creative production cost data. They cannot calculate creative return because they do not have the numerator. This is not laziness. It is a structural gap in how the industry shares information.

The Granularity Problem

Even if creative cost data were available, MMM operates at the wrong level of granularity for creative evaluation. MMM analyses aggregate, channel-level contribution over multi-year periods. Creative teams need to evaluate individual campaigns, individual assets, individual production investments—often within weeks of a campaign's completion.

A CMO who needs to justify next quarter's creative budget cannot wait for a two-year retrospective. And a creative director who needs to understand which campaign approach generated better returns cannot find that answer in a channel-level decomposition.

The Impact Horizon Problem

This is the problem Avinash named. Performance creative—direct response, activation, conversion—operates on a short time horizon. Its effects are immediate and measurable within existing frameworks. Brand creative—awareness, salience, emotional connection—operates on a longer horizon. Its effects compound over time but are invisible in short attribution windows.

MMM can partially address this with adstock decay modeling. But it cannot separate the brand-building contribution of a creative idea from the brand-building contribution of sustained media presence. Both show up as long-term effects. MMM cannot attribute which is which.

What Creative Mix Modeling Would Look Like

If Media Mix Modeling asks "which channels should I invest in?," then Creative Mix Modeling asks the complementary question: "what did my creative investment yield, and how should I allocate creative resources?".

CMM is not a replacement for MMM. It is the missing layer. Where MMM decomposes outcomes by channel, CMM decomposes outcomes by creative contribution. Together, they give you both sides of the equation:

A functioning Creative Mix Modeling system would need to deliver on several requirements:

1. Campaign-Level Financial Metrics

For each creative production investment, the system must calculate a financial return—not a score, not a prediction, but a ratio of investment to measurable media value generated. This needs to be expressed in the same currency as media metrics (CPMs, cost-per-outcome) so that creative and media can be compared on equal terms.

2. Cross-Channel Measurement

Creative does not live in a single channel. A campaign idea might manifest as a TV commercial, social content, out-of-home executions, experiential activations, PR, and earned media. CMM must aggregate the total value generated by a creative concept across every touchpoint—not just the paid media where impressions are easy to count.

3. Earned and Owned Value Capture

This is where CMM diverges most sharply from MMM. A significant portion of creative value is generated not through paid media but through earned media (press coverage, social sharing, cultural conversation) and owned channels (website traffic, CRM engagement, community). MMM is blind to these because they are not paid media inputs. CMM must capture them as outputs of creative investment.

4. A Creative Multiplier

The core output of CMM is what we call the Creative Multiplier—the ratio of total measurable value (paid + earned + owned) to total creative investment (production cost plus paid media spend). A Creative Multiplier of 3.0x means that for every dollar spent producing the creative work and the paid media to run it, three dollars of measurable media value was generated. This is the engine efficiency metric that MMM cannot provide.

5. Benchmarking Across Campaigns, Categories, and Markets

For CMM to be useful as a decision tool, results must be benchmarkable. Is a 2.5x Creative Multiplier good or bad? That depends on category norms, market context, and competitive benchmarks. A functioning CMM system needs a growing database of benchmarks that let teams evaluate their creative ROI relative to meaningful comparisons.

6. Addressing the Impact Horizon

CMM must account for the different time horizons of creative value. A performance campaign may generate immediate, measurable return. A brand campaign may generate lower immediate return but higher sustained value through earned media, brand salience, and cultural resonance. CMM should be able to model both—measuring the direct yield of performance creative and the compound yield of brand creative over time.

How Creative CPM Enables Creative Mix Modeling

Creative CPM was built to solve the foundational problem that makes CMM possible: giving creative a financial metric in the same language as media.

Here is how each element of CCPM maps to the CMM requirements outlined above:

Campaign-Level Financial Metrics

CCPM calculates a Creative CPM for every outcome—the cost per thousand impressions or engagements generated by the creative work, across all channels. It also calculates a Creative Multiplier—the financial return ratio of total creative investment (production plus paid media) to total measurable value generated. These are the same metric types (CPM, ROAS) that media has used for decades. Creative can now be evaluated in the same boardroom language.

Cross-Channel Measurement

CCPM aggregates value across paid, earned, and owned channels. A campaign that generates television impressions, social engagement, press coverage, and website traffic has all of those outputs valued using market-rate CPMs and summed into a total creative yield. The creative concept is evaluated as a whole, not channel by channel.

Earned and Owned Value Capture

This is where CCPM fills the gap that MMM structurally cannot. Earned media—the impressions, engagement, and coverage that creative generates without additional media spend—is valued at equivalent market rates and included in the total return. A campaign that generates $500K in paid media value and $800K in earned media value has a total creative yield of $1.3M. MMM would only see the $500K. CCPM sees both.

Benchmarking

As CCPM's database of reports grows, benchmarks emerge by category, market, campaign type, and creative format. Teams can evaluate whether their 2.5x Creative Multiplier is above or below the norm for their category. This is the feedback loop that allows creative investment decisions to be made on evidence rather than instinct.

The Impact Horizon

CCPM currently measures the direct yield of a creative investment within a defined campaign window. This addresses the performance creative horizon effectively. For brand creative's longer horizon, CCPM's roadmap includes longitudinal tracking—measuring how a creative concept's earned media value compounds over time as it enters culture, sustains conversation, and continues generating organic impressions beyond the initial campaign window.

The Combined Model: MMM + CMM

The most powerful application of Creative Mix Modeling is not as a standalone tool but as a complement to existing MMM infrastructure. Consider the difference:

MMM Alone

Your MMM tells you that television drove 18% of incremental sales, digital video drove 12%, and social drove 8%. You decide to shift budget from television to digital video because the ROAS is higher.

But you don't know whether television's 18% was driven by the creative or by the media weight. You don't know whether digital video's 12% was efficient because the creative was strong or because the audience targeting was precise. And you don't know whether the social creative is underperforming its potential or whether 8% is actually exceptional given the production investment.

MMM + CMM

Your MMM tells you the channel contributions. Your CMM (powered by CCPM) tells you the creative contributions within those channels. Now you can see:

• Television's 18% contribution came from a creative concept with a 4.2x Creative Multiplier—the creative was genuinely strong.
• Digital video's 12% came from a concept with a 1.1x Creative Multiplier—the channel ROAS was driven by targeting, not creative. The creative was average.
• Social's 8% came from a concept with a 6.8x Creative Multiplier—the creative was exceptional but media-starved. More investment here would likely yield disproportionate returns.

The allocation decision changes completely. Instead of shifting budget from TV to digital video (which the MMM-only view suggested), you now know to shift budget to social where the creative engine is most efficient, and to invest in better creative for digital video where the media is working but the creative is not.

Why Now

Three forces are converging to make Creative Mix Modeling both possible and urgent:

1. MMM Is Having a Renaissance—And Exposing Its Own Limits

The open-sourcing of MMM tools by Google (Meridian) and Meta (Robyn) has democratised media mix analysis. More brands than ever are running MMM. But as adoption grows, so does the frustration with what MMM cannot answer. Every CMO who has seen an MMM output has asked some version of: "But what about the creative?" The demand for CMM is a direct consequence of MMM's success.

2. MMM Vendors Are Approaching the Ceiling—But Cannot Break Through It

The most sophisticated MMM platforms are aware of this gap and are moving toward it. Mutinex's GrowthOS, for example, now offers "Campaign-Varying MMM" that decomposes channel contribution into creative, format, publisher, and audience attributes. This is a meaningful advancement—it shows which creative assets drove more incremental sales within paid channels.

But it is still measuring creative effectiveness within the media model. It answers "which creative performed better in this paid channel?" It does not answer "what did my total creative investment yield across paid, earned, and owned?" It cannot calculate a Creative Multiplier because it does not incorporate creative production cost or earned media value—both of which sit outside the MMM data model.

This is not a criticism of Mutinex or any MMM platform. It is a structural observation. The MMM framework is designed around media spend as the independent variable and sales as the dependent variable. Creative as a separate financial input with its own yield equation requires a different model entirely. That model is CMM.

The global MMM solutions market is projected to reach $9.4 billion by 2030. The fact that an industry of that scale still treats creative as residual variance tells you how large the opportunity is for the complementary layer.

3. AI Is Flooding the Market With Creative Variants

Generative AI has made it possible to produce hundreds of creative variants at near-zero marginal cost. This is powerful. It is also dangerous without measurement. When you can produce 200 versions of an ad, the question shifts from "can we make it?" to "which ones worked?" Volume without valuation is waste. CMM provides the valuation layer that AI-generated creative desperately needs.

4. CFOs Are Demanding Creative Accountability

In a tightening economic environment, every budget line needs to justify its existence in financial terms. Media spend has always had this accountability because it has always had metrics. Creative spend has been protected by subjective judgment—and that protection is eroding. The CMOs who survive the next budget cycle will be the ones who can answer: "Here is what our creative investment returned." CMM gives them that answer.

What This Means for the Industry

The shift from Media Mix Modeling to Creative Mix Modeling is not a replacement. It is an evolution. MMM was always half the equation—the media half. CMM completes it.

For brands, CMM means creative budgets can be justified with the same rigour as media budgets. Production investment becomes a strategic lever, not a discretionary cost.

For agencies, CMM means the ability to prove the financial value of creative work—not through awards or subjective judgment, but through market-based financial metrics. This changes the negotiation from "how much do your people cost" to "how much value did your work create."

For MMM practitioners, CMM is not a threat. It is an additive data input. Imagine an MMM that includes creative quality scores—generated by CCPM—as a variable in the model. The explanatory power increases. The residual variance decreases. The recommendations get sharper.

For CFOs, CMM means a single, comparable metric for the largest unquantified variable in marketing spend. No more "trust us, the creative was good." Instead: "The creative returned 3.4x on production cost, with 60% of that value coming from earned media."

The Road Ahead

When I proposed CMM in our exchange, I described CCPM as "the missing link" and "the only way to avoid flying blind on the most expensive part of the mix." Avinash responded by correctly identifying the structural challenges—the rows of data problem, the impact horizon, the noise-to-signal ratio—and said he would "invest in learning more about CCPM."

Those challenges are real. Creative Mix Modeling is not a finished product. It is a direction—one that CCPM is building toward with every report, every benchmark, and every client engagement that proves creative value can be quantified in financial terms.

The industry spent two decades building increasingly sophisticated models for where to spend. It is time to build equally sophisticated models for what that spending creates.

We have modeled the fuel long enough.

It is time to model the engine.