Media Mix Modelling (MMM): How It Drives Smarter Investment Decisions and Growth

How can Media Mix Modelling (MMM) lead to better investment decisions and higher growth? Our partners, Kinase, have taken a closer look at what it is and how it can help.

Accurately identifying what drives a customer to make a purchase is notoriously difficult—even when asked directly, many consumers may not be able to articulate the real reason behind their decision. While digital marketing cookies once promised transparency, the reality has proven far more complex. Geo-testing, although informative, can be disruptive. Today, Media Mix Modelling (MMM) has emerged as the gold standard, and with advancements in machine learning, it's more accessible than ever.

The problem used to be the familiar one of attribution: What could be the source of truth if different platforms and models gave different results? Data or AI-driven attribution purported to solve the issue, but these were biased toward the lower funnel, and ingesting data from outside web analytics immediately posed a new problem: how can an attribution model that ignores the analogue world ever be relied upon?

These pressures have led to further disruption, including the decline of first-party cookies, growing privacy concerns driven by companies such as Apple, and the deprecation of default tracking on iOS. The result is a fragmented and increasingly opaque data landscape.

Results from well-planned and executed experiments can calibrate real-time data sources, such as attribution models. The Kinase team have long been at the forefront of designing such incrementality tests. They began with 'online to store' tests for many of their hybrid retail clients, such as Dreams and Goldsmiths. Since then, they have designed and executed a plethora of tests on everything from budget incrementality, margin bidding, the real contribution of a new channel to a business (such as Programmatic or YouTube), to different attribution models.

These methodologies can be thought of as nesting in and alongside each other. Kinase always consults with its clients on the best KPIs for their goals and uses a combination of measurement tools to generate the best reporting model possible. A third layer has become a crucial part of this optimisation kit.

MMM takes elements of attribution and econometrics methodologies and builds them into new, customisable models that can provide a long-term view of your channel mix without short-term disruption.

Viewed through a layered lens, attribution remains at the core of what can still be tracked in detail on a daily basis, through tools such as consent mode, server-side solutions, and on-site analytics. Around this, incrementality testing can evaluate the effectiveness of individual channels at specific investment levels. MMM broadens that view to incorporate all channels (both on and offline), determining the contribution of each and the optimal budget allocation.

How can MMM achieve this 'everything at once unity'? It uses multiple data sources and can integrate with and confirm the results of parallel incrementality tests and attribution data. The data sources used break down into:

• Contextual: This is the bigger picture. Market positioning, vertical data, and macroeconomic trends can be fed into the model. These are elements that affect performance regardless of media type.
• Media: This is the basics of media spending and performance, which add up to contribution and effectiveness by channel.
• Total Performance: This is the overview of business performance by market.

These are fed into open-source (as with Meta's Robyn) or open-access models with proprietary data unavailable elsewhere (as with Google's Meridian). Meridian pulls in YouTube data and actual Google query volume data rather than the relativised Google Trends data that is publicly available. Kinase uses both models in its MMM solutions to create bespoke models for each MMM client's business.

With a long-running analytics department that has a track record of delivering cross-channel incrementality testing for clients such as Dreams and Nisbets, Kinase was perfectly placed to quickly establish new MMM benchmarks for its clients.

The Kinase team found from projects with clients such as Purplebricks that simply by reallocating budgets between channels, businesses can gain a 4% increase in total revenue returned. This means that MMM pays for itself and delivers a profit straight off the bat, even for businesses with well-established and individually optimised channel budgets. After this initial step, businesses have the confidence to invest further in channels where the biggest opportunities have been identified to drive further growth.

MMM best fits medium—to large businesses with multiple media lines that are looking for an efficiency boost and a clear path to growth through marketing investment.

Often, the most significant opportunities lie in upper funnel activity, less easily measured by old attribution-based methods. There, the incremental return is likely to be higher, and MMM can reveal where to allocate the spend. The more channels, the greater the benefit MMM can deliver, thus potentially unlocking more growth. It can also help answer questions such as whether high spend budget lines like PPC deliver incremental value or waste money.

However, it's not all or nothing, and that's the beauty of today's tools and flexible strategies. A combination of MMM analysis, well-run testing, and calibrated attribution can get every layer complementing each other and driving forward your success online.

Fresh Egg partners with Kinase to bring cutting-edge Media Mix Modelling solutions to our clients, combining strategic insight with technical excellence. Integrating MMM with well-designed testing and calibrated attribution, we help brands unlock more value from their marketing spend and confidently navigate an increasingly complex media landscape. If you're ready to explore how MMM can drive growth for your business, get in touch.