Why Irish and EU Businesses are Returning to Marketing Mix Modelling

The department store creator, John Wanamaker, famously griped: “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”

That quote is over 100 years old, and we’ve all nodded along to this quote at marketing conferences as a shared painful joke.

The reality is it’s no longer unsolvable. Marketing Mix Modelling is the modern scientific answer to Wanamaker’s dilemma and the future of marketing measurement that suitable Irish and EU businesses need to consider adopting as digital attribution deteriorates.

Previously I discussed why ad platform dashboards have a bias to over-report and why third-party attribution delivers a fairer view. However, even 3rd-party attribution has limits: Attribution is digital-only, tracking-dependent, and can’t account for offline channels (TV, radio etc.), seasonality, brand equity or macroeconomic factors driving revenues.

Marketing Mix Modelling answers not ‘which ad drove this click?’ but instead ‘across everything we spend, what is actually driving revenues and how can we allocate budget accordingly?’.

What Is Marketing Mix Modelling?

Marketing Mix Modelling (MMM) is a statistical technique that incorporates the analysis of historical business data, looking at revenues achieved in relation to ad spend, pricing changes, promotions, seasonality, economic conditions and competitor activities.

It delivers the view that Wanamaker sought by quantifying what is actually driving sales and by how much. The real value it delivers is separation. MMM is the only methodology that can reliably establish the difference between your base sales and incremental sales.

• Base sales are those that would have occurred anyway due to your brand equity, seasonal demand and baseline pricing.
• Incremental sales are those you achieve as a direct result of your marketing efforts.

It’s important to differentiate Marketing Mix Modelling (MMM) from Media Mix Modelling. Media Mix Modelling focuses strictly on budget allocation to paid media channels, while Marketing Mix Modelling assesses broader factors including pricing, distribution, competitor activity, product launches and strategy.

Marketing Mix Modelling isn’t a shiny new toy, but it is making a very real comeback and will only become more important as we move towards more privacy-focused advertising methods. It was originally developed in the 1980’s for fast-moving consumer goods (FMCG) like Coca-Cola, where large data volumes allowed for its use to plan budgets across print and TV.

MMM fell out of favour when digital tracking began to deliver cheap and easy last-click attribution, but the irony is that digital tracking is now breaking and has been degrading for the last few years, becoming structurally less reliable. As AI and cloud computing have made the data processing capabilities required for MMM much more accessible, MMM is returning as the best scientific method available.

If your business has always primarily budgeted based on Ad platform results and short term return on ad spend (ROAS) measures, then Marketing Mix Modelling will likely challenge assumptions that have grown from this.

What Types Of Businesses Should Use Marketing Mix Modelling?

Marketing Mix Modelling works best for B2C Brands with at least 2 years of consistent marketing spend data across multiple marketing channels to help build a model. The sectors that particularly benefit from MMM are eCommerce, retail, insurance services, financial services and FMCG. MMM is most relevant for businesses that run both offline and online marketing activities that need to understand how their combination affects revenues and drives lead generation.

Accurate Marketing Mix Modelling is more difficult but not impossible for:

• B2B brands where complex multi-year sales cycles make tracing back to source challenging; they find it harder to connect MMM to CRM-level deal attribution.
• Businesses that are in an early stage with limited marketing data available yet.
• Businesses that market on a single marketing channel or where multi-channel businesses’ marketing spend variation is too small to meaningfully model to results.

Why Is Marketing Mix Modelling Making A Comeback?

To understand how important Marketing Mix Modelling has become, you only need look at Gartner and their creation of their first-ever dedicated ‘Magic Quadrant‘ for Marketing Mix Modelling solutions in 2024, with the second edition in 2025. If you don’t speak ‘Gartner’, they only create Magic Quadrants for markets that have reached enterprise-level maturity with sufficient size and vendor competition, signalling MMM has gone mainstream.

If you’re feeling a touch of whiplash from the loss of reliable digital tracking over the last few years due to privacy measures, cookie consent and more, you’re not alone. A 2024 eMarketer survey found 53.5% of US marketers now use MMM specifically to overcome privacy-driven tracking limitations. The data shows we’re at an important inflexion point in the state of marketing and how it’s best measured.

Marketing channel fragmentation is greater than ever, and user journeys have never been so diverse. Your customers are interacting with your brand across traditional search, AI answer engines, social media, podcasts, influencer content, smart TVs, and Out-of-Home (OOH) advertising, and YouTube before becoming a customer, and no digital pixel-based tracking tool can see this full picture. The technology available has caught up though; MMM can account for all of these by using Bayesian statistical methods and machine learning, making it cheaper and much faster than the 6-month enterprise-level projects it used to require.

We always see US businesses adapting first to these changes, with Irish businesses slow to follow but eventually getting there. For Irish and EU businesses, third-party cookies are dying, and user-level tracking is degrading. GDPR and cookie consent requirements see consent rejection rates, resulting in large portions of user journeys becoming invisible to in-browser pixel-based tracking tools. Analytics and ad platforms respond by estimating to try to close the gaps. MMM works on privacy-safe aggregated data requiring no consent.

The paradox of the marketing industry globally currently is that 85% of marketers describe themselves as ‘extremely’ or ‘very’ confident in their holistic ROI measurement, according to Nielsen’s 7th Annual Marketing Report. That confidence is inspiring, but the report also shows only 32% of marketers globally are measuring across both digital and traditional channels and in Europe that drops to just 23%. That’s a 53-point gap between global marketers’ confidence (85%) and our capability (32%), effectively driving blindfolded while insisting we can see the road.

Adoption is catching up to confidence levels, though. Kantar reports that 51% of sophisticated US marketing teams are using MMM in combination with other tools; Emarketer surveys show that 46.9% of US marketing teams plan to increase MMM investment in the next year, and MMM is rated the most reliable measurement methodology by 27.6% of US marketers, ahead of Multi Touch Attribution at 19.4%.

All of which shows that the triangulated approach achieved by including Attribution, Incrementality and MMM is becoming the new standard, which we need to work towards. When marketing budgets are squeezed, and the tolerance for misallocating budgets based on flawed data is low, MMM can help guide the way.

What Are The Benefits Of Marketing Mix Modelling?

If you’re still unsure why you need to consider adopting Marketing Mix Modelling, below are the specific data-backed benefits MMM brings to the table:

No Tracking Or Consent Dependencies

MMM is structurally privacy-sound. All historical data is aggregated and has zero reliance on cookie-based tracking or in-browser pixel-based tracking and requires no consent for use.

Cross Channel Neutrality

MMM includes TV, radio, OOH, events, influencers, email, SEO / GEO, and digital ad spend in a single unbiased measurement model, which is impossible with any attribution tool. No ad platform reporting bias is introduced, as no channel has a ‘home advantage’.

Budget Scenario Planning

Your MMM model answers questions like “what if we shift 20% of Meta spend to YouTube?” before spending, not after.

Catches The ‘Brand Building’ Short & Long Term Sales That Digital Attribution Misses

A Nielsen study conducted for Google on the longer-term effects of awareness and brand building efforts found that a 1% increase in upper and mid-funnel brand awareness drives a 0.6% increase in long-term sales and a 0.4% increase in short-term sales. Typical ad platform attribution models often end 7 days post-click, and according to Analytics Partners’ ROI Genome data, two-thirds of advertising impact occurs after a week. A focus on click-based models with short attribution windows is undervaluing your brand-building and top-of-funnel activities, while MMM can measure the compound effect over weeks and months.

MMM Exposes The ‘Halo Effect’ & Synergies Across Channels

The ‘halo effect’ is simply the effect one advertising channel has on another. Your TV ad will boost your branded search and paid search efforts; your radio ad drives branded search again. Pixel-based tracking measures each channel entirely in isolation, so it fails to connect the cross-channel amplification entirely. MMM identifies it and helps guide teams to maximise the effect.

It Reveals The 60:40 Rule

This is one of the key points for any performance marketer reading this, and was recently highlighted in another post on why performance marketing fails. The Institute of Practitioners in Advertising, analysing 996 case studies from 700 brands across 83 sectors and 30+ years (Binet & Field’s The Long and the Short of It), found that the optimal budget split is 60% brand building and 40% activation. However, because short-term attribution optimises toward activation (the 40%), most of us are starving the 60% that drives compounding growth. MMM forces you to see the whole picture.

Reduces The Costs Of Measuring Incorrectly

Marketers estimate they waste an average of 26% of their budgets on ineffective channels, according to a survey of 1,000 marketing professionals by Rakuten Marketing. That’s €130,000 of a €500,000 budget being spent badly. Marketing Mix Modelling shines a light on wasted spend.

The 3 Layers Of Measurement: Attribution, Incrementality & Marketing Mix Modelling

The three core areas of marketing measurement are complementary tools, not competitors and using any in isolation means you’re missing the big picture.

Multi-Touch Digital Attribution

This is the tactical layer of measurement. The benefits of multi-touch digital attribution are that it’s near real time and can reveal behavioural data at the user level. It’s ideal for day-to-day paid media campaign optimisation by Meta or Google Ads experts, helps diagnose creative performance and dials into audiences that are ready to buy or sign up right now.

Limitations include the degradation of digital tracking affected by privacy changes, the bias introduced by the ad platforms being used as the single source of performance data, and naturally it doesn’t include the effect of your offline marketing channels.

Incrementality Testing

A quick reminder that incremental sales are those achieved by your marketing activities that wouldn’t have occurred otherwise. Measuring incrementality requires experimentation using testing methods like geo holdouts and conversion lift studies. Essentially, these are studies of audiences that were served ads (the ‘treatment’ group) or weren’t served ads (the holdout group). As the old data science saying goes, “correlation is not causality” and incrementality is the most direct way to understand if a channel is actually causing conversions, not just correlating with them.

One test is used to answer a single question. An example could be “Does running paid Search for products not in promotion increase sales of these products?” The answers from incrementality testing are used to help calibrate and validate MMM models. Both Meta and Google provide native conversion lift testing tools, which are a good start.

Marketing Mix Modelling

MMM is the strategic layer that sits above all your marketing activities, including offline and across your channels, while accounting for external factors. MMM is what should guide long term marketing planning and budget allocation; it’s not for real time campaign tweaks. MMM answers the questions that digital attribution is structurally unable to.

The Triangulation Approach: Using All Three

When used together, it’s multi-touch digital attribution that guides day-to-day optimisation, incrementality tests allow specific marketing channel investments to be validated, and Marketing Mix Modelling powers strategic direction and decisions. In reality, WARC reports that only 2% of marketers are using the combination of digital attribution, incrementality experimentation and MMM that is recommended by marketing measurement experts.

Meta & Google’s Free Open-Source MMM Tools

Advertising tech giants Google and Meta releasing free Marketing Mix Modelling tools should be an eye-opener for anyone sleeping on MMM and clinging to the hope that digital attribution will ever return to its former accuracy. Those days are gone and aren’t coming back.

I critically examined Google and Meta’s ad attribution methods previously, and their release of free MMM tools is no coincidence; it’s the signal that MMM has crossed from an academic methodology to mainstream marketing measurement available to teams of all sizes. Harvard Business Review announced the same, way back in March 2023, when they declared MMM as “a new gold standard for digital ad measurement”.

“Marketing mix models have a specific advantage: They’re able to produce dependable measurements – and insight – purely from natural variation in aggregate data, and don’t require user-level data.”

— Harvard Business Review, March 2023.

Google Meridian

Google Meridian was publicly released in February 2025 and is an open-source Bayesian MMM framework. Naturally, it has deep integration with Google products including Google Ads, YouTube and DV360. It’s feature-rich for running scientific experimentation with sophisticated calibration options, but this also makes it better suited to marketing teams with strong data science capabilities.

Meta Robyn

Meta Robyn is a free open-source MMM in R and Python, free on GitHub (MIT licence). It’s the most widely adopted open-source MMM tool currently available, and with good reason; it’s more accessible for most marketing teams. Its calibration against real-world lift tests is built in, and many of the common tests that marketing teams want to run can be automated.

Meridian Vs Robyn

Both are free, so that’s not where the cost lies. Meridian is a more scientifically rigorous platform which prioritises statistical accuracy over accessibility for marketing teams and requires more technically capable operators. Robyn is more widely adopted because it’s easier for small teams to get to grips with and it prioritises automation where possible.

What Are The Blockers to Adopting Meta Robyn Or Google Meridian?

The truth about integrating open-source models into your marketing measurement is that they’re not plug and play. For starters, your historical data needs to be reviewed, cleaned and transformed as needed for processing. Typically, your existing data infrastructure and analytical capabilities need to be reviewed which some businesses simply won’t have the in-house talent to complete. This often makes managed MMM platforms or agency partnerships a more practical entry point.

Managed Marketing Mix Modelling Platforms

The data available shows that the vast majority of businesses that want to adopt MMM don’t have the resources or talent to build it internally using open-source platforms. Only 26% of in-house marketing teams actually conduct their MMM in-house; most businesses rely on external expertise, which includes the options below.

Analytic Partners

Enterprise-level and consultancy-led MMM that draws from large datasets from over 1,000 brands, 50+ countries and hundreds of billions in spend. High-end and high investment.

Nielsen Marketing Mix Modelling

Nielsen’s MMM offering is built on datasets that include 60 countries and more than 3,000 brands, making it strong for benchmarking against industry categories.

Ekimetrics

An obvious choice for Irish and European advertisers, as they’re an EU-headquartered data science consultancy with strong GDPR sensitive MMM capabilities.

Northbeam & Triple Whale

Triple Whale and Northbeam are two platforms which are an excellent first step in moving away from trusting only Ad platform reporting and moving to non biassed 3rd party digital advertising attribution, which also offer MMM-adjacent capabilities. This lowers the MMM entry bar yet again for DTC brands that are already using their attribution tools.

How Can Businesses Get Started With Marketing Mix Modelling?

1. Your MMM model will be informed by not only your historical marketing data but also your incrementality tests, so get started with these. Running geo-holdouts (excluding geographic areas from ads) or Meta Conversion Lift Studies will create the causal data needed. It’s also the quickest way to understand the difference between ad platform reported numbers and the actual impact on your business. The more of this data you have, the more reliable your model will become.
2. Clean and centralise your historical marketing data. Break it down by week and spend per marketing channel and resulting revenues, with any pricing changes, promotions or key seasonal periods of demand included. This is an organisation-wide challenge as much as a technical one, and you’ll need a minimum of two years of clean data. The sooner you plan for this, the better.
3. Start exploring the free open-source MMM tools from Meta and Google. By reviewing their documentation, they’ll help you understand which questions MMM can and can’t answer, what data is needed and whether your business has the internal data science capabilities to adopt MMM.
4. If not possible using internal teams, consider a managed partnership or agency.

Start Planning Now

While ad platform reporting is still the most useful method for daily measurement and optimisation of campaign results, and if your digital conversion tracking is properly implemented, it’s around 75-80% accurate depending on your marketing consent rates, even higher if you’re using server-side tracking; however, it’s best to start investigating whether your business should adopt the best measurement methods possible to help understand where your marketing budget delivers real impact.

If you want to take the first step and start using 3rd party attribution tools or running conversion lift studies with a Google Ads or Meta Ads agency that understands them, let’s have that chat.