Prove Ad Spend Value Post-Cookie With MMM

©PA

Cookies are gone, but attribution isn't. Learn how Marketing Mix Modelling (MMM) provides a privacy-safe way to measure your ad performance.

'Marketing Mix Modelling' (MMM): The 'Cookieless' Way to Prove Your Ads Are Actually Working

For the last decade, digital marketers have lived by a single, addictive metric: the last-click. We built entire careers on Multi-Touch Attribution (MTA) models that told us, with precise-looking dashboards, exactly which Facebook ad or Google search keyword drove a sale. Now, thanks to privacy updates and the "cookiepocalypse," that world is gone.

Most marketers are now flying blind. Our attribution models are broken, and our ROAS (Return on Ad Spend) figures are unreliable guesses. How can you confidently ask for a budget when you cannot prove what is working?

The answer, ironically, comes from a method that predates the internet itself: Marketing Mix Modelling (MMM). This "old-school" statistical approach is making a massive comeback, giving brands a holistic, privacy-safe way to measure the true impact of their marketing.

What Is Marketing Mix Modelling (And Why Is It Back?)

At its core, Marketing Mix Modelling is a "top-down" statistical analysis that uses aggregated data to quantify the impact of various marketing and non-marketing drivers on a specific outcome, usually sales or revenue.

Instead of tracking individual users with cookies (a "bottom-up" approach), MMM looks at the big picture. It takes months or years of data—such as weekly ad spend, sales figures, competitor activity, and even the weather—and finds correlations. It answers the big questions: "For every £1 I spent on TV ads, how many pounds did I get back?" and "What would happen to my sales if I cut my search budget by 20%?"

This macro-level view is precisely why MMM is the perfect solution for the modern, privacy-first era.

How Is MMM "Cookieless"?

MMM is "cookieless" by design because it never needed cookies in the first place. It does not track individuals, only "inputs" and "outputs."

Here is the simple logic:

  • MTA (Cookie-Based): "This specific user, 'User 123,' clicked Ad A, then Ad B, and then bought."
  • MMM (Cookieless): "In the week we spent £50k on TV, sales were £200k. In the week we spent £100k, sales were £300k. The model calculates that the £50k increase in spend contributed to a £100k lift."

Because it relies on aggregated, anonymous data, MMM is completely unaffected by an individual's privacy settings, ad blockers, or browser changes.

MMM vs. Multi-Touch Attribution (MTA)

For many digital-native marketers, MMM can feel like a step backwards. It is slower and less granular than the real-time MTA dashboards we loved. However, this is because they are designed to answer different questions.

Relying solely on last-click MTA was always a high-stakes bet. It felt precise, but it was a gamble that ignored the complex journey of a real customer, like betting all your chips on a single number at https://fortunica-online.com/en-gb. The momentary thrill of seeing a "direct" conversion often overshadowed the fact that 99% of the customer's journey (brand recall, TV ads, word-of-mouth) was invisible. In the post-cookie world, that last-click gamble is a guaranteed loss; MMM is the strategic, odds-based approach to managing your entire portfolio.

This table breaks down the fundamental differences between the two methods:

Feature

Multi-Touch Attribution (MTA)

Marketing Mix Modelling (MMM)

Data Level

User-Level (clicks, impressions)

Aggregate (weekly spend, sales)

Data Source

Cookies, pixels, user IDs

Ad logs, sales reports, external data

Privacy

Dependent on user tracking (failing)

Privacy-safe by design

Scope*

Digital channels only

All channels (TV, radio, digital, print)

External Factors

Ignores them (e.g., seasonality, price)

Includes them (e.g., seasonality, economy)

Key Question

"Which ad creative got the last click?"

"What is the total ROAS of my YouTube budget?"

As the table shows, MTA was good for in-platform, micro-optimisations. MMM is built for strategic, high-level budget allocation—exactly what CMOs and CFOs need to see.

The Key Components of a Modern MMM

This is not your grandfather's MMM, which took six months and a team of statisticians to build. Modern MMM is faster, more agile, and powered by open-source tools (like Meta's "Robyn" or Google's "Meridian").

Any modern model has three core components.

Aggregated Data Inputs

The quality of your model depends entirely on the quality of your data. You need to gather:

  • Marketing Data: Weekly or daily spend and impression data for every channel (e.g., Google Ads, Facebook, TV, radio).
  • Conversion Data: Total sales, revenue, or sign-ups per day or week.
  • External Factors: Data on seasonality (e.g., Christmas), holidays, competitor spending, and economic indicators (e.g., inflation).
  • Non-Media Drivers: Information on your own price changes, promotions, product launches, or even website downtime.

The Statistical Model

This is the "engine" that finds the patterns. The model (often a form of Bayesian regression) sifts through all the data and isolates the impact of each individual driver.

It learns, for example, that a price promotion creates a huge, immediate sales spike, while TV advertising creates a smaller, but much longer-lasting, lift in "base" sales. It also accounts for the "lag effect" (e.g., a TV ad seen today might not cause a purchase for two weeks).

The Outputs: What You Actually Learn

This is where the magic happens. A good MMM does not just give you a single "ROAS" number. It gives you a set of actionable, strategic tools.

  • Contribution Breakdowns: It shows you a chart of your total sales, with "slices" showing what percentage was driven by your brand, seasonality, and each specific marketing channel.
  • Spend vs. Response Curves: This is the most valuable part. The model shows you the "point of diminishing returns" for each channel. It answers: "At what point does spending more on Facebook stop being profitable?"
  • Budget Optimisation: The model can run simulations, answering: "If I have a £500k budget, what is the optimal way to split it across all channels to maximise revenue?"

Your Path to Cookieless Attribution

The shift away from user-level tracking is forcing marketers to be smarter, more strategic, and more holistic. The era of easy, last-click attribution is over, and it is not coming back.

Marketing Mix Modelling is the only proven, privacy-compliant methodology that can measure the entire business, from TV ads and pricing to Facebook campaigns and competitor noise. It allows you to prove your ads are actually working, without tracking a single user.

Your challenge is to get started. You do not need a perfect, all-encompassing model tomorrow. Your task is to start the conversation. Talk to your analytics team. Audit your data. Can you pull your weekly spend and sales data into a single spreadsheet? That is the first step. By starting now, you will be building a future-proof attribution model while your competitors are still grieving the death of the cookie.