The marketing industry is undergoing a major transformation. Privacy regulations are becoming stricter, user tracking is being limited, and traditional attribution models are rapidly losing their effectiveness. In this new reality, companies are seeking smarter ways to allocate marketing budgets effectively without relying on user-level data.
According to the IAB State of Data 2024 report, 95% of advertising and data executives expect further signal loss — leading to increased limitations on personalization — and over 80% of companies have already been forced to fundamentally restructure their organizational processes and measurement strategies.
As a result, there’s a growing demand for approaches that leverage aggregated data while complying with new privacy requirements. Marketing Mix Modeling (MMM) has returned to the spotlight as one of the most effective tools for optimizing budgets and accurately measuring the impact of each channel.
We spoke with Stanislav Petrov, Senior Data Scientist at Capital.com, who has over a decade of experience in analytics, to explore how MMM enables companies to build sustainable marketing strategies, make precise decisions, and unlock new growth opportunities — even when traditional tracking tools no longer work.
How has the approach to measuring marketing effectiveness evolved, and what are the key challenges?
In recent years, marketing analytics has undergone a profound shift, primarily driven by stricter privacy regulations and growing restrictions on data collection. Access to user-level signals has drastically declined. In the EU, the enforcement of GDPR and, more recently, the Digital Markets Act (2024) introduced major constraints on how companies can process, combine, and activate personal data — particularly within the ecosystems of dominant platforms. In the U.S., state-level legislation like the California Privacy Rights Act (CPRA) has significantly limited cross-site tracking.
Tech platforms have responded accordingly. Apple’s App Tracking Transparency framework requires apps to request explicit permission for tracking, which most users now deny. Google, meanwhile, is phasing out third-party cookies in Chrome and advancing its Privacy Sandbox, shifting from user-level identifiers to more aggregated, cohort-based targeting approaches.
Previously, attribution was relatively straightforward: cookies allowed marketers to track user paths, measure conversions, and link outcomes to specific channels. Today, such direct tracking is no longer possible in most regulated markets. This has accelerated the shift toward privacy-first, aggregated measurement methods — including MMM and incrementality testing.
Interestingly, while the methods have changed, the core challenge remains the same: marketers still need to allocate budgets effectively and understand which channels are truly driving results. What has changed is the degree of uncertainty. Without user-level data, it has become significantly harder to evaluate the incremental contribution of individual channels, campaigns, or tactics. In this new context, durable approaches like MMM are becoming critical — providing a stable foundation for investment decisions when traditional tools fall short.
How does MMM differ from traditional attribution, and why is it becoming an increasingly valuable tool?
Traditional attribution models are facing growing challenges in today’s fragmented and heavily regulated data environment. They often produce inconsistent or conflicting results, especially when they rely on user-level tracking, which is now restricted by privacy legislation.
MMM takes a different approach. It doesn’t track individual user actions and doesn’t rely on clicks or cookies. Instead, MMM analyzes aggregated data, evaluating the overall contribution of each marketing channel to key business outcomes — such as revenue growth or brand awareness.
The methodology accounts for seasonality, carry-over effects (adstock), and diminishing returns, helping identify the optimal investment level for each channel. MMM provides a holistic view of the entire marketing mix, rather than a narrow focus on individual actions or creatives.
As privacy regulations tighten and access to personal data declines, interest in MMM continues to grow. It’s increasingly seen as a reliable foundation for strategic planning and budget allocation — especially when traditional analytics tools no longer deliver actionable insights.
What tools and frameworks are used to build MMM models?
MMM is based on linear regression, adapted to reflect marketing-specific dynamics. While companies can build custom models internally, many rely on open-source frameworks developed by major platforms and research communities. The most popular include Robyn (by Meta), PyMC-Marketing, and Lightweight-MMM (by Google). Each offers different levels of customization, scalability, and statistical sophistication.
How does MMM account for seasonality, lag effects, and channel interactions?
Frameworks approach these factors differently. Robyn, for instance, uses Prophet to decompose time series into trend and seasonality before feeding them into regression. PyMC applies Fourier transformations to account for seasonal fluctuations. Lag effects are captured using adstock transformations, which reflect how the advertising impact is distributed over time rather than being immediate.
One limitation of traditional MMM is its handling of nonlinear interactions. Linear models often struggle to accurately reflect how marketing channels amplify each other’s effects. This remains an area for further innovation and hybrid modeling.
Is it possible to automate budget allocation using MMM?
In theory, MMM can be integrated into automated systems. In practice, this is not always advisable. MMM is best used as a diagnostic and strategic tool, not a fully automated decision engine. Combining MMM with other methods — such as multi-touch attribution or Markov chain modeling — often provides a more balanced and reliable view. Using multiple lenses reduces the risk of over-optimization based on a single framework. In some cases, MMM yielded inconsistent results due to its probabilistic nature and poorly defined inputs. If it had been automated, it could have turned into a big waste of money. Automation at the individual user level is still more sensible and feasible. For example, this would be useful when using value-based or real-time bidding.
What does the future of marketing analytics look like under increasing restrictions?
Marketing analytics is shifting toward hybrid models: aggregate data, causal inference, and privacy-safe experimentation are becoming the new standard. The value of first-party data will continue to rise, especially as companies learn how to responsibly activate and leverage it.
At the same time, the ecosystem is maturing: tools are becoming more accessible, automation is evolving, and AI is lowering barriers to entry. These trends are accelerating the adoption of even advanced methods like MMM and helping organizations move away from individual-level tracking toward more sustainable and compliant measurement strategies.
In the long run, marketing analytics will become more holistic, transparent, and focused on long-term effectiveness — with trust, privacy, and sustainable data sources playing a central role.