A recap of the MSI webinar
As global markets continue to face uncertainty from the impact of tariffs, inflation, and supply chain disruptions, the panel explored a pressing question: Can marketing mix models (MMM) help companies mitigate risk from tariffs and similar economic shocks?
The Challenge: Modeling the Unknown
Scott McDonald opened the webinar by framing the challenge: marketing mix models traditionally incorporate variables within a brand’s control-such as pricing, promotion, and advertising-as well as external factors like competitive activity and economic shocks in financial markets. While many models have absorbed the effects of past shocks such as the 2008 financial crisis, tariffs from previous administrations, and the COVID-19 pandemic, the question remains whether these models can adequately predict or help companies prepare for unprecedented economic impacts like current tariff disruptions.
Koen Pauwels explained the fundamental difficulty with relying on historical data to predict future shocks, referencing the “Lucas critique,” which warns that models based on past data may fail when market expectations change. He shared an example from his consultancy work with a Japanese car manufacturer during the Toyota recall, where sales unexpectedly increased due to a competitor’s troubles. This required updating the model not only to adjust production forecasts but also to reassess advertising effectiveness, considering possible negative halo effects on Japanese cars. Koen also emphasized the importance of distinguishing between short-term tactical adjustments and long-term structural changes in model relationships.
Jason McNellis reflected on the lessons learned from COVID-19, higher tariffs, supply chain disruptions, and inflation. He noted that while no prior event exactly matches the current tariff environment, companies can leverage analogues such as port congestion and inflationary periods to inform their models. Jason stressed that instead of treating tariff announcements as isolated events, companies should identify correlates-like consumer confidence, exchange rates, or inflation-and incorporate these into scenario planning to anticipate demand shifts. He expressed confidence that, despite the uniqueness of the current tariff policy situation, there is a rich history of shocks that marketing systems in other countries have successfully modeled.
The Limits of Prediction-and the Power of Causal Inference
Henry Innis, CEO and Co-founder of Mutinex, took a more cautious stance, expressing skepticism about the predictive power of MMMs in chaotic economic times. He argued that treating models as “crystal balls” risks misleading decision-makers. Instead, he advocated for focusing on causal inference-rapidly updating models to understand how business performance responds to shocks rather than trying to predict exact outcomes. In Henry’s own words:
I’m dubious about presenting MMMs as crystal balls. What’s imperative is to have a very strong model structure that’s causally sound, that you can then update quickly and start to see how your business is responding.
Innis also highlighted three key truths companies should build into their models: price points will change both competitively and for consumers; consumer confidence disruptions will have cascading effects on discretionary spending; and tariff announcements likely represent a regime change that shifts model coefficients significantly. He urged marketers to prioritize causal modeling over elaborate scenario planning.
Baseline Effects and Structural Shifts
Ryan Dew offered a complementary perspective, suggesting that while some aspects of shocks-such as baseline effects driven by consumer sentiment-can be reasonably forecasted using historical data on proposed tariffs, other structural shifts are more difficult to anticipate. He noted that if consumer sentiment declines during an economic downturn, models that incorporate this baseline effect may still provide useful projections. However, regime changes that alter fundamental relationships, such as shifts in price elasticity, present a greater challenge for marketing mix models. Ryan also stressed the importance of distinguishing between these types of changes to understand the limits and opportunities of MMMs going forward.
Key Takeaways
• Marketing mix models retain value but require adaptation. Historical analogues can inform models, but companies must be cautious about overreliance on past data when facing novel shocks.
• Causal inference is more valuable than prediction in volatile times. As Henry Innis stressed, “Doing causal inference quickly… is a far more valuable technique to be using in these sorts of times.”
• Agility and continuous model updates are essential. Companies must rapidly reassess assumptions and update models to reflect changing market dynamics.
In conclusion, the webinar underscored that while marketing mix models remain a vital tool for navigating economic shocks, their effectiveness depends on combining historical insights with agile, causally sound modeling approaches. The future belongs to those who can learn, pivot, and model change faster than the shocks themselves.
Watch the webinar here: