The Media Mix Mirage: Unveiling the Flaws in Attribution During Peak Shopping Seasons

By Chris Marine, Campfire Consulting

As the holiday season gets underway, brands are gearing up for what is traditionally the most lucrative time of the year. Marketing budgets swell, new campaigns are launched, and every advertiser hopes to capture a slice of the consumer's heightened willingness to spend. In this flurry of activity, many are turning to Media Mix Modeling (MMM) to measure the effectiveness of their strategies. But could this approach give marketers a false sense of success? We think so, and we're not alone.

In recent years, digital advertisers and platforms have rekindled their interest in MMM—a statistical analysis method that evaluates the performance of various media channels like TV, radio, print, and now digital and social giants such as Google, Meta, and Amazon. Over months-long campaigns, MMM attempts to attribute sales contributions to these channels, helping marketers decide where to allocate their budgets.

However, a surge of academic research is casting doubt on the reliability of MMM. Julian Runge, an assistant professor of marketing at Northwestern University, and William Grosso, CEO of Game Data Pros, recently highlighted significant flaws in the model in a column for Mobile Dev Memo. One of the most pressing issues they point out is the timing of new media tactics.

Consider a brand that decides to implement a new advertising strategy in late October, perhaps increasing spend on Connected TV (CTV) or social ads. As sales naturally spike during the holiday shopping season, this new tactic appears to yield impressive results. On paper, the strategy looks like a resounding success, but did it genuinely drive the sales uptick? Or was it merely riding the wave of seasonal consumer behavior?

"The core problem with MMM during peak seasons is that it often conflates correlation with causation," says Runge. "Sales are increasing anyway due to external factors like holidays or economic stimuli. Without proper controls, it's easy to misattribute that growth to new marketing efforts."

This misattribution isn't just a statistical oversight—it has real financial implications. We often see companies pour more money into channels or lower funnel tactics that aren't actually delivering incremental value, leading to inefficient budget allocations and missed opportunities elsewhere.

Adding another layer of complexity is the inherent bias within marketing teams. There's a natural tendency to seek out data that confirms the effectiveness of one's strategies. This confirmation bias can blind marketers to the broader context influencing consumer behavior.

"Marketers are under immense pressure to demonstrate ROI," Grosso explains. "The allure of attributing success to one's own efforts is strong, especially when using models that lack the granularity to separate out external factors."

So, what's the solution? At Campfire we suggest a multifaceted approach to attribution, especially during peak periods. This includes incorporating more granular data analysis, utilizing control groups in our measurement, and combining MMM with other measurement methods like Multi-Touch Attribution (MTA).

"Diversifying measurement approaches can mitigate the shortcomings of any single model," advises Runge. "It's about triangulating the true impact of marketing efforts by looking at the data from multiple angles."

As brands navigate the high stakes of the holiday season, it's crucial to remain vigilant about the tools and models relied on. Blindly trusting in MMM without accounting for its limitations may lead to overestimating the effectiveness of certain campaigns, ultimately affecting the bottom line, long-term relevance in culture and growth.

In an industry where every dollar counts, especially during the most competitive time of the year, a nuanced understanding of attribution models, a holistic approach to data activation and omnichannel strategy, isn't just beneficial—it's essential.

Stay informed. Stay critical. And most importantly, question the numbers before they become the narrative.