5 Ways to Adjust Marketing Mix Models for Unexpected Events

Marketing mix models (MMMs), are a time-tested method for measuring the impact of your marketing and developing more well-informed strategic plans. However, in order to continue to access useful planning insights, businesses must adjust their models to account for the impact COVID-19 has had. Most existing MMMs have been built without specific consideration for dealing with unexpected events, meaning they may not be able to accurately account for many of the changes businesses are going through. By updating your models now, you can better understand the pandemic’s impact and get useful insights for your strategic planning.

How unexpected events impact marketing mix models

Marketing mix models amalgamate historical data into patterns and trends that can then be used to make predictions about channel performance. This is usually an essential practice for many leading brands, however, the COVID-19 outbreak has changed consumer and business behaviors and disrupted marketing across the globe. Even looking back on relatively recent data often can’t fully account for this impact.

To make matters worse, the impact differs greatly across the world. Consumer behaviors vary by country, region, etc., and in turn, policies and restrictions vary as well. Even when looking at 2 neighboring cities you can see extremely large differences in terms of impacts. Most MMMs just aren’t granular enough to capture these local differences.

Many businesses have significantly changed their creative & copy in response to the current climate, using new messaging and tones to appropriately connect with their audiences. Creative impact tends to be poorly measured by Traditional MMMs, and so it’s difficult to assess the effects of creative changes without adapting the models.

Accounting for disruptive events in MMMs

I have seen many “Marketing Guru’s” lately that have been preaching that MMMs are a thing of the past and that COVID is proof of that, I believe that rather than abandon MMMs altogether, we should be seeking to adapt our models to account for these disruptive events. By following the five key steps outlined below, businesses can update their models and continue to access useful insights to help aid in strategic planning.

1. Account for changes in consumer behavior.

Running MMMs more frequently allows you to include more recent data, which helps ensure your businesses output reflects the rapidly shifting digital environments. As marketers, we should be considering the long-lasting implications of consumer behavior; for example, e-commerce sales may remain higher than pre-pandemic conditions if there is a more permanent shift in how consumers are purchasing. Businesses can use recency variables, which allow models to calculate when a purchase was made, and custom adstocks for ads according to the target audience, demographic, media channel, and format, but also optimization goals, target audience, and other campaign-specific characteristics.

2. Include more granular data.

By using more granular, precise data, such as data focused on specific geographies will result in much more targeted insights, especially in unusual times. Marketing solutions must focus on how COVID-19 has affected specific markets and how that correlates with the brand’s category and vertical. By using granular data, such as daily, campaign-level, geo-region, direct marketing area (DMA) level data, businesses can better understand the impact of evolving consumer behavior. When modeling by geography to account for COVID-19, it is important to keep in mind that the timing of both impact and response has varied.

3. Consider how creative quality and media type impact performance.

Ad campaign quality can drastically vary, meaning not all impressions are of equal value. However, as mentioned before, typically MMMs don’t account for creative quality and media type. This issue is especially important right now, as most businesses have shifted their creative strategies such as updating copy and using different imagery in response to COVID-19. By comparing campaign performance according to adherence to best practices, businesses can better contextualize the value of impressions and incorporate these assigned values into their MMMs.

4. Calibrate and validate MMMs.

Constantly running experiments helps businesses check their assumptions, choose the correct models, test against known outcomes, and tweak their MMMs in order to make them more accurate. When there is increased scrutiny on marketing budgets or less relevant historical data to reference, calibrating marketing mix models with experiments is not only an effective but essential way to ensure models are as accurate as possible. I am always encouraging marketers and business owners to look at their marketing efforts holistically. This includes looking at all your marketing activities and how they work as a whole instead of separate entities. This will really help you maximize the effectiveness of your experiments and ensure that your strategy is solid moving forward especially during uncertain times.

5. Adjust optimization and scenario planning.

It is important to remember that current audience behavioral patterns may not indicate long-term trends. Businesses should evaluate findings from high-impact periods of the pandemic to determine if these should be included in planning. If there is one key thing to remember about optimizations during unusual times like COVID, it is that there is no “one” solution. All businesses are going to operate under a different set of assumptions and may see very different findings when doing experiments.

Key takeaways:

While marketing mix models are a time-tested method for measuring the impact of your marketing and developing more well-informed strategic plans, they may need to be adjusted to account for disruptions and unexpected events, such as the COVID-19 pandemic. Accounting for changes in consumer behavior, including more granular data, calibrating marketing mix models with experiments, and adjusting scenario planning is key to adjusting models to account for the pandemic’s impact.