Model Mayhem: Analytics in the Aftermath of COVID-19

COVID-19 has accelerated the marketing model evaluation process, but regular review is something brands should have been doing all along.

The COVID-19 pandemic has turned marketing models upside down, forcing brands to respond to rapidly changing customer behaviors and motivations. Which of these new behaviors and motivations will last? What happens if the virus spikes in the fall? Will your traditional customers ever come back? Will your new customers stick around when the pandemic is over?

The only thing we know for sure is this: The old models don’t apply anymore. Insights that worked in the past will no longer accurately predict the future!

As you may know, advanced analytics algorithms are trained to predict future behavior based on patterns from the past. Such machine learning models built before the pandemic will continue analyzing data received during the pandemic, and that will likely reduce their predictive capabilities.

Garry Rosenfeldt, Principal, Analytics and Business Intelligence for Allant Group, says that pandemic-induced changes have been a wake-up call for many marketers. “Whether we’re in a pandemic or not, brands should always be monitoring customer flow and adjusting their analytics,” he says. “It’s always better if you have a process in place to see and respond to trends as they occur. Many brands are unfortunately just now waking up to this reality due to the COVID-19 pandemic.”

As shown in the chart below, there is ample evidence to verify the assumption that consumers are shopping in new ways and visiting competitor brands not previously in their consideration set.

His colleague Jimmy Yi—Allant Group’s Principal of Strategic Consulting and Analytic Strategy—says it’s more important than ever for companies to do extra reporting and tracking. “Marketers have to know if their traditional customers are engaging in different channels, and whether new customers can form sustainable customer segments,” he says.

The identification of new consumer behaviors could very well lead to new business strategies.

The identification of new consumer behaviors could very well lead to new business strategies, which are likely to require adjustments to existing models or the creation of new models altogether.

One Allant Group client, for example—a leading women’s retailer—has reached a younger, more rural demographic since the pandemic hit. In order to respond more quickly to this new audience, and to changing patterns from traditional customers, the marketing department has been running monthly reports comparing current customer geographies, demographics, and transaction volume to the same time frame a year ago.

“What many companies are scrambling to do today should really become a habit after the pandemic uncertainty ends,” asserts Rosenfeldt. “Models may not need to be checked every month, but they should be evaluated on a consistent, perhaps quarterly, basis.”

In the meantime, Yi recommends running two sets of models, one with your COVID customers and one with your non-COVID customers. “The differences among each customer set will determine how much effort you put into promoting different products, channels, and offers,” he says.

“The goal is to determine which models are working, and which models need to be monitored and adjusted,” Yi says. “That exercise begins with an understanding of the data points coming into a model, how those data points are weighted, and what assumptions were made during the model’s development.”

According to a McKinsey report on leadership’s role in fixing analytics broken by COVID-19, existing models may need “new, fresher inputs to provide more accurate insights in this volatile period. …We find that much data exists that can help organizations better understand current trends and customer needs. Business leaders charged with sourcing this data—whether a business head, chief data officer, or other executive—should expand their sources…”

Rosenfeldt says companies already have many good first-party data sources at their disposal, such as data from transactions, website interactions, and mobile app engagement. “Many of our clients are relying more heavily on digital sources, and, by doing so, are strengthening their analytics capabilities,” he says.

External data can also help get analytics models back on track.  For example, public health data, from external sources, in addition to a rich set of consumer demographic, interest and behavior data, combined with a Brand’s first-party data, can and should be used to further augment the model’s strength.

Marketers, analytics professionals, and business leaders must work together to fix broken models and develop new models.

Marketers, analytics professionals, and business leaders must work together to fix broken models and develop new models that show consistent patterns month over month. Data influenced by COVID-19 will need to be weighted, or possibly set aside, when creating the new models of the future.

As consumer habits change due to COVID-19 concerns, data interest and behavior data will also change. Because consumers are spending less on travel, sports and entertainment, while spending more through e-commerce and streaming services, this data will become the new set of predictive data. 

The more data you have, and the more sophisticated tools you use to analyze your data, the more effective your models will be. “Modeling and analytics are designed to measure how people are engaging with your brand and what motivates them to do so, so you can capitalize on the customers that show the highest potential for revenue growth,” says Rosenfeldt. “Brands can never do too much of that.”

Yi agrees, adding that marketers who commit to detecting market changes—and responding to those changes quickly— will have the greatest chance for sustainable growth. “Getting through the current crisis requires great analytics agility,” he says. “Continuing this practice after the pandemic goes away will make marketers much more prepared for the next upending surprise.”

Allant Group offers a depth and breadth of analytics and data expertise that enable brands to discover nuances in customer behaviors that maximize profitability. To learn more about normalizing your analytics in a post-pandemic world, contact us online or call 800-367-7311.