What Have Marketers Learned During The Pandemic And What Role Can Analytics Play In Forecasting The ‘Next Normal’
Written by Eric Bradlow, GBK Co-Founder and Vice Dean for Analytics at the Wharton School
2020 has been a year of incredible change with the global pandemic impacting nearly every facet of life and business. As part of The Wharton School’s Fast Forward: COVID-19 video series, I recently answered a series of questions on what the future may hold for business and sports through the lens of analytics. I’ve also had dozens of conversations with colleagues, clients and students on the evolving challenges leaders and organizations face during the pandemic, and the role analytics can play in shaping decisions.
Based on those conversations, along with my own experience as Vice Dean of Analytics at Wharton, here are a few key takeaways and observations:
Analytics as a decision-making tool
During periods of crisis and uncertainty, analytics act as a critical decision-making support tool. Already we’ve seen statistical forecasts play a vital role in understanding the impact of COVID-19 to better inform health and public policy decisions. As leaders consider when and how to re-open our economy, decisions have to be made with some amount of ambiguity, while still taking the best available data into account.
Among the many variables data scientists and epidemiologists have to consider: How do factors such as population density, travel patterns and other behaviors impact the rate of infection? As new data is interpreted, how much weight should be given to contributing factors such as age, pre-existing health conditions, among other variables? These questions are difficult to answer as often the data is not collected or measured in a consistent way.
“During periods of uncertainty, analytics act as a critical decision-making support tool.”
As businesses apply analytics, they are also dealing with high levels of ambiguity and less than perfect data, and the current health crisis has added even more uncertainty. Given the rate of change, companies can’t afford to base their analytical models on outdated or historical data. At the same time, they have to continue to make key decisions based on the best available data they have, while constantly seeking out better data.
Changing customer behaviors
Every company regardless of industry has needed to pivot and adapt their approach during the current health crisis, and both primary research and analytics play a critical part in understanding what the next normal will be. Already we have seen many customer preferences and buying patterns shift. For example, my family never used to use a grocery delivery service for our weekly shopping. Now we shop that way all the time and will likely continue to do so even after the pandemic has ended. As all of us marketers know, “state dependence” (inertia) is a strong force and now that our previous choices are delivery services, we may stay that way.
Many online and home delivery businesses such as Amazon, Instacart and others have boomed this year as people sheltered in place from home. And many retail brands and experts believe these shifts in behavior will become permanent, as consumers continue to trial and adopt online services ahead of traditional retail.
“As all of us marketers know, “state dependence” (inertia) is a strong force and now that our previous choices are delivery services, we may stay that way.”
In response, companies across categories are making moves to bolster their online experience to drive stronger engagement with customers. With Covid-19 cases still surging around the country, we’re also seeing companies rethink their retail strategy. Late last month, Microsoft permanently closed its brick and mortar stores, a growing trend.
Increase experimentation during times of uncertainty
In a time when so much rapid change is happening, companies need to experiment more, not less, to improve ROI for their analytics efforts. The first step before you apply data or define your approach is to clearly define the business problem you are solving for. Once you have clearly defined the outcomes you are attempting to predict, next ask what assumptions are reasonable to make based on the data and customer insights that we have? What assumptions can support a forecasting model?
“Companies need to experiment more, not less, to improve ROI for their analytics efforts.”
Predictive analytics becomes particularly effective when you begin to infer customer needs, and are able to distinguish between and isolate the contribution of individual factors, through randomized controlled experiments. By applying better, real-time data (including behavioral, transactional, demographic, and more), supplemented with external data like surveys, firms are able to separate the signal from noise.
Stay focused on your most valuable customers
We all know the classic marketing statistic that 80% of your revenue comes from 20% of your customers. Now the question is which 20%? Even as states reopen, I find it difficult to believe that the number of unique customers that businesses have will go back to the level that it was before the pandemic.
“80% of your revenue comes from 20% of your customers. Now the question is which 20%?”
Brands need to apply analytics to better understand who their most valuable customers are right now, and just as importantly, who they are not. As the economy reopens, many companies are in a rush to acquire new customers, or re-acquire new customers, when the focus should be on their most valuable customers. Other businesses have offered deep pricing discounts in recent months, or chased other opportunities down market. While this may help to drive incremental sales, it has potential to negatively impact brand equity in the long run.
Maximize CLV by encouraging binge consumption
Many brands also fail to maximize Customer Lifetime Value (CLV) to the extent they could, and during times of crisis or market turbulence, there can be a tendency to compound the problem. There are several challenges that brands face as it relates to how they measure CLV. One, the mathematical model or framework called RFM (recency, frequency and monetary value) that most firms use to calculate CLV is fundamentally wrong! Through my research at Wharton, we have demonstrated another crucial dimension of buying behavior that is highly predictive of CLV – binge consumption or what I call “clumpiness”. Customers who consume or buy content in bunches, then go away and come back, and buy in bunches, are more valuable than customers who buy at a steady pace.
“Many brands also fail to maximize CLV to the extent they could, and during times of crisis or market turbulence, there can be a tendency to compound the problem.”
Second, the profile of your most valuable customers is dynamic and constantly evolving. During the pandemic, with consumers predominantly buying products and services online, many brands are seeing binge purchase behavior from customers on an even more frequent basis. As customer behavior changes, brands need to revisit their segmentation to ensure their marketing, product and brand strategy are 100% focused on fueling growth with their most valuable customers (MVPs).
Another key to success here is focusing on better data, not big data. Now is the time to throw out data sets or insights that are no longer valid and make sure you are building actionable strategies around the best available data on current customer behavior. As I discussed in my previous article, brands and marketers that hyper-focus on their MVPs and track those customer’s clumpiness over time, are far more successful in predicting and encouraging future binge purchases.
Analytics a recession proof opportunity
In many ways, the Covid-19 crisis has acted as the largest scale social science experiment of our time. Never before has the modern global economy been shut down and reopened, nor have we had to manage through change with so little certainty about the future.
“In many ways, the Covid-19 crisis has acted as the largest scale social science experiment of our time.”
As the global economy reopens, companies are taking varying approaches to weather the storm, but one constant is the demand for better analytics. As I recently shared with Ramona Schindelheim on the Working Nation podcast, I believe analytics is one of the jobs that is truly “recession proof”.
Companies and leaders around the world need to learn how to ingest data, analyze data and make business decisions based on those data. In fact, in my role as Vice Dean of Analytics at Wharton, we’re hearing from companies that there is a shortage of people who can do applied data science and analytics effectively, and that demand will only continue to grow.
