While COVID-19 is having a significant impact on how retailers currently operate, many experts believe these developments will result in a long-term shift in consumer behavior. In a recent interview with Retail Dive, Simeon Siegel, managing director at BMO Capital Markets, remarked “Right now it’s very unclear how long this impact will be — not only because we don’t know the duration of the virus but because there’s a latent fear that’s emerging as well. The longer the impact, the more lingering the fear and the more evolution there will be of consumer processes.”

What this means for retailers is that they will need to assess the future impact of changing consumer patterns on their brick and mortar sales.  The impact on future store performance will result in the need for a fresh approach to location modeling – one that can’t be adequately addressed through traditional modeling techniques. As a quick refresher, traditional models use historical observations to determine future performance – a rear-view mirror approach to forecasting. They are incapable of considering complex, non-linear relationships, or of updating rapidly to reflect new and refreshed data.

Let’s dig into this a little more, and discuss how machine learning-powered models will help retailers understand the new normal quicker, enabling them to react faster and recover sooner.

The New Normal … Maybe

While there is a lot of speculation on what the new normal will look like, no one knows for sure. But we do know that sales forecasting models built on last year’s data might not be predictive in a forward-looking post-COVID world. While traditional models may serve as a good benchmark, the previous relationship between an individual unit’s competitors and customers may no longer be accurate in the post-COVID world.

Some theorize that older populations, who have the disposable income, will be less eager to resume behaviors that were common before the pandemic, such as going to the movies, sitting in cafes or eating in restaurants due to lingering fears about the dangers of contracting the coronavirus.  On the other hand, younger populations may be quicker to resume activities, but are more likely to be unemployed or have lower incomes to support these behaviors. Further, it will likely take consumers a long time to consider going back to restaurants, movie theaters, crowded stores, or a public gym resulting in changes in annual spending. If/when these shifts in performance happen, applying an old model to a future site without reflecting the changed consumer shopping patterns or spending habits, will result in making ill-informed real estate decisions at best, and making impactful mistakes with Cap-Ex at worst.

In addition, there will be significant changes to the competitive environment, which will vary from location to location and trade area to trade area. Retailers that were struggling prior to the pandemic may go bankrupt resulting in an abundance of closed competitors.  All in all, retailers and restaurant companies will likely have to close units, or at least reposition, to accommodate higher levels of omnichannel activity. So, the competitive landscape for each unit may look very different when the dust settles.

The other change that will impact individual unit traffic, particularly for restaurants, is the possibility of a new, broader, work-from-home culture. During the pandemic, companies may realize that certain roles are well-suited to work from home, which will forever change the daytime population relative to certain store locations. In fact, some experts are predicting that the percentage of individuals who work from home one or more days a week may increase from 4% pre-pandemic to 25% in under the new normal. This shift will result in significant impacts on breakfast and lunch sales for restaurants or traditional retail that relies on a high density of worker-based traffic to drive sales (e.g., downtown locations in Chicago, NYC, San Francisco, etc.)

The limitations of traditional models – and the benefits of machine learning

Traditional models can’t adapt quickly, which is exacerbated in the current environment. Tango’s models, in contrast, are powered by machine learning and can rapidly recalibrate based on changes in underlying factors – something critical in a post-COVID world. Techniques of the past are very manual. In contrast, machine learning provides the ability to sift through old, and new, data sets to find the best combination of models to interpret the current environment. By using machine learning techniques, we can process large volumes of information and discern patterns that are beyond the capabilities of human analysis, especially when the underlying data is changing month to month. These techniques are tailor-made to address the rapidly changing impact of closed competition, or shifting consumer patterns and provide the operator with a forward-looking Sales Forecast that reflects a post-COVID environment.

The other advantage is that the calibration of machine learning fueled models takes a fraction of the time of a trained analyst as compared to more manual processes. This speed will be essential in the post-COVID world as retailers and restaurant companies scramble to adjust to the new normal. Additionally, as we go through the recovery period, which likely won’t be a straight line, Tango can help navigate the choppy waters. These models can be run repeatedly to reflect changes in consumer behavior so that decisions are being made with the most recent and accurate data available.

Tango is spending a significant amount of time and effort monitoring change; collecting, compiling, aggregating and sifting through new raw data sources to figure out what is relevant – something market research analysts within retail organizations do not have the time to do. This allows us, on a dynamic basis, to recalibrate models as we go. Tango’s models are able to quickly let retailers know which areas are recovering and which are not, which competitors are thriving in specific locations, and which ones are closing.

This ability to quickly and accurately predict the new normal is as relevant to retailers who face losses as well as increased sales as a result of COVID-19. For some retailers, the question is, “How long will it take for my sales to ramp up and get back to normal?”, and for others, like grocery, the question is,  “How long will it take me to get back to normal when my comp sales start decreasing?”

If you are a retailer that is currently using a traditional location model, get in touch and we can show you how Tango can help prepare you for the post-COVID world and beyond.

Also, check out our COVID-19 Retail Location Benchmark tool here.