For years, retailers and commercial property developers have relied on trade area analysis to help them better understand where people live, work, and shop in relation to commercial businesses. This involves both knowing where customers are traveling from and calculating the average drive times for getting from point A to point B.
The Huff Model takes this a step further by using this information—combined with data about store size and distance from shoppers—to predict the likelihood of customers visiting a given store (over another in the same area). This is a way to assess a store’s relative “attractiveness.”
While this can help estimate general foot traffic to any given store on any given day, where the traditional Huff Model falls short is in its inability to break these predictions down to dayparts.
In other words, you can use the static Huff Model to see whether a grocery store, for example, gets more visits on weekends than on weekdays. It can’t tell you, however, at what times those different spikes in foot traffic may occur. This is a critical piece of missing information and, without it, we can’t actually paint a full picture of retail foot traffic that explains at what point, during the day or night, people visit different retail trade areas with greater or less frequency.
Convinced that the Huff Model could take our knowledge about foot traffic patterns a step further, researchers at the University of Wisconsin-Madison and the University of Virginia, in a paper entitled, “Calibrating the Dynamic Huff Model for Business Analysis Using Location Big Data,” explored how weaving ‘time-awareness’ into the traditional Huff Model could give retail trade analysis a whole new level of precision in its predictions.
To do this, they created a time-aware dynamic Huff Model (“T-Huff” for short), powered by SafeGraph Patterns foot traffic data and U.S. Census Block Group data, to more accurately calculate and predict store visit patterns to supermarkets and department stores in the 10 largest U.S. cities.
The end result is a much more accurate and predictive Huff Model that uses ‘time’—more precisely, the exact hours in which people visit certain stores more frequently—to assess the anticipated market share for different kinds of businesses over time.
The benefit here is simple: the knowledge gained by making the Huff Model time-aware allows commercial developers to build better cities that address the unique needs of its local residents.
The T-Huff Model, therefore, helps us understand customer behavior with an entirely new level of specificity while also shedding light on how visits to different retail trade areas may vary significantly during different day-parts. And when this information is calibrated against the relative proximity between consumers and stores or other demographic and socioeconomic factors, like gender or median household income, we can begin to understand what truly impacts a customer’s choice to visit one store over another—at different times of the day.
This is incredibly useful for setting business hours, allocating transportation resources, and implementing the right infrastructures to improve the accessibility and success of any given retail trade area during its traffic peaks and valleys.
What’s clear in this paper is that SafeGraph data has the power to fuel retail analytics in ways that not only expand our knowledge of retail trade analysis but also can begin to explain consumer behavior (around shopping patterns) in a whole new light.
To learn more about how SafeGraph enabled this research, watch Professor Song Gao’s webinar on Calibrating the Huff Model.
Academic researchers, non-profits, and government organizations: Join SafeGraph’s COVID-19 Data Consortium to get SafeGraph data at no-cost.
If you are with a for-profit business, please contact our team today to learn how these datasets can help your business navigate this unfolding health and economic crisis.