Recency Analysis – Mall Stores vs. Quick Service

Recency Analysis – Mall Stores vs. Quick Service

By Dave Strauss, Analytics | #EuclidData | 30 May, 2014

At Euclid, we have the ability to understand visit patterns for smartphone-toting individuals who frequent physical locations. We have been working with some clients for multiple years now and have had time to look at some long term trends in recency behavior at different types of retail locations. Two distinct trends can be seen, a weekly trend and a yearly trend. The yearly trend is much more prevalent at specialty retail locations located in a mall than at quick service locations.

We establish these trends by looking at one of our primary metrics, recency, which measures how people behave around a store over long periods of time. Recency measures the number of days between successive visits of a device to a physical location. If I show up at a store on April 1 and on April 20, then I have a “recency” of 20 days.

In the plot below, we show the likelihood of seeing a device with any recency averaged across many stores over a multiple years of observation. The x-axis is the recency measurement, the number of days between successive visits, and the y-axis measures the likelihood. The plot shows that we are much less likely to observe a device with a recency of 300 days than a device with a recency of 5 days. However, we can also see two additional trends. The short term oscillations occur at 7-day intervals which means that people are more likely to return on the same day of the week as their original visit than any other day of the week. Generally, this makes sense. Weekends tend to be busier for a large majority of retail locations.

Additionally, we see a strong peak right at the 365 day mark which shows that returning to a physical space once a year is far more likely than returning once every 350 days. For mall locations, this effect is more pronounced than for quick service locations. In fact, in a mall location, a device is 2.5 times as likely to return in exactly 365 days as it is to return in 345 days, while for a quick service location, a device is only 1.2 times as likely to return in exactly 365 days than it is to return in 345 days. Interestingly, those 20 days make a significant difference for repeat customers.

Dave Strauss, Analytics

Dave Strauss, Analytics

More posts by Dave Strauss, Analytics