Discussions

Hi,

We’re using the Advan monthly visit patterns data to approximate monthly employment at fast food restaurant locations, defining employment as the number of visitors who stayed for more than 4 hours. Our goal is to treat these long-duration visits as a proxy for employee shifts.

I have created using the data. It looks at the number of visitors with bucket_dwell_times >4 hours as a proxy for the number of employee shifts, since we assume most people staying at a fast food restaurant for more than 4 hours are employees not patrons.

In reviewing the series, we’ve noticed a few patterns that we’d like to clarify:

Large Dips: Certain months show sharp declines that seem too large to be explained by normal seasonality. Could these be due to panel changes, data coverage gaps, or adjustments?

High Shift Counts: Industry benchmarks suggest employees typically work ~12–16 shifts per month. The data often implies far more shifts per employee, especially before 2020. Could you explain why these counts are so elevated?

Instability Over Time: The series doesn’t remain stable across months — employment fluctuates in ways that don’t align with true headcount. Is this driven by panel expansion, POI updates, or another methodological factor?

Any guidance on how to interpret these issues?

Best,
Emma-Jane

(sorry i've tried to contact advan directly but they don't seem to have a contact protal/forum)