What counts as a comparable period for lot counts vs. reported comps
When a retailer reports same-store sales up 3.2% and your lot occupancy series is flat for the same ten weeks, the first question isn't whether the satellite data is wrong. It's whether you compared the same stretch of calendar in the first place. Most mismatches between parking lot counts and reported comps trace back to this, not to vehicle-counting error.
Retailers don't report against the Gregorian calendar you're using to pull imagery. They report against a fiscal calendar built for their own internal accounting, and that calendar rarely lines up with the one on your wall the way you'd assume.
Retail's fiscal calendar isn't your calendar
Most retail and restaurant chains run a 4-4-5 fiscal quarter: three months split into reporting periods of 4 weeks, 4 weeks, and 5 weeks, so every quarter ends on a Saturday (or sometimes Sunday) rather than the last day of a calendar month. That means "Q3" for a retailer with a January fiscal year-end can run from late April through late July, not July 1 through September 30. If you chart lot counts for the calendar quarter and hold it against a 4-4-5 comp, you've already built in a week or two of drift before you've looked at a single vehicle.
Then there's the 53-week year. Because 4-4-5 calendars track to 52 weeks of 7 days, the arithmetic falls short of the solar year by about a day annually, and roughly every five or six years a company tacks on an extra week to true it up. Any company that did this last year is now reporting a comp period shifted by seven days relative to companies that didn't, and relative to its own prior-year comp. Pull the investor relations fiscal calendar for each tenant before you set your date range, not after.
Day-of-week and holiday placement shift the comparison too
Even inside a correctly aligned fiscal quarter, the day-of-week composition matters. A comp period with five Saturdays draws a different lot occupancy pattern than one with four, independent of any real change in traffic. The same goes for how major holidays fall. Easter can land in March or April depending on the year, and a retailer's "Q1 comp" either captures the pre-Easter shopping surge or doesn't, which shows up as a swing in reported comps that has nothing to do with underlying demand. Thanksgiving's fixed to the fourth Thursday of November, but the number of shopping days between it and Christmas still shifts by up to six days year over year, which matters a lot if you're trying to read a holiday-quarter lot series against last year's.
The fix is the same one retail finance teams already use internally: align by day-of-week position within the fiscal period, not by raw date. If last year's comp window ran Sunday through Saturday across weeks 14 through 22, this year's window should be defined the same way, even if the calendar dates it lands on are a few days off from last year's.
Building the window before you chart anything
In practice, that means doing three things before you overlay a lot occupancy series on a reported comp number: pull the tenant's own fiscal calendar from its 10-Q or investor deck rather than assuming calendar quarters, check whether either comparison year contains a 53rd week, and confirm the day-of-week alignment of the start and end dates rather than just matching month and day. Skip any of these and you're explaining a variance that's really just a calendar artifact.
Once the window is set correctly, a daily occupancy series is actually easier to work with than most footfall proxies, because you can resample it to whatever period boundary the tenant uses instead of being stuck with whatever bucket a third-party panel already baked in. Parking Lot Counts builds that series per site from repeat satellite passes, so you're the one deciding where the comp period starts and ends, not inheriting someone else's week.
Get the fiscal window right first. The chart will tell you more once it is.