Field notes: the Black Friday count that didn't match the earnings call

Every year there's at least one name on the coverage list where the lot looked packed on the Friday and the comp came in flat, or worse, slightly negative. The analyst who called it "soft" on the print gets a polite correction in the Q&A, something about timing shifts or a promotional calendar that moved spend into October. Nobody's lying. The lot really was full. The comp really was flat. Both things happened, and reconciling them is most of the job.

Start with what "full" means on the day. A lot that reads near capacity at 11 a.m. on Black Friday can sit at a fraction of that by 2 p.m., then spike again after dinner for the folks who waited out the doorbuster line. A single pass, or even three passes, tells you the shape of one day. It doesn't tell you whether that day was up or down against last year's Black Friday, which is the only comparison that matters to the comp. The series, indexed to the same site over enough weeks, is what keeps a single loud Friday from getting mistaken for a trend.

Then there's the mix problem. Black Friday traffic skews toward categories that don't convert at the register the way they used to. A car in the lot might be there for a return, a price-match, a scroll through the app while sitting in the warm vehicle before walking in, or a curbside pickup that never puts a body through the door at all. BOPIS killed the old one-to-one between a parked car and a transaction years ago, and the gap is biggest at exactly the retailers and exactly the weekend where it gets discussed most. A tenant can post a genuinely full lot and a genuinely soft comp in the same quarter and not be hiding anything.

Store count is the other variable that gets dropped in the hallway version of the story. If a chain closed a dozen underperforming boxes since last November, same-store-sales math drops those locations out of the base entirely, but the aggregate lot-fullness read across the portfolio can still look strong because the remaining stores are the better-performing ones that were always going to draw a crowd. Compare the wrong set of sites and you'll swear the company is sandbagging. Compare the right set, store for store, and the numbers usually come back into line.

The parking lot is a different signal than the one management reports, built on different math, and the two read best side by side rather than as a check against each other. The value of a daily read is having a per-site occupancy series going back far enough that you can see whether the softness started in week one of the promotional calendar or week four, whether it's a weather week or a four-week pattern, and whether the site that worries you is the outlier or just the one you happened to be watching that day.

A single loud day is one data point among many. The count earns its keep over a quarter, once enough weeks stack up to show whether the Friday spike held or faded. This trade works best when the chart gets built before the call happens, so nobody's pulling satellite images the morning after to explain a note that already went out wrong.

Run your own site list against a daily series and see where the lot and the comp actually agree.