Field notes: the week guidance said one thing and the lot count said another

Mid-September, a retailer I cover stood up at a sell-side conference and reaffirmed the quarter. Comps tracking "in line," the CFO said, traffic "stable," the usual language you write down without really hearing. I wrote it down. Then I went back to the lot data for the chain's top forty stores and the two lines weren't telling the same story.
The counts had been softening for about three weeks. Nothing a single regional store closure or a construction detour would explain away, just a slow bleed, maybe four to six points off the seasonal baseline, concentrated in the Sun Belt cluster that had carried the chain's comps for the last two years. The Northeast stores were fine. The flagship urban boxes were fine. It was specifically the suburban strip-center locations in Texas and Florida that had gone quiet.
I sat with that for a day before doing anything with it, because a single region softening against a reaffirmed guide is exactly the kind of thing that looks like signal and turns out to be noise. First question: weather. There'd been a string of afternoon storms through the region that month, and foot traffic is genuinely sensitive to that in a way same-store sales eventually smooth over. I pulled the counts against the prior two years for the same calendar weeks. The storm pattern explained maybe a third of the gap. The rest was just fewer cars.
Second question: was this a lot-sizing artifact. Some of the chain's newer stores sit on shared pads with a grocery anchor, and if the grocery tenant was running a promotion that pulled extra cars onto the lot, my index could be reading phantom traffic in the comp period, making a real decline now look worse than it is. I checked the store list against known shared-pad locations and excluded six of them. The pattern held.
By early October the gap had widened to the point where it stopped being an interesting curiosity and became a thing I needed to write up. Guidance implied traffic roughly flat to slightly positive. The lots said mid-single-digit negative, isolated to the exact two states where the chain had leaned hardest on new unit growth the prior eighteen months. That pattern reads as cannibalization from those new units eating into the existing base, something a CFO either hasn't caught yet in the daily flash numbers or has caught and isn't ready to say on a call three weeks before the print.
I wrote the note cautious, because one region's parking lots are not a same-store sales print and I said so explicitly. But I also said the division between what management described on the call and what the asphalt showed was wide enough to flag, and that the next data point worth watching was whether the softness bled into the chain's other high-growth markets or stayed contained to the two states.
It stayed contained, mostly, and the quarter came in a little light on comps with management citing "pockets of softness in certain Sun Belt markets" on the call, almost word for word what the lot counts had been saying for five weeks by then. Nobody's claiming prophecy here. The counts don't know why traffic dropped, they just know it dropped, and figuring out the why is still the job. What they're good for is telling you where to start asking before the press release tells you where you should have been looking.
The whole case for building a daily lot-count series into the pre-print routine, rather than reaching for it after guidance gets walked back, comes down to this: the gap between what gets said on a call and what's happening on the ground shows up in the data weeks before anyone puts it in a sentence. Sales Nowcast builds that series chain by chain, store by store, so the next divergence shows up on your screen before it shows up in a revised guide.