Four moves between a satellite pass and a number you can trade on. Set nothing up — the pipeline runs whether you're watching or not.
Each stage stands on its own. You only touch the parts you want to.
Every week, fresh overhead imagery covers each location in the universe. Angle, season, and resolution are normalized, so a lot in Phoenix and a lot in Buffalo are measured on the same ruler.
A vision model finds every vehicle in the frame and tallies it. It's tuned to skip the delivery bays, the employee row, and the overflow field next door — so the count reflects shoppers, not staff.
Raw counts mean nothing alone. Each site is scored against its own seasonal history and its peers, turning a number into a trend: busier than usual, or quietly emptying out.
Site trends aggregate up to the chain. The output is a demand read for the whole company — beat, in-line, or miss taking shape — with a confidence score and the lead time attached.
What actually happens between the satellite and your screen.
Pulls commercial overhead imagery for every site on a weekly cadence and normalizes it, so counts are comparable across geography, weather, and time of day.
Finds and tallies vehicles in each lot, trained specifically on retail parking to filter out staff, delivery, and adjacent traffic that isn't a shopper.
Benchmarks each site against its own history, rolls counts into a chain-level demand index, and attaches confidence and lead time to every read.
The best signal is the one already in the ground before anyone thinks to ask for it.
— How Lotwyse is meant to feel
Request access and we'll run the whole pipeline on a live location for a chain you actually care about.