Store locator

The problemThe old locator ran on ProMap at $839.88 a year, and it was slow enough that support kept filing bugs about it. When it was raised in May 2025 the answer was that it could not be improved: "I can not improve this, it is related with count of locations." So the constraint was never really technical. A limit in the tool had become a limit on the business.

The decisionRebuild it rather than tune it, and refuse to pay for the part everyone assumed you had to pay for. Storefront photos would have meant the Places API at around $700. Street View has a free metadata endpoint that returns where a panorama is shot from, so I computed a camera heading per store from that instead — 43,943 of them — and pointed each one at the building rather than due north.

The result46,434 stores, 22,891 of them confirmed stockists. Riverside went from the 26 ProMap could find to 60. Storefront photography on 97.5% of stores for nothing. ProMap uninstalled.

The before and after is one image: heading 0 is due north, which is what a naive build ships. Milwaukee at due north is an empty alley. Milwaukee at its computed heading is the door of the shop.