Case study · Proof of concept · Palm Beach County, FL
A map-based tool that ranks possible locations for a new franchise site using public data. I built it on my own initiative to show how public data can rank franchise sites. It was not commissioned by, and is not affiliated with or endorsed by, any brand or franchisee.
The question
The example scenario is a single auto-service site for a single franchisee of a national auto-service brand, inside Palm Beach County, FL. The tool places the brand's existing locations on a map, draws an adjustable exclusion radius around each one, and ranks the places outside those circles where a new site could make sense.
It is a working proof of concept, not a client project. Everything it shows comes from public sources, and it says so plainly when an input is an assumption or a proxy.
How it works
Existing locations are read from the brand's public store pages and cross-checked against OpenStreetMap. Each one gets an exclusion circle with an adjustable radius (default 3 miles, range 0.5 to 15). The radius is labeled as an assumption, because real territory terms live in franchise documents the tool has not seen.
Candidate points are sampled every 0.4 miles along FDOT-counted roads in the county, with interstates, freeways, and ramps removed. Points inside any exclusion circle are left out.
A second view ranks individual Palm Beach County Property Appraiser parcels that pass the property filters and lie outside every exclusion circle. Parcels are potential sites only. They are not confirmed listings, prices, or rents.
Each factor becomes a percentile rank among the sites that pass the filters, and the score is the weighted average × 100. Every weight is adjustable, and the map, ranked list, and match counts update as you change weights or filters.
Parcel records don't include service-bay counts, so automotive and industrial/flex parcels are labeled likely multi-bay, likely single-bay, or unknown, based on Florida Department of Revenue building area and an adjustable cutoff (default 3,000 sq ft). The cutoff is shown as an assumption.
On a phone, the map, filters, and ranked sites sit behind a bottom tab bar. A "?" button next to each setting opens a short explanation of where its data comes from and how it is scored, and a "How to use" panel walks through the steps.
Screenshots
Screenshots have been edited to remove brand names and private details.
Public data
Commercial listing sites are not scraped. Each ranked site links out to listing searches instead.
The data refresh wraps every source so a failure is recorded instead of stopping the run. If a source fails, the page sets that factor's weight to 0 and shows a note explaining why. It never fills in estimated values in place of missing data.
How it's built
The app is static HTML, CSS, and vanilla JavaScript, with Leaflet for the map, stored locally rather than loaded from a CDN. There is no build step. The data is refreshed with a separate script, and the page does all of its scoring in the browser.
See it working
Want to see the site finder in action, or talk about ranking locations with your own data? Get in touch to set up a walkthrough.