Method
What this measures, what it cannot, and why it is built this way.
The ceiling is the ZIP code
Neither Meta nor Google can target a census block group or tract. Meta bottoms out at a ZIP or a 0.63-50 mile pin radius; Google at a postal code or a 1 km radius. There are roughly 240,000 block groups against 33,000 ZIP-equivalents, about seven block groups per ZIP, so anything modelled finer has to be aggregated back up before a dollar can be spent against it. Meta's own household-income targeting is bucketed by ZIP average income - the platform is making the same approximation. This tool therefore models and reports at the ZIP, which is the smallest unit that is actually spendable.
Reach is verified, not modelled
Every other product in this category stops at
"here is your segment". This one asks Meta's delivery_estimate endpoint whether
the audience exists and how big it is. Meta sometimes answers with a not-ready placeholder of
exactly 1,000 to 1,000; that is the API computing asynchronously, not an audience of a
thousand people, and it is never shown as a number.
No appended data, ever
Customer lists are geocoded to a ZIP and counted. Nothing is appended to them and nothing is uploaded anywhere. Attaching purchased demographic attributes to a client's list and pushing it to an ad platform is a "sale" or "share" under CPRA whether or not money changes hands, and doing the appending would make an agency look like a data broker rather than a service provider. Modelling geography from public aggregate data sidesteps the question entirely.
Sources
| Demographics | US Census ACS 5-year, public domain, free |
| Boundaries | Census TIGERweb ZCTA, free |
| Geocoding | US Census geocoder, free, no key |
| Drive time | OpenRouteService, Mapbox or Geoapify - all with free tiers |
| Competitors | Overture Maps (CDLA-Permissive-2.0), checked against Google Maps via DataForSEO |
| Reach | Meta Marketing API |
Every data source above is free.