Originally published on proxymint.
Geo targeted proxies filter exits by country, region or city. How accurate each level is, why city pools run thin, and when country targeting is enough.
TL;DR
- A geo target is a label plus a live pool. The label comes from an IP geolocation database, and the pool is whatever devices are online in that place when the order runs. Both get weaker as the target gets finer.
- City labels are estimates. MaxMind estimates 99.8% accuracy at country level but 66% for US cities within 50 km (checked September 2026). Mobile carrier addresses may resolve only to country and state, and the target runs its own lookup.
- Pool depth drops at every level. Per-request rotation in a small city cycles the same few exits, so the request rate has to stay at what the target allows.
- proxymint targets country, region and city only. Residential and mobile carry all three; ISP and datacenter IPv4 are chosen by country at order time. There is no ZIP, ASN, ISP or carrier selection on any tier.
- The IP is one location signal of several. Accept-Language, DNS, WebRTC and the site's saved store or location can contradict a correct exit.
Key takeaways
- Match the level to the data. Target city only when the page, price or result changes by city; for state rules use region, and for national prices use country and keep the deeper pool.
- Check the exit before the job. Echo the exit IP, look it up in two geolocation databases, and only then run the real traffic.
- Count unique IPs before scaling. Send a few hundred requests per level through an IP echo endpoint, so the plan matches the exits a city actually has.
- Set the browser for the market. Set locale to the market whose pages the job needs, resolve DNS at the proxy, and start each market in a fresh profile.
- Set ZIP and store inside the site. A city exit plus the site's own postcode or store picker gets the local page without ZIP targeting.
Read the full guide on proxymint, with the commands, tables and sources.
Top comments (0)