Struggling with geo-IP accuracy for our 'What is my City Name' tool

Author
Siddharth Chopra Author
|
15 hours ago Asked
|
3 Views
|
1 Replies
0

Hey folks,

  • Context: we've got this simple web tool, "What is my City Name," our main goal for it is quick, accurate city-level identification based on a user's IP.
  • The Core Problem: we're consistently hitting a wall with geo-IP accuracy, especially for mobile users and edge cases. the data just isn't granular enough or isn't frequently updated, causing real headaches for us.
  • What We've Implemented/Tried:
    • started with MaxMind GeoLite2, then upgraded to their paid GeoIP2 City database for better resolution.
    • experimented with IPinfo.io and AbstractAPI for comparison and redundancy, hoping to find a more reliable source.
    • implemented client-side IP detection (via WebRTC STUN servers) as a fallback, but that's privacy-sensitive and not always reliable, plus it adds complexity.
    • caching strategies for frequent IPs to reduce API calls and improve speed, which helps performance but not core accuracy.
  • Specific Issues Encountered:
    • significant discrepancies between providers for the same IP, sometimes off by hundreds of miles, which is just crazy.
    • mobile carrier IPs often resolve to major cities far from the actual user, making our tool pretty useless for those users.
    • VPN/proxy detection complicates things, but even without them, accuracy is shaky.
    • database update frequencies don't seem to keep up with dynamic IP assignments, especially with ISPs recycling IPs so often.
  • Our Goal: achieve sub-10-mile accuracy for at least 80% of our traffic, or understand if this is an unrealistic expectation for geo-IP technology right now.
  • Looking For Advice On:
    • are there lesser-known, highly accurate geo-IP providers specifically for city-level resolution that we might have missed?
    • advanced techniques for combining multiple geo-IP sources to create a more reliable "consensus" location? maybe some statistical methods?
    • strategies for handling mobile carrier IPs more effectively without resorting to client-side geolocation?
    • any open-source projects or algorithms that provide better geo-IP data processing or correction?
  • appreciate any insights or experiences you guys can share! Help a brother out please...

1 Answers

0
Karan Yadav
Answered 9 hours ago
Hello Siddharth Chopra, I'll help a brother out with that geo-IP accuracy challengeโ€”yes, sometimes the data *is* just crazy! Sub-10-mile precision for 80% of traffic via IP geolocation alone is unrealistic due to inherent provider limits from ISP routing and dynamic mobile IPs; instead, for better city-level identification, consider:
  • A weighted ensemble of MaxMind GeoIP2, IPinfo.io, and AbstractAPI, prioritizing data freshness and confidence scores for a consensus location.
  • Integrating the browser's Geolocation API as a *privacy-aware fallback* (with explicit user consent) for mobile users and critical cases where IP data is insufficient.
  • Exploring specialized B2B IP intelligence providers beyond standard geo-IP databases for deeper network topology analysis and better handling of mobile carrier IPs.
Are you logging discrepancies to refine your weighting algorithm?

Your Answer

You must Log In to post an answer and earn reputation.