Struggling with IP API geolocation accuracy for user segmentation?

I'm experiencing significant challenges with geolocation accuracy from various IP API providers, which is directly impacting our user segmentation and analytics. We rely heavily on this data for regional targeting and fraud detection, and the current inconsistencies are proving to be a significant technical blocker.

• Current Setup: We integrate with multiple IP APIs (e.g., MaxMind, ipstack) to get geo-IP data on inbound requests. This multi-provider approach was an attempt to mitigate single-point-of-failure issues and improve data quality.
• Core Problem: Despite using reputable services, we frequently encounter discrepancies. Users reported as being in specific countries are often misidentified, especially for mobile IPs, VPN users, or smaller, less common regions. This leads to inaccurate segmentation, erroneous analytics, and consequently, wasted marketing spend due to misdirected campaigns. It also complicates our fraud detection efforts, as legitimate users are sometimes flagged incorrectly.
• Attempts to Resolve: We've tried cross-referencing data from 2-3 providers, implementing weighted averages, and even historical data analysis to identify consistent biases. However, the inconsistencies persist across different network types and geographic locations. It feels like we're constantly fighting a losing battle against IP data reliability.
• Specific Questions:
  • Are there specific IP API providers known for superior geolocation accuracy, particularly for niche regions or mobile networks, that might offer a higher fidelity of geo-IP data?
  • What advanced validation techniques or secondary data sources (e.g., client-side signals, user-provided data with consent) do you use to improve the trustworthiness of your geo-IP data beyond just multiple API calls?
  • Has anyone implemented a robust fallback or verification system to handle rate limiting or temporary API outages without compromising data quality or introducing significant latency?
• Looking For: Best practices or architectural patterns to build a more resilient and accurate geo-IP identification system. I'm especially interested in strategies that move beyond simple API aggregation.

Any insights from those who've tackled similar issues and achieved a higher degree of precision would be greatly appreciated. Waiting for an expert reply!