Optimizing IP geolocation data accuracy for dynamic ranges?

Context: We operate an IP Lookup Tool primarily focused on providing detailed IP geolocation data. Our goal is to offer users highly precise geo-location information for any given IP address, including city, region, and country.

Problem Statement: We're encountering significant challenges with the accuracy of this IP geolocation data, particularly for dynamic IP ranges frequently assigned by ISPs and mobile carriers. Standard database solutions, such as GeoLite2 or MaxMind, often lack the necessary precision for city and region-level detail in specific geographic areas or become quickly outdated due to frequent reallocations. This leads to user frustration and diminishes the perceived value of our tool.

Specific Technical Block: Our primary technical hurdle is how to effectively update and maintain high-precision geolocation for IP blocks that are either rapidly changing ownership/location or are frequently reallocated within a broad geographical region. The static nature of many commercial databases doesn't keep pace with these dynamic assignments.

Core Question: What advanced methodologies, real-time data sources, or third-party APIs have proven most effective for enhancing IP geolocation data accuracy and update frequency, especially when dealing with highly dynamic IP address allocations?