IP Geolocation Data Quality in 2026: How to Actually Evaluate Accuracy, VPN Coverage, and False Positives

Why Vendor Accuracy Claims Mean Almost Nothing

When a vendor advertises "99.9% accuracy," that number usually refers to country-level resolution on a curated test set of known IPs. It does not tell you how accurate their city-level results are for your specific traffic mix, how well they detect residential proxies used in your region, or how many legitimate VPN users they will misclassify.

Accuracy is not a single number. It is a matrix of dimensions that vary by geography, IP type, and use case. The vendor whose overall accuracy is highest may perform worse than a competitor on the specific dimension that matters most to your application.

Five Dimensions of IP Intelligence Data Quality

Before you sign a contract or write integration code, evaluate vendors across these five dimensions. Each one affects a different part of your application, and the relative importance depends on your use case.

How to Build Your Own Accuracy Test

Do not rely on vendor-provided benchmarks. Build your own ground-truth dataset from IPs whose real location you can verify independently. Here is the methodology:

# Test script: Query an IP against the geolocation API
curl -s "https://ip-info.app/api/v1/geolocate/8.8.8.8" \
  -H "x-api-key: YOUR_API_KEY" | python3 -m json.tool

# Expected response fields to evaluate:
# countryCode, city.name, city.latitude, city.longitude,
# asn, aso, isVPN, isProxy, isTor, threatLevel,
# city.accuracy_radius, registeredCountryCode

# TypeScript: Calculate accuracy by region
interface TestResult {
  ip: string;
  knownCountry: string;
  vendorCountry: string;
  knownCity?: string;
  vendorCity?: string;
}

function scoreByRegion(results: TestResult[]) {
  const regions: Record<string, { total: number; correct: number }> = {};

  for (const r of results) {
    const region = r.knownCountry;
    if (!regions[region]) regions[region] = { total: 0, correct: 0 };
    regions[region].total++;
    if (r.knownCountry === r.vendorCountry) {
      regions[region].correct++;
    }
  }

  // Sort by volume to prioritize your largest markets
  return Object.entries(regions)
    .sort((a, b) => b[1].total - a[1].total)
    .map(([region, data]) => ({
      region,
      accuracy: ((data.correct / data.total) * 100).toFixed(1) + '%',
      sampleSize: data.total,
    }));
}

VPN and Proxy Detection: What Accuracy Actually Means

VPN detection is not binary. The landscape includes commercial VPNs (NordVPN, ExpressVPN), corporate VPNs and SD-WAN platforms, mobile carrier NAT pools that look like proxy traffic, consumer privacy networks (Apple Private Relay, Cloudflare WARP), Tor exit nodes, and residential proxy networks. Each category has different detection difficulty and different false-positive risk.

When you evaluate vendors, ask for detection rates broken down by these categories, not a single aggregate number. A vendor that detects 99% of commercial VPNs but only 60% of residential proxies is not necessarily worse than one that detects 95% of both — it depends on whether your fraud exposure comes from VPN users or proxy networks.

Regional Accuracy Gaps That Catch Teams Off Guard

IP geolocation accuracy is not uniform. It follows the distribution of internet infrastructure investment, BGP visibility, and the density of ground-truth measurement networks. Here is what to expect across major regions:

These ranges are typical across the industry, not specific to any single vendor. The point is that if your application serves users in Africa or Southeast Asia, you need to test those regions specifically — a vendor who excels in North America may not be the best choice for your actual traffic distribution.

Buyer's Scorecard: What to Ask Every Vendor

Use this framework when evaluating IP intelligence providers. These are the questions that separate marketing claims from operational reality:

Frequently Asked Questions