NETWORK INSIGHTS · IP GEOLOCATION
IP Address Geolocation Accuracy Report - IPv4
The industry's only public accuracy benchmark. Updated daily with real ground-truth GPS data tested against multiple providers.
How to read this report
This report compares IP geolocation estimates with real-world GPS observations. It is updated daily and shows how close each dataset gets at country level, by distance from the actual location, and — where supported — whether the real location falls inside the provider's stated confidence area.
Why it matters
IP geolocation is probabilistic. A useful benchmark should show uncertainty, not just a single dot on a map.
METHODOLOGY
How this benchmark measures IP geolocation accuracy
Ground-truth data comes from many hundreds of websites and mobile apps operated across the globe. These services ask real users to share device location, sending anonymous GPS + IP pairings to BigDataCloud in near real time. Each daily report is based on the latest 24 hours of observations, measured from midnight UTC to midnight UTC. We believe this sample is a strong representation of the actual user traffic that a typical IP geolocation service customer would expect to see.
Reference data
The report compares provider results against recent GPS-derived observations from real end-user traffic, not against registry, WHOIS or routing metadata alone.
Processing
Each IP/location pair is tested against all provider datasets simultaneously, so every provider is evaluated on identical samples under identical conditions.
Fair comparison
Every sample is tested against all available providers in near real time, including BigDataCloud's own production dataset as a normal provider result.
Fairness and limitations
This is a real-user traffic benchmark, not a universal benchmark for every IP type. It is not primarily a physical-location test for routers, servers, VPN exit nodes or other passive infrastructure.
Training-window separation
The report is compiled from the latest 24-hour reporting window — midnight UTC to midnight UTC — before that window is incorporated into BigDataCloud's live dataset. We apply the same process to every provider and are doing our best to make this report as fair and unbiased as possible.
Exclusions and privacy
Abusive or clearly invalid samples are excluded. Anonymiser traffic — including VPNs, proxies and Tor — is included at natural proportions. No personal data is collected or stored in the report.
CONFIDENCE AREA ACCURACY
Does the real location fall within the predicted area?
All network types shown · 39,499 total distinct IPv4 addresses
What is this?
BigDataCloud's patented technology predicts a geographic area where an IP is most likely located. MaxMind uses an accuracy radius for a similar purpose. This chart shows what percentage of test samples had their actual GPS location fall within the provider's predicted area.
Why only two providers?
Only BigDataCloud and MaxMind publish geographic area estimates. Other providers report only a point coordinate without an area boundary.
Why it matters
A higher hit ratio means fewer false positives in your application — the predicted area reliably contains the real location.
| Provider | All networks | Wired | Cellular | Hosting / mixed |
|---|---|---|---|---|
| BDC | 97.0% | 98.0% | 94.8% | 26.2% |
| MaxMind GeoLite2 | 40.7% | 41.3% | 35.3% | 10.2% |
Table values use the same live API data as the chart above.
POINT ACCURACY
How close is the estimated location to reality?
Each line shows the cumulative percentage of IPs geolocated within a given distance of the actual GPS location. Higher lines at every distance threshold mean better accuracy — the gap at close range (5–50km) matters most for city-level use cases. Results are broken down by network type below.
All networks combined
39,499 distinct IPv4 addresses
Point accuracy measures the straight-line distance between the provider's estimated location and the actual GPS coordinates. This view combines all network types — wired, cellular, and hosting — for an overall picture. Results are based on distinct IP addresses regardless of the number of location samples received.
| Provider | ≤ 5km | ≤ 10km | ≤ 25km | ≤ 50km | ≤ 100km | ≤ 250km | ≤ 500km | ≤ 1000km |
|---|---|---|---|---|---|---|---|---|
| BDC | 12.8% | 23.2% | 42.3% | 49.0% | 58.7% | 69.8% | 84.8% | 98.0% |
| IP2Location | 4.6% | 12.9% | 32.5% | 36.1% | 45.0% | 58.7% | 79.0% | 97.3% |
| MaxMind GeoLite2 | 6.4% | 17.0% | 42.0% | 47.6% | 57.1% | 69.6% | 84.3% | 97.7% |
| db-ip | 3.8% | 9.9% | 27.8% | 32.2% | 39.7% | 49.5% | 69.1% | 96.5% |
| WhoisXmlApi | 5.1% | 13.3% | 34.2% | 38.7% | 46.8% | 58.9% | 78.9% | 97.1% |
| ipapi.is | 4.7% | 12.5% | 33.0% | 36.7% | 44.8% | 57.0% | 78.4% | 97.0% |
Cumulative accuracy by distance threshold for all networks combined.
Wired consumers
37,419 distinct IPv4 addresses
This segment primarily represents fixed network installations such as wired and WiFi connections, including both home and office consumers. Fixed networks are the most accurately geolocated because the IP address stays in one place. Understanding wired consumer geolocation is crucial for delivering tailored content, improving service delivery, and enhancing security measures.
| Provider | ≤ 5km | ≤ 10km | ≤ 25km | ≤ 50km | ≤ 100km | ≤ 250km | ≤ 500km | ≤ 1000km |
|---|---|---|---|---|---|---|---|---|
| BDC | 13.0% | 23.6% | 43.1% | 50.0% | 59.6% | 70.6% | 85.7% | 99.0% |
| IP2Location | 4.6% | 13.2% | 33.3% | 37.0% | 45.5% | 59.1% | 79.7% | 98.4% |
| MaxMind GeoLite2 | 6.5% | 17.3% | 42.9% | 48.6% | 57.8% | 70.4% | 85.2% | 98.8% |
| db-ip | 3.9% | 10.1% | 28.6% | 33.2% | 40.6% | 50.6% | 70.2% | 97.5% |
| WhoisXmlApi | 5.2% | 13.6% | 35.1% | 39.7% | 47.6% | 59.7% | 79.7% | 98.1% |
| ipapi.is | 4.8% | 12.7% | 33.8% | 37.5% | 45.3% | 57.4% | 79.1% | 98.0% |
Cumulative accuracy by distance threshold for wired consumers.
Cellular consumers
1,641 distinct IPv4 addresses
Cellular consumers access the internet through mobile networks — a group that has seen exponential growth. Mobile networks are harder to geolocate because carriers often route traffic through centralised gateways that may be hundreds of kilometres from the user's actual location. Accurate cellular geolocation is crucial for mobile advertising, content customisation, and fraud prevention.
| Provider | ≤ 5km | ≤ 10km | ≤ 25km | ≤ 50km | ≤ 100km | ≤ 250km | ≤ 500km | ≤ 1000km |
|---|---|---|---|---|---|---|---|---|
| BDC | 9.0% | 18.9% | 33.0% | 37.8% | 52.1% | 67.3% | 83.9% | 95.7% |
| IP2Location | 4.6% | 10.6% | 19.5% | 24.2% | 43.3% | 61.2% | 81.4% | 94.5% |
| MaxMind GeoLite2 | 4.3% | 13.0% | 30.1% | 37.3% | 53.1% | 68.8% | 83.8% | 95.9% |
| db-ip | 2.3% | 6.0% | 13.5% | 16.5% | 27.2% | 35.5% | 59.1% | 95.0% |
| WhoisXmlApi | 3.0% | 9.0% | 19.5% | 24.4% | 39.9% | 52.9% | 78.7% | 95.5% |
| ipapi.is | 3.9% | 10.7% | 21.9% | 27.0% | 44.4% | 59.3% | 79.4% | 95.2% |
Cumulative accuracy by distance threshold for cellular consumers.
Hosting and mixed network consumers
493 distinct IPv4 addresses
This category represents IP addresses detected as servicing hosting networks, VPNs, proxies, and mixed networks. Mixed networks are subnets that serve legitimate customers like cellular users but are also subleased to host VPN or proxy services. This category also includes residential proxies. Accuracy here is inherently lower due to the geographic separation between infrastructure and actual users.
| Provider | ≤ 5km | ≤ 10km | ≤ 25km | ≤ 50km | ≤ 100km | ≤ 250km | ≤ 500km | ≤ 1000km |
|---|---|---|---|---|---|---|---|---|
| BDC | 6.9% | 7.9% | 11.8% | 12.8% | 13.6% | 14.2% | 21.5% | 28.9% |
| IP2Location | 0.8% | 2.8% | 5.3% | 6.3% | 9.1% | 10.2% | 13.4% | 18.9% |
| MaxMind GeoLite2 | 2.0% | 4.9% | 7.1% | 7.9% | 9.3% | 12.6% | 15.0% | 21.3% |
| db-ip | 2.2% | 4.7% | 8.9% | 9.3% | 10.3% | 11.4% | 15.0% | 20.5% |
| WhoisXmlApi | 2.2% | 4.7% | 8.3% | 9.1% | 9.3% | 10.4% | 14.6% | 19.5% |
| ipapi.is | 2.0% | 3.9% | 5.9% | 6.9% | 9.1% | 10.2% | 14.6% | 19.9% |
Cumulative accuracy by distance threshold for hosting and mixed network consumers.
COUNTRY ACCURACY
Correct country identification rate
All network types shown · 39,499 total distinct IPv4 addresses
What is this?
The most basic geolocation test — did the provider identify the correct country? Country detection is relatively easy for wired networks. Real differences emerge with hosting and mixed infrastructure.
Why the scores are similar
All major providers get country right most of the time for wired networks. The gap matters most in hosting/CDN traffic where shared infrastructure crosses borders.
| Provider | All networks | Wired | Cellular | Hosting / mixed |
|---|---|---|---|---|
| BDC | 98.5% | 99.6% | 95.7% | 28.7% |
| IP2Location | 98.5% | 99.6% | 95.7% | 22.2% |
| MaxMind GeoLite2 | 98.5% | 99.6% | 95.7% | 22.2% |
| db-ip | 98.5% | 99.6% | 95.7% | 22.1% |
| WhoisXmlApi | 98.5% | 99.6% | 95.7% | 21.7% |
| ipapi.is | 98.5% | 99.6% | 95.7% | 21.7% |
Table values use the same live API data as the chart above.
INCLUDED PROVIDERS
Tested datasets — Saudi Arabia
We are aiming to include as many IP geolocation original data vendors as possible. We guarantee that we make absolutely no use of another provider's data beyond the scope of this report.
If you are an IP geolocation provider, please get in touch with us and let us access your data — API access is preferred. We will happily include you in our automated report and promise to handle your data with great respect and confidentiality.
| Provider | Product | IP Blocks* | Distinct Locations | Dataset |
|---|---|---|---|---|
| BDC | BigDataCloud API | 500,716 | 7,344 | Jul 5, 2026 |
| IP2Location | DB5 IP2Location IP-Country-Region-City-Latitude-Longitude Database | 15,217 | 77 | Jul 3, 2026 |
| MaxMind GeoLite2 | MaxMind GeoLite2 Free Geolocation Data | 4,639 | 56 | Jul 3, 2026 |
| db-ip | DbIp IP to Location with Monthly updates | 29,593 | 466 | Jul 1, 2026 |
| WhoisXmlApi | WhoisXmlApi IP Geolocation Data Feed | 19,447 | 343 | Jul 4, 2026 |
| ipapi.is | IP to Geolocation Database (commercial) | 16,367 | 310 | Jul 5, 2026 |
* The blocks count field represents the total number of unique records in each dataset. This value is generally applicable to flat data sources. However, the BigDataCloud database does not have such a metric available as we resolve down to a single IP address. We therefore provide the total number of network segments detected by the likelihood of serving the same territory and purpose.
| Network segment | Sample count (IPv4) |
|---|---|
| All networks | 39,499 distinct IPv4 addresses |
| Wired consumers | 37,417 distinct IPv4 addresses |
| Cellular consumers | 1,641 distinct IPv4 addresses |
| Hosting and mixed network consumers | 492 distinct IPv4 addresses |
Want this accuracy in your application?
Start with our free tier. Same patented technology, same daily-updated data.