Troubleshoot before replacing an IP

12CE editorial · Updated 2026-09-09. Scenarios below illustrate a diagnostic method; they are not benchmark results or quality guarantees.

Conflicting risk scores

First confirm that the results describe the same exit IP and are current. A rotating proxy can change addresses between tests. Missing data and historical caches are not a zero score.

Our earlier troubleshooting included a public endpoint returning HTTP 403. That was a rejected query, not a high risk score for the tested IP. Retry the failed provider later. Do not average scores: fraud prediction, proxy classification and report-based abuse confidence measure different things. See IPQS fields and AbuseIPDB FAQ.

1. Check the provider, timestamp/cache label and flags.
2. Retry missing providers individually.
3. Verify your actual intended service.
4. Contact the supplier if concerning signals persist and affect that use. A low score is not a guarantee.

Different DNS exits

The test observes recursive resolvers querying our authoritative node, not necessarily the DNS address configured on your device. A different IP or country alone is not a leak verdict.

1. Write down your intended routing policy.
2. Run the DNS test in the same browser/network.
3. Change one DNS or proxy setting at a time and compare.
4. Investigate a reproducible mismatch with that policy. No observed resolver is inconclusive, not proof of no leak.

WebRTC observes candidate paths, not DNS resolvers. These checks complement each other. Read MDN's WebRTC connectivity explanation. Our current node cannot validate IPv6; that is not a statement about your device.

Streaming is unconfirmed

Connection and page features do not prove playback. A consent page, challenge, template change or timeout may affect the result. Even a normal page cannot prove subscription eligibility.

1. Check that your browser and tested proxy use the same exit.
2. Retry only the unconfirmed service.
3. Verify normal playback with an authorized account, without bypassing access controls.
4. Share the time, test status and observed behavior with support. Do not share cookies or proxy credentials.

Use the redacted text report as a starting point. Add an IP separately only when necessary, through a trusted channel. These examples do not establish a measured false-positive rate.

Redacted historical observations

Our audit on 2026-09-15 covered 500 saved checks. In 346 records, IP2Location was at least 66 while Scamalytics was 0. Scores are not interchangeable; this alone does not identify the cause of disagreement.

Of 323 initial YouTube failures in that batch, 90 matched follow-up records obtained an HTTP 200 and Premium page signal; 233 did not. This is not verified playback or subscription availability. The sample was not random and concurrency was not controlled, so it is not a general error-rate estimate or proof of a concurrency effect. Account names, IPs and ports are not published.