Understand the limits of a test first

Proxy-service performance changes with the user's location, ISP, time of day, and server load. Results on this site describe a specific sample and do not promise that every user will obtain the same result.

• Reproducible

  Retain dates, environments, tools, time windows, and the number of test runs whenever possible.

• No unsupported extrapolation

  One speed test or one server cannot represent long-term performance across the whole service.

• Time-sensitive

  Plans, routes, IP addresses, and unlock status can change at any time.

Basic test record

A complete test should record as much of the following as possible: date and time window, location, ISP, baseline bandwidth, device and client, server region, and number of test runs. Data without this context can serve only as limited supporting information.

Primary metrics

Metric	How it is recorded	Important limitation
Latency and packet loss	Test the same server multiple times and favor the median	A single lowest-latency result does not represent long-term performance
Download and upload	Record baseline bandwidth and the test tool	Results depend on the local network and test server
Evening peak stability	Normally retest between 20:00 and 23:00 local time	One evening cannot establish long-term availability
Streaming and AI services	Record the server, region, service, and date	Unlock status may change when the IP address changes
Support response	Record the support channel and time to a useful response	One ticket cannot represent every user's experience

Evidentiary limits

• Route labels such as “IPLC,” “IEPL,” and “BGP transit” are described as provider-advertised specifications when they come only from the seller.
• A “no-logs” claim usually cannot be independently verified from outside the service and should not receive definitive endorsement based only on marketing.
• Absolute promises such as “no speed limits,” “will never shut down,” or “perfect unlocking” are not treated as verifiable facts by this site.
• When comparable quantitative data from a consistent environment is unavailable, the article should clearly identify the material as a source summary or preliminary experience.

Evidence levels

Level	Statements the evidence can support	Conclusions it cannot establish directly
A: Repeated tests by this site with environment records	“In this test” or “the median in these tests was”	Long-term stability in every region
B: One test or screenshot from this site	“Connected in this test” or “unlocked in this test”	Permanent unlocking or an enduring absence of speed limits
C: Official page or announcement	“The official specification states” or “the provider says”	Independent verification
D: User report or indirect material	“A user reported” or “there are related indications”	A general fact or final determination

Purchasing guidance

For a first purchase, the site generally recommends a short billing period and a low-cost plan, together with a backup option. No amount of historical performance can guarantee future availability.

---

Related policies: Editorial Policy · Affiliate Disclosure · Report a Data Error

Contributors: ermaozi