Methodology
This page is not just a promise of how we plan to measure. It exposes the real probe network, sample size, coverage thresholds and current limitations so the source and boundaries of every data point can be inspected directly.
Current Measurement Framework
These are not static marketing statements. They describe the actual measurement and presentation rules used by the platform today.
Global Active Probing
Real probes launch requests from multiple geographies, covering mainland China and overseas regions instead of relying on a single network perspective.
Unified Sample Scope
Observation targets and aggregation rules are standardized across providers so results remain comparable.
Continuous Sampling & Aggregation
Samples are continuously written and aggregated by hour and region, while public pages read snapshots and aggregates for stable performance and consistent definitions.
Measured vs Claimed Layers
Security metrics are split between measured items and vendor-declared items so unverifiable claims do not pollute measured scores.
Probe Network Distribution
Current real probes are distributed across the regions below. This list will expand naturally as more real probes are deployed.
Show all 344 probesCollapse additional probes
Current Public Thresholds
We do not fabricate data just to look complete. Immature dimensions remain visibly downgraded.
Security Score Notes
Measured Items:TLS 1.3, weak cipher suites, certificate chain integrity, HSTS, CSP, X-Content-Type-Options, HTTP/3 and OCSP Stapling.
Declared Items:Vendor-declared capabilities such as WAF or DDoS protection are shown separately instead of being mixed into measured scores.
Known limitations
- Node coverage is inferred from real probe observations and is not an official full node list from the vendor.
- The number of real probes is still limited, especially for multi-region coverage.
- Website speed tests and observation results can be affected by target-site volatility, throttling and temporary network jitter.
- Dimensions without enough real samples remain downgraded instead of being padded with fake completeness.