Testing & Measurement Policy
Published: March 2026 • Last Reviewed: September 14, 2026
VelocityVerify provides browser-based broadband diagnostics designed to make test conditions and limitations clear. Measurements are delivered through Cloudflare infrastructure and should be interpreted alongside local Wi-Fi conditions, device limits, routing, and congestion.
Core Guarantee: We never accept payment, hardware loan conditions, or sponsorship that influences testing outcomes. All test results represent objective physical network telemetry measured under documented, scientific conditions.
1. Independent Measurement Infrastructure
Most commercial speed tests prioritize endpoints co-located directly inside the user's ISP network. While this measures the raw local line rate to the nearest central office, it conceals real-world peering congestion, transit bottlenecks, and bufferbloat. Our measurement topology operates under three foundational rules:
- Hosting disclosure: The current service is hosted on Cloudflare Pages and uses Cloudflare Pages Functions for test endpoints.
- Multi-CDN Probing: Diagnostic evaluations utilize multi-CDN concurrent requests to test real-world application pathways rather than a single synthetic channel.
- Versioned Algorithms: Our diagnostic algorithms (including bufferbloat P90 calculations and jitter variance analysis) are explicitly versioned and documented in our Engineering Methodology.
2. Browser Test Conditions
Results depend on the visitor's device, browser, local network, Wi-Fi conditions, route, and current congestion. For repeatable comparisons, use the same device and connection method, pause unrelated traffic, run several tests, and compare like-for-like conditions.
- Wired baseline: Use Ethernet when possible to separate local Wi-Fi limitations from the internet connection.
- Repeated runs: Run multiple tests at different times rather than treating one result as conclusive.
- Document limitations: Browser measurements are diagnostic indicators, not proof of intentional throttling or formal carrier certification.
3. Transparent Metric Definitions
Network performance cannot be reduced to a single misleading number. Our tests report disaggregated, scientifically verifiable metrics:
| Metric | Measurement Protocol | What It Validates |
|---|---|---|
| Idle Latency | ICMP/HTTP round-trip time across 10 consecutive probes before data transfer | Baseline physical route distance to neutral peering exchanges |
| Loaded Latency (Bufferbloat) | Continuous HTTP ping sampling while saturating downstream and upstream pipes | Router queue management (FQ-CoDel/CAKE) and ISP modem buffer sizing |
| Jitter | Statistical standard deviation (P90) of consecutive packet arrival times | Connection stability required for gaming tickrates and VoIP streams |
| Sustained Throughput | Multi-threaded payload transfer discarding initial burst ramp-up TCP windows | True usable application bandwidth delivered to the user device |
4. Defensible Diagnostic Language
A single browser speed test cannot legally or technically prove intentional ISP traffic discrimination or throttling. Our reports and diagnostic tools maintain strict scientific caution: we distinguish between local Wi-Fi contention, neighborhood node congestion, and protocol-specific traffic shaping. We state observations as engineering hypotheses backed by evidence, never sensationalized accusations.
5. Review Cycles & method reviews
Our testing infrastructure is audited monthly by site maintainers (Younes Kirat) to verify calibration against industry standards (RFC 6349, ITU-T Y.1564). When broadband standards evolve or protocol updates occur, documentation is updated with public change logs.