Metrics & methodology
How we measure proxy performance.
We are deliberately publishing the methodology first and the live numbers second. This page exists so that any number we ever publish on a marketing page can be traced back to a written definition, a measurement window, and a product scope.
Last reviewed: 2026-06-20.
Principles
What we will and will not claim about performance.
- Each metric has a written definition before we publish a number for it.
- Each published number is tied to a measurement window, sample size, and product scope.
- We distinguish gateway availability, upstream connect success, and target-request success — and never collapse them into a single "uptime" figure.
- If a number cannot be measured from the live system, it is not published. We will name the missing data feed instead of estimating.
- We do not publish unaudited pool-size counts. Pool composition for your account is shown in the customer portal.
Defined metrics
The four metrics we are committing to publish.
These metrics are scaffolded today. Definitions are frozen; live numbers are gated on the public metrics data feed and will appear here once the feed is connected.
| Metric | Definition | Scope | Status |
|---|---|---|---|
| Gateway availability | Percentage of seconds in the measurement window during which the rotating gateway health endpoint reported at least one healthy upstream provider. | Rotating gateway, rolling 30-day window, measured from internal health checks. | Scaffolded — data feed pending |
| Upstream provider success rate | Successful upstream connects ÷ total upstream connect attempts. Counts TCP-level connect success against the selected provider, not target-application response codes. | Per provider type (datacenter, residential, mobile), rolling 30-day window. | Scaffolded — data feed pending |
| Target-request success rate | Customer-facing requests through the gateway that returned a 2xx/3xx target response ÷ total target requests. Excludes 4xx caused by customer credentials and 5xx caused by the target site. | Rotating gateway, rolling 30-day window, aggregated across active accounts. | Not currently published |
| Assigned-product uptime | Percentage of minutes our assigned (static) endpoints accepted authenticated connections during the measurement window. Does not measure target-site reachability. | Assigned datacenter, ISP, residential, mobile, and IPv6 endpoints. Rolling 30-day window. | Scaffolded — data feed pending |
Definitions reviewed quarterly. Live numbers will refresh at most once per 24 hours once the metrics feed is connected.
What is not on this page
Things we do not publish here.
- Per-target benchmarks (e.g. success rate against a specific site). Real-world success is highly target-dependent and we do not publish vendor-curated benchmarks.
- Unaudited IP pool size. See /capabilities for our disclosure policy.
- Real-time status. For incident updates and live service status, see /status.