Data sources
Every benchmark figure on ProxyLook comes from one of three source types: our own hands-on test data, independent third-party lab reports, or the provider's published specifications. We never invent, extrapolate or estimate a number.
Test
Hands-on runs against providers we hold retail accounts with.
Collect
Independent lab reports and providers' published specs, with the source recorded.
Vet
Each figure is checked for source quality, recency and internal consistency.
Publish
Numbers go live with their sourcing; no reliable data means "—", not an estimate.
Where we test hands-on, we pay full retail for every account — no comp accounts, no "press" tier. Where a number comes from a third-party lab report, it is attributed on the provider page.
Source hierarchy
When more than one source exists for the same metric, we use this order of preference. A lower-preference source is only used when nothing better is available — and the page says which one was used.
Compiled figures across providers are sortable on our benchmarks dashboard →.
Vetting rules
Before a number is published it has to pass all three gates. Any one failing means the metric stays at "—" — this is what keeps marketing spin and stale figures out of the tables.
Identifiable source
We can name where the figure came from — our own run, a specific lab report, or a specific provider document.
Recent enough to matter
Figures from sources that have gone stale (product relaunches, network changes, old reports) are dropped rather than carried forward.
Internally consistent
A number that contradicts the provider's own documentation or other credible measurements gets investigated, not averaged.
What we report
Five metrics per provider, filled in only where a vetted source exists. Different metrics for the same provider can come from different source types — each keeps its own sourcing.
Success rate
From our tests or attributed lab reports; never a vendor marketing claim.
Response time
Average / percentile latency where a measured source exists.
IP pool size
The provider's published pool count, labeled as provider-stated.
Uptime
From status pages and published SLAs, cross-checked against lab reports.
Entry price
Verified against the provider's live pricing page, dated.
What we don't claim
Our numbers come from mixed sources — our own tests, independent lab reports and published specifications — measured at different times, from different vantage points, with different harnesses. Treat them as directional comparisons, not as a guarantee of what you'll see in production.
Real-world performance depends on your origin region, your target site's anti-bot regime that day, your concurrency, and your session strategy. A provider that scored 99% in one lab's test may score 95% in your stack with different headers and a different rotation pattern.
Because the sources differ, small gaps between two providers' numbers (a point or two of success rate) are not meaningful. Large gaps, consistent across sources, usually are. The right way to validate any provider is to run a few hundred requests through their cheapest tier against your actual targets before committing.
Checking a number
Email [email protected] with any figure on the site and we'll tell you exactly where it came from — our own run, the specific lab report, or the provider document — and when it was last verified. Find an error and we'll correct it in place and credit you in the changelog.
Methodology changelog
Every change to how we source or vet numbers is dated here. We never silently re-baseline.