How We Test
Our testing methodology
Every ranking on this site is the product of structured, hands-on testing. This page documents exactly how we do it – the hardware, the rubrics, the scoring, and the safeguards.
Four principles that guide every test
1
We buy or download it ourselves
We use the same version available to the public – no pre-release builds, no vendor-configured demo environments. If the free tier is meaningfully different from paid, we test both.
2
Fixed scoring rubric per category
Each category has a documented rubric with weighted criteria. We do not adjust criteria after seeing a product’s performance. Rubrics are published in full below.
3
Two-reviewer sign-off
The primary reviewer scores the product and writes the draft. A second reviewer performs an independent check of the score and testing notes before publication.
4
Regular re-tests
Top-ranked products are retested at least twice per year. If a major update ships, we retest within 30 days and update the article with a timestamped revision note.
Test environments
Our lab runs standardised hardware profiles for each category to ensure consistent benchmarks.
We maintain three standardised test machines:
Profile
Spec
Used for
Desktop (Windows)
Intel Core i7, 32 GB RAM, 1 TB NVMe, Windows 11 Pro
Productivity, security, developer tools
Desktop (macOS)
Apple M3 Pro, 36 GB RAM, 512 GB SSD, macOS 15
Creative software, cross-platform comparisons
Network bench
Dedicated machine with dual NIC, pfSense router, controlled WAN link
VPN, firewall, DNS tools
Mobile testing uses current-generation iOS and Android devices on a clean profile with no third-party apps pre-installed.
Scoring rubric
Each product is scored 1 – 10 on each criterion. The overall score is a weighted average. Category-specific rubrics add or replace criteria as appropriate.
Criterion
Weight
What we measure
Core feature quality
30%
Does it do its primary job reliably and well?
Ease of use
20%
Onboarding, UI clarity, documentation quality
Performance
15%
Speed, resource usage, stability under load
Value for money
15%
Pricing vs. feature set vs. alternatives
Privacy & security
10%
Data handling, permissions, update cadence
Support & updates
10%
Responsiveness, update frequency, changelog quality
What we do not do
—
Accept payment to influence scores or rankings.
—
Publish reviews of products we have not personally tested.
—
Rely on vendor-supplied benchmark results without independent verification.
—
Adjust scores after publication without a public correction notice.
—
Keep negative reviews private to protect affiliate relationships.
Questions about our methodology?
If you believe a test result is inaccurate or our methodology has a gap, we want to hear about it.
Contact the editorial team