Our methodology
How we rate routers
Every router we review receives a single score from 1.0 to 5.0 (to one decimal), rolled up from six weighted categories. The rubric is publicly defined, applied consistently across products, and — crucially — independent of any commercial relationship we have with the manufacturer.
What the numbers mean
The 1–5 scale
4.5 – 5.0
Exceptional
Category-defining. Best-in-class Wi-Fi, flawless firmware, outstanding value for its tier. A product we would recommend without caveats.
4.0 – 4.4
Excellent
Strong across the board with only minor limitations. Most readers will be very happy with a 4-star router.
3.0 – 3.9
Solid
Does its job well. Has noteworthy trade-offs — maybe limited USB support, fewer VPN options, or a dated mobile app.
2.0 – 2.9
Mediocre
Works, but better options exist at the same price. Recommend only if a specific feature locks you in.
1.0 – 1.9
Avoid
Significant flaws: unreliable firmware, deceptive marketing, serious security gaps, or wildly overpriced.
Weighted criteria
What makes up the score
Performance
30%Real-world throughput and latency across 2.4 / 5 / 6 GHz. Measured at 1, 5, 10, and 25 meters with a standardized client fleet. Sustained transfers, not burst numbers.
Features & Software
25%Breadth and polish of built-in features: VPN client/server, QoS, parental controls, guest networks, USB services, mesh support, IPv6. Points added for open-source firmware compatibility.
Reliability & Firmware
20%Stability over long sessions, frequency of manufacturer updates, response to published CVEs, and firmware change quality.
Value
15%Feature density and performance per dollar at current street price. A $500 router needs to earn its premium.
Design & Build
5%Thermals, port placement, cable management, physical footprint, LED etiquette, and mobile app UX.
Support & Warranty
5%Warranty length, ease of RMA, vendor responsiveness, documentation quality, and community knowledge base.
Weights occasionally shift by category — a mesh-first product is evaluated more heavily on coverage uniformity than raw throughput, for instance. Any deviation is called out at the top of the review.
Our test rig
How we actually test
Controlled environment
A 150 m² residential test space with representative RF noise (microwave, Bluetooth speakers, 2.4 GHz smart-home hubs) kept consistent between tests.
Standardized clients
iperf3 on Windows 11, macOS Sonoma, and iPhone 15 Pro. Wired 2.5 GbE baseline host to isolate LAN bottlenecks.
Benchmarks, not demos
10-minute sustained throughput runs — not 5-second peaks. Latency measured under concurrent load from 4 streaming clients.
Multi-day stability
Each router runs as our primary gateway for at least 7 days before verdict. Firmware quirks show up on day 3, not hour 1.
What we don't do
Independence guarantees
- No manufacturer previews a review before publication.
- Review units are returned within 30 days of verdict, never retained as personal equipment or resold.
- Payment for placement, paid rankings, or "sponsored best-of" lists are refused.
- Affiliate links exist and are clearly labeled. They do not change scoring.
Read the full policy in our Ethics Statement or reach out if anything here needs clarification.