Review & Testing Methodology

Published September 28, 2026.

This page sets the evidence required for our different article formats. It does not certify a laboratory, a particular test rig or a contributor's credentials.

Researched guides

A researched guide explains information from publisher documentation, official storefronts and other identified sources. It should state its scope, link its sources and identify the date or version checked. It is not a controlled test or proof of personal play time.

Hands-on reviews

A hands-on review should specify who played, the platform, game version, period and extent of play. It should distinguish observed behavior from information supplied by the publisher and disclose relevant commercial relationships. A score requires a stated rationale; a download overview does not need a review score.

Hardware and performance tests

A benchmark should identify the hardware models, firmware, operating system, drivers, game build, resolution and settings. Explain the test scene, capture tool, number of runs, warm-up conditions and how results were summarized. Where possible, provide the underlying measurements and their variation.

Latency measurements must identify the start and end of the measured interval. A polling interval is not the same as end-to-end click-to-display latency. Manufacturer specifications and theoretical calculations must be labeled separately from observed results.

Limitations and updates

A result applies to the configuration and conditions tested. Do not generalize a single run to all systems. A new patch, driver or device revision may change the outcome. Mark a retest only when a new test was performed; identify documentary corrections separately.

Missing evidence

If an article does not include the information needed to reproduce a numerical claim, treat that claim as unverified and request clarification. Older articles are subject to revision under our editorial policy.