1. Start with the job
We define the user, intended output, constraints, and cost of failure before recommending software. A universal ranking without this context is rarely useful.
2. Build a primary-source record
We prioritize official documentation, repositories, release notes, pricing pages, licenses, and terms. Secondary reporting may add context but does not replace a product’s current primary sources.
3. Label evidence honestly
Source-checked means claims were verified against cited sources. Hands-on tested will be used only when we directly ran a disclosed scenario. We will not invent performance, testimonials, or user counts.
4. Score for fit
Our scores consider capability breadth, ease of first success, output fit, workflow reliability, control and privacy, rights clarity, and total cost. Scores are editorial summaries, not laboratory measurements.
5. Publish disqualifying limitations
A recommendation must show the main reason not to choose it. Hardware constraints, missing features, licensing ambiguity, price volatility, and workflow friction belong next to the verdict.
6. Date volatile facts
Prices, versions, supported platforms, and project statistics can change. We attach checked dates and avoid presenting volatile numbers as timeless facts.
7. Keep commercial influence visible
Affiliate relationships do not purchase rankings or suppress limitations. Affiliate links, sponsored placements, and supplied access will be clearly labeled.
Correction policy
Material corrections should update the page’s modified date and, where meaningful, its verdict. Readers and product teams may provide evidence, but final editorial judgment remains independent.