Research methodology

How AI Orbit turns raw data into useful intelligence.

Our methodology is designed around traceability: what a metric means, where it came from, and which parts are measured, editorial or community-provided.

Core principles

AI Orbit prioritizes structured, comparable data and avoids presenting unavailable fields as facts. Public pages should use published/active records, preserve source context and distinguish calculated scores from direct observations.

Database firstDisplay values should come from stored records or defined calculations.

Verification visibleVerified benchmark/news states are shown separately from unverified data.

Transparent labelsEditorial, community and measured signals should not be blended silently.

AI Tool ratings & discovery

Tool pages can use existing rating, popularity and benchmark fields alongside categories, pricing models, capabilities and reviews. These signals serve different purposes and should not be treated as interchangeable.

RatingStored product rating used for ordering and discovery where available.

PopularityRelative discovery signal stored for a tool; it is not presented as audited market share.

Benchmark scorePerformance-oriented score linked to benchmark data where available.

ReviewsEditorial/community assessments shown separately from automated or benchmark signals.

AI Model benchmarks & leaderboards

Leaderboards prefer verified benchmark results. Where a composite score is shown, normalized benchmark performance can be combined using the benchmark's configured weight and maximum score. This is a comparative interface metric, not a claim that one model is universally best.

Normalized result(score ÷ benchmark max score) × weightComposite rankings depend on available coverage. Missing benchmarks can affect comparability.

Pricing Intelligence

Pricing pages separate current plans, historical snapshots, source records and detected price changes. A price shown without its billing period or plan context can be misleading, so the UI preserves those distinctions whenever the source data supports them.

News sourcing & verification

News intelligence can include source identity, publish time, AI summary/tags, importance, sentiment, verification state and duplicate detection. AI summaries are treated as summaries—not replacements for the original source—and source links remain important context.

Reviews & community signals

Published reviews are moderated before public display. Editorial reviews and community reviews can coexist, but their origin should remain visible. Reviews are subjective evidence and are not equivalent to controlled benchmark measurements.

Editorial policy

Provider marketing claims should not automatically become AI Orbit conclusions.

Commercial relationships should not silently determine rankings.

Corrections should update structured records rather than being hidden in presentation text.

Material uncertainty should be communicated when evidence is incomplete.

Limitations you should know

AI products change rapidly. Pricing, model capabilities, benchmark methodology and provider policies can change after collection. Benchmarks also measure specific tasks—not every real-world workflow. AI Orbit should therefore support decisions, not replace independent testing for high-stakes purchases or deployments.

Report a methodology or data concern