How the Flywheel Radar Scores Tools

The Flywheel Radar evaluates tools across 20 standardized metrics. Every tool in every category is measured against the same framework, which means a project management platform and an AI coding assistant go through identical evaluation dimensions. This page summarizes how that framework works. For the full technical methodology — including normalization formulas, EWMA smoothing, and hysteresis rules — see the complete methodology page.

The 20-Metric Framework

We group the 20 metrics into five evaluation pillars. Each pillar carries equal weight in the composite score:

• Functionality — core feature completeness, API coverage, integration breadth, and workflow automation depth.
• Pricing & Value — free tier availability, per-seat cost at scale, pricing transparency, and total cost of ownership.
• User Experience — onboarding friction, documentation quality, UI consistency, and learning curve for new teams.
• Performance & Reliability — uptime track record, latency benchmarks, incident response history, and scalability under load.
• Ecosystem & Trust — security certifications, compliance posture, community health, and vendor longevity signals.

Within each pillar, individual metrics are weighted by their impact on real-world adoption decisions. A tool that scores well on pricing but poorly on reliability will not receive a recommendation — the framework is designed to surface balanced strength, not isolated highlights.

Data Sources

No single data source tells the full story. The Radar draws from five independent channels to build each tool's profile:

• Public pricing pages — we capture published plans, per-seat costs, and enterprise pricing where available.
• Community forums and developer sentiment — Reddit, Hacker News, Stack Overflow, and GitHub Discussions surface real user friction points.
• Analyst reports — G2, Gartner, and Forrester data provide enterprise adoption context and peer comparison benchmarks.
• Hands-on testing — our team evaluates tools directly, running standardized test scenarios to verify vendor claims against actual behavior.
• API documentation and changelogs — release cadence, breaking change frequency, and documentation quality are tracked over time.

When sources conflict, we weight direct testing and community sentiment over vendor-provided materials. Analyst reports provide useful context but are never treated as ground truth on their own.

What Earns a Recommendation

A tool receives a Radar recommendation when it meets three conditions simultaneously: it scores above the 70th percentile in its category's composite ranking, it has no critical failures in any single pillar (no score below 40 in any of the five dimensions), and it demonstrates stable or improving momentum over a rolling 12-week window.

Tools that score well overall but show a sharp decline in one area — a security incident, a pricing change that alienates users, or a stalled development cadence — are flagged for review and may lose their recommendation until the issue is resolved. For a practical example of how these scores translate into category-level recommendations, see our Notion alternatives comparison.

What We Don't Do

The Radar does not accept vendor sponsorship, affiliate placement fees, or pay-to-rank arrangements of any kind. No vendor can purchase a higher score or a more favorable position. We do not run "sponsored reviews" or accept free enterprise licenses in exchange for coverage.

We also don't rely on a single evaluator's opinion. Every score is the output of a structured framework applied consistently across tools. Individual preferences are surfaced as qualitative notes, never baked into the quantitative score.

Deep Dive

This page covers the what and why of the scoring framework. For the how — including the exact normalization formulas, source weighting coefficients, movement classification rules, and known limitations — read the full Radar methodology documentation.