How recommendations work

Facts, estimates, and judgment have different jobs.

Official fact

Plan prices, limits, commercial-use language, and documented features map to official source records.

Calculated result

Costs and Match Scores come from deterministic TypeScript and explicit assumptions.

Editorial judgment

Verdicts explain who benefits and what the published facts do not settle.

Match Score

Each use case defines weights totaling 1.0. The calculator evaluates budget fit, required features, usage volume, commercial use, workflow, languages, API access, and editing. Sound quality is excluded because no standardized hands-on test exists in this release.

Unknown fields are removed from the score denominator rather than scored as zero. They reduce Data confidence. Ties break by required conditions met, estimated monthly cost, data confidence, then tool name. Near-equal results are described as a close match.

Costs and assumptions

Narration defaults to 150 words per minute and six characters per word including a space. Annual rates are labeled as annual commitments. Free restrictions, unknown overage, commercial-plan requirements, and exceeded limits remain visible.

Sources and freshness

Only official pricing, feature, terms, and help pages enter public product facts. Data is Fresh for 14 days, Aging through day 30, Stale from day 31, and excluded from automatic recommendation after 60 days. Verification dates change only after a source is checked.

Editorial and automation

This release is based on published data only; it does not claim hands-on product tests. Future automation may normalize official sources and draft structured JSON, but unsupported claims fail validation and automatic publication is disabled.

Commercial independence

Commission rates, affiliate approval, and revenue do not enter recommendation data or tie-breaking. Affiliate URLs are resolved only after rankings are complete.