How we review software
Every claim in a KanzenAI review traces to a source. This page documents exactly what those sources are, how scoring works, and what we refuse to publish.
Where the data comes from
- Vendor pricing pages — pulled directly, quoted with the date we pulled them. If pricing is "contact sales," we say that; we never invent a number.
- Product documentation and changelogs — feature claims come from what vendors publish, not from marketing copy.
- Public demos and walkthroughs — for workflow and interface assessments.
- Aggregated user reports — patterns across public reviews and operator communities, weighted toward recurring complaints and recurring praise, not one-off anecdotes.
Research is AI-assisted and editorially reviewed. We are a research desk, not a testing lab — we do not claim hands-on trials, test benches, or lab measurements, and any review that implied otherwise has been corrected.
How scoring works
- Pricing transparency — public, complete pricing scores higher than gated "contact sales" pricing at equal capability.
- Capability vs. the job — tools are scored against the specific operator's workflow (a roofing contractor's estimating needs, a restaurant's ticket flow), not against a generic feature checklist.
- Total cost honesty — required add-ons, per-seat charges, and contract minimums count against a low sticker price.
- Trajectory — active development and shipping cadence versus stagnation.
Commission size plays no role in scoring. Vendors cannot pay for placement, cannot preview reviews, and cannot pay to remove criticism.
Freshness
Pricing is re-verified on a rolling basis and each review states when its pricing was last checked. When a vendor changes pricing materially, the review is updated and the modification date reflects it.
What we refuse to publish
- A product section whose vendor sources we could not verify at research time
- Fabricated test results, fabricated user counts, or invented pricing
- Sponsored reviews or paid placements presented as editorial
- Ratings based on a vendor relationship rather than the rubric
Corrections
Factual errors are corrected in the article with a note. Report one: hello@kanzenai.com.
Related: About KanzenAI · Affiliate disclosure · Contact