Method version: criteria-only-v1
How the recorded opinions are ordered
The ranking method publishes its inputs and limits. It summarises documented editorial judgement without turning catalogue copy into product-quality evidence.
Which records qualify
A tool needs all five numeric criterion values between 0 and 10 in the published criteria file. Automatic or incomplete records do not qualify. Retired and historical product entries are excluded, so an assessment of an old product does not become an endorsement of its replacement.
There are 93 currently applicable opinion records. The original criterion judgement dates were not recorded. This method was published on 9 October 2026; that publication does not reset the dates or establish that every opinion was re-tested.
The arithmetic
Opinion index = round((functionality + pricing_value + ease_of_use + reliability + founder_fit) × 2)
Each criterion has equal weight. A record containing 9, 8, 8, 7 and 9 gives an index of 82/100. Category lists are ordered by the underlying average; equal averages share the same position. Alphabetical ordering only resolves display order within a tie and adds no quality points.
| Criterion | What the opinion considers |
|---|---|
| Functionality | Fit of the documented capabilities for the stated job. |
| Pricing value | The editor’s value judgement for the intended founder workflow. Verify current plans separately. |
| Ease of use | The editor’s view of setup and workflow complexity. This is not a measured onboarding-time result. |
| Reliability judgement | The editor’s opinion about operational confidence. It is not a measured uptime or security score. |
| Founder fit | Suitability for the solo or small-team context described by the assessment. |
What never adds points
Legacy verdict labels, affiliate payments, ownership class, name or domain length, marketing keywords, public discussion counts and catalogue popularity add no ranking points. The index does not predict revenue or traffic.
Catalogue discovery is separate
A catalogue-only entry has no editorial index, position or recommendation. These listings are shown alphabetically. A category with zero qualifying records says “No editorial pick yet”; it does not appoint the highest automatic estimate as a winner.
The former Best Score mixed automatic catalogue estimates with editorial and public-momentum inputs. The current ranking pages replace that blend with the criteria-only method above. Old copied score snapshots and automatic keyword-derived scores are not current rankings.
Different jobs need different choices
The index compares recorded opinions within a category. The main assessment page is alphabetical rather than an overall quality leaderboard. A tool’s rank does not remove the need to check your required features, region, pricing, data handling and exit path.
For example, a one-page Carrd route and a more complex Framer or Webflow website have different requirements. The route’s scoped recommendation does not make Carrd a universal website-builder winner. Read the website category scope.
Documentation and testing
Provider links support factual feature descriptions. Their check dates are separate from criterion opinions and pricing-table dates. Documentation does not prove comparative output quality, speed, reliability or hands-on testing.
The published workflow records identify executed tasks, environment, interaction method, synthetic inputs, outputs and limits. They add no points to the opinion index. A one-sided product test does not establish a comparative winner.
Corrections and editorial review
Factual source corrections can be published with their receipt. New opinions, changed criterion values and changed recommendations require editorial review. Send the tool URL, disputed detail and primary source.
One label, one meaning
On 20 October 2026 we removed the undocumented Our Pick distinction. Recommended retains its recorded editorial meaning. Category order still uses the same five stored criteria; no score, weighting, tie rule or rank changed in this correction.