Replacement scorecard: pass or fail first, then score
List what the new tool must do and which candidates you are weighing. Mark each cell Yes, Partly, No or Not checked yet. Hard requirements knock a candidate out on a No; the rest are weighted. The result says whether the winner is settled or which unchecked cells could still change it. Everything runs in your browser and nothing is sent anywhere.
must: for a hard requirement, or w1: to w5: for a weight (no prefix means weight 3). Up to 15 lines.
How it scores
- Hard requirements first. A No on a
must:line knocks the candidate out, and the result names the requirement. A Not checked yet on a hard requirement does not knock it out, but the candidate cannot be called the winner until it is checked. - Weighted score. Yes counts 1, Partly counts 0.5, No counts 0, multiplied by the weight and shown as a percentage of the total weight. Not checked yet is shown as a range: the lowest score if every unknown turns out No, and the highest if every unknown turns out Yes.
- Settled, tied or not settled. The leader is settled only when its lowest possible score is strictly higher than every other surviving candidate's highest possible score and it has no unchecked hard requirement. If two candidates finish with the same checked score, the result says Tied. Otherwise the tool lists the unchecked cells worth checking first.
- Only hard requirements. If every line is a
must:line there is nothing to weight, so the result is pass or fail only: it lists who is not knocked out and shows no percentages. - Notices. Past 15 requirements or 8 candidates the extra lines are dropped and a notice says how many. A weight outside w1 to w5 is counted as weight 3 and flagged, and duplicate candidate names are flagged.
The tool cannot find facts for you. A Yes with no source is a guess: write the source next to each cell in your own notes, and read the price or limit on the vendor's own page, dated, before you trust it. For a worked example with dated sources see the GitHub Actions replacement sample.
When you would rather not do the reading
The slow part is filling the matrix with checked facts. Switch Brief does that for $39 within 24 hours: your must-haves become pass or fail tests, three candidates plus staying put are read from live public pages with source and date, and the unknowns are listed. It is AI-written research from public pages, not security, legal or compliance advice, and nothing you supply is run.