A score that refuses, rather than a score that rounds up
Five weighted criteria produce a number. Five hard rules stop the number mattering. An answer that scores 86 and cites a certificate lapsing before the return date is red, and no amount of good writing moves it.
Every bid tool scores answers. The scoring is the easy half and the part that sells demonstrations: a number, a traffic light, a dial that moves when you improve a paragraph. The hard half is what the number is not allowed to do.
Bid-On scores an answer on five criteria — compliance, evidence, buyer relevance, specificity, and clarity against the stated limits — weighted so that answering the question actually asked counts for most and writing nicely counts for least. The weights are copied onto the bid when it is created, not read live, so a rubric reversioned next quarter cannot restate what a submission already sent was scored against.
Above eighty is green, above fifty is amber, and below that is red. Then come the five rules that produce a red whatever the prose scored: a mandatory question nobody answered, a pass/fail question answered “no”, a cited certificate or insurance that expires before the deadline, a declaration the customer has not confirmed, and an answer past a word limit the buyer stated.
Each of those is a reason a submission is rejected rather than marked down, and that difference is the whole design. A marked-down answer costs you points. A rejected submission costs you the bid, and everything else in it was spent for nothing.
The expiry rule is judged against the day the buyer reads it, never against today. A certificate that lapses the week before return is not evidence for that submission, and a tool that showed it as current because it is current this morning is a tool that lets a pack go out wrong. On a rolling submission with no return date — a dynamic market, a maintained registration — the judgement day is the day the pack is read, and the position says which day it used, because a readiness figure somebody screenshots is worthless without the date it was true on.
Where a hard rule fires, the screen also shows the band the prose alone would have earned. “This would have been green” is the sentence that tells a reviewer the writing is fine and the problem is a certificate — which is a ten-second fix rather than a rewrite.
And the scorer is a separate act from the writer. The command that records an answer carries no score field at all, so the only way a score exists is a second pass by somebody or something that did not write it. Every criterion needs a reason a reviewer could argue with, recorded beside it.
See the record being made rather than described.
Launch the demo environment