Hard gates

If business impact scores 0 or 1, stop the candidate as scored. Technical feasibility doesn't justify a project with too little business value.

If adoption friction scores 0 or 1, stop it as scored as well. The people involved cannot absorb the change dependably enough. A system can work technically and still fail to help because it never becomes part of the job. If likelihood of success is 0, stop: the proposed approach has no credible path to the outcome under the conditions being scored.

Any of these conditions overrides the total. Record what would have to change before reconsidering, or compare another candidate. You may choose preparation or a deliberate wait, but don't use a high score elsewhere to push the same blocked experiment ahead. A genuinely changed scope needs a fresh assessment.

Cost and risk is deliberately not a gate. Three dimensions can end a candidate on their own: business impact, adoption friction, and likelihood of success. A poor cost and risk rating lowers the total and should be discussed openly, but a company may still decide that an important opportunity is worth an expensive or exposed first step. That is an owner's judgment about appetite, not arithmetic, so the model reports it rather than deciding it.