Start by separating leading explanations into mutually useful buckets: transport or oxygen limitation, regulation or stress response, feed strategy mismatch, measurement ambiguity, or downstream interpretation errors.

Then define which 3 to 5 runs would actually discriminate between those buckets. If a proposed run cannot rule out an important alternative explanation, it is probably noise dressed up as activity.

This is where a decision memo matters. It forces the team to state what is most likely true, what is still unresolved, and what evidence would change the plan.