What evidence would justify moving a series forward when its mechanism of action is uncertain?

Why

Phenotypic discovery has produced many drugs whose mechanism was worked out later, so demanding mechanism before progression is not a rule. The question is whether uncertainty can be managed. A series with a smooth structure-activity relationship, where small chemical changes track with activity, behaves like something binding a real site rather than like interference.

Activity reproduced in several independent models, including patient-derived material, lowers the chance of a model-specific artifact. An inactive close analog that does nothing is strong evidence. What mechanism normally provides, and must be replaced, is a way to tell whether the drug is doing its job in a person.

Without a target, a pharmacodynamic marker of the phenotype itself can serve. Without that, a trial can only read out on clinical endpoints, which is slow and expensive. Risks also grow.

Selectivity cannot be assessed against an unknown target, patient selection has no biomarker, and an unexpected toxicity is hard to attribute. Target deconvolution work should run in parallel rather than be deferred. The honest framing is that progression is a bet on a well-characterized phenotype, with the mechanism work funded alongside it and a decision point where the absence of a marker becomes blocking.